%!PS
%%Version: 3.3.1
%%DocumentFonts: (atend)
%%Pages: (atend)
%%EndComments
%
% Version 3.3.1 prologue for troff files.
%

/#copies 1 store
/aspectratio 1 def
/formsperpage 1 def
/landscape false def
/linewidth .3 def
/magnification 1 def
/margin 0 def
/orientation 0 def
/resolution 720 def
/rotation 1 def
/xoffset 0 def
/yoffset 0 def

/roundpage true def
/useclippath true def
/pagebbox [0 0 612 792] def

/R  /Times-Roman def
/I  /Times-Italic def
/B  /Times-Bold def
/BI /Times-BoldItalic def
/H  /Helvetica def
/HI /Helvetica-Oblique def
/HB /Helvetica-Bold def
/HX /Helvetica-BoldOblique def
/CW /Courier def
/CO /Courier def
/CI /Courier-Oblique def
/CB /Courier-Bold def
/CX /Courier-BoldOblique def
/PA /Palatino-Roman def
/PI /Palatino-Italic def
/PB /Palatino-Bold def
/PX /Palatino-BoldItalic def
/Hr /Helvetica-Narrow def
/Hi /Helvetica-Narrow-Oblique def
/Hb /Helvetica-Narrow-Bold def
/Hx /Helvetica-Narrow-BoldOblique def
/KR /Bookman-Light def
/KI /Bookman-LightItalic def
/KB /Bookman-Demi def
/KX /Bookman-DemiItalic def
/AR /AvantGarde-Book def
/AI /AvantGarde-BookOblique def
/AB /AvantGarde-Demi def
/AX /AvantGarde-DemiOblique def
/NR /NewCenturySchlbk-Roman def
/NI /NewCenturySchlbk-Italic def
/NB /NewCenturySchlbk-Bold def
/NX /NewCenturySchlbk-BoldItalic def
/ZD /ZapfDingbats def
/ZI /ZapfChancery-MediumItalic def
/S  /S def
/S1 /S1 def
/GR /Symbol def

/inch {72 mul} bind def
/min {2 copy gt {exch} if pop} bind def

/setup {
	counttomark 2 idiv {def} repeat pop

	landscape {/orientation 90 orientation add def} if
	/scaling 72 resolution div def
	linewidth setlinewidth
	1 setlinecap

	pagedimensions
	xcenter ycenter translate
	orientation rotation mul rotate
	width 2 div neg height 2 div translate
	xoffset inch yoffset inch neg translate
	margin 2 div dup neg translate
	magnification dup aspectratio mul scale
	scaling scaling scale

	addmetrics
	0 0 moveto
} def

/pagedimensions {
	useclippath userdict /gotpagebbox known not and {
		/pagebbox [clippath pathbbox newpath] def
		roundpage currentdict /roundpagebbox known and {roundpagebbox} if
	} if
	pagebbox aload pop
	4 -1 roll exch 4 1 roll 4 copy
	landscape {4 2 roll} if
	sub /width exch def
	sub /height exch def
	add 2 div /xcenter exch def
	add 2 div /ycenter exch def
	userdict /gotpagebbox true put
} def

/addmetrics {
	/Symbol /S null Sdefs cf
	/Times-Roman /S1 StandardEncoding dup length array copy S1defs cf
} def

/pagesetup {
	/page exch def
	currentdict /pagedict known currentdict page known and {
		page load pagedict exch get cvx exec
	} if
} def

/decodingdefs [
	{counttomark 2 idiv {y moveto show} repeat}
	{neg /y exch def counttomark 2 idiv {y moveto show} repeat}
	{neg moveto {2 index stringwidth pop sub exch div 0 32 4 -1 roll widthshow} repeat}
	{neg moveto {spacewidth sub 0.0 32 4 -1 roll widthshow} repeat}
	{counttomark 2 idiv {y moveto show} repeat}
	{neg setfunnytext}
] def

/setdecoding {/t decodingdefs 3 -1 roll get bind def} bind def

/w {neg moveto show} bind def
/m {neg dup /y exch def moveto} bind def
/done {/lastpage where {pop lastpage} if} def

/f {
	dup /font exch def findfont exch
	dup /ptsize exch def scaling div dup /size exch def scalefont setfont
	linewidth ptsize mul scaling 10 mul div setlinewidth
	/spacewidth ( ) stringwidth pop def
} bind def

/changefont {
	/fontheight exch def
	/fontslant exch def
	currentfont [
		1 0
		fontheight ptsize div fontslant sin mul fontslant cos div
		fontheight ptsize div
		0 0
	] makefont setfont
} bind def

/sf {f} bind def

/cf {
	dup length 2 idiv
	/entries exch def
	/chtab exch def
	/newencoding exch def
	/newfont exch def

	findfont dup length 1 add dict
	/newdict exch def
	{1 index /FID ne {newdict 3 1 roll put}{pop pop} ifelse} forall

	newencoding type /arraytype eq {newdict /Encoding newencoding put} if

	newdict /Metrics entries dict put
	newdict /Metrics get
	begin
		chtab aload pop
		1 1 entries {pop def} for
		newfont newdict definefont pop
	end
} bind def

%
% A few arrays used to adjust reference points and character widths in some
% of the printer resident fonts. If square roots are too high try changing
% the lines describing /radical and /radicalex to,
%
%	/radical	[0 -75 550 0]
%	/radicalex	[-50 -75 500 0]
%
% Move braceleftbt a bit - default PostScript character is off a bit.
%

/Sdefs [
	/bracketlefttp		[201 500]
	/bracketleftbt		[201 500]
	/bracketrighttp		[-81 380]
	/bracketrightbt		[-83 380]
	/braceleftbt		[203 490]
	/bracketrightex		[220 -125 500 0]
	/radical		[0 0 550 0]
	/radicalex		[-50 0 500 0]
	/parenleftex		[-20 -170 0 0]
	/integral		[100 -50 500 0]
	/infinity		[10 -75 730 0]
] def

/S1defs [
	/underscore		[0 80 500 0]
	/endash			[7 90 650 0]
] def
%
% Tries to round clipping path dimensions, as stored in array pagebbox, so they
% match one of the known sizes in the papersizes array. Lower left coordinates
% are always set to 0.
%

/roundpagebbox {
    7 dict begin
	/papersizes [8.5 inch 11 inch 14 inch 17 inch] def

	/mappapersize {
		/val exch def
		/slop .5 inch def
		/diff slop def
		/j 0 def
		0 1 papersizes length 1 sub {
			/i exch def
			papersizes i get val sub abs
			dup diff le {/diff exch def /j i def} {pop} ifelse
		} for
		diff slop lt {papersizes j get} {val} ifelse
	} def

	pagebbox 0 0 put
	pagebbox 1 0 put
	pagebbox dup 2 get mappapersize 2 exch put
	pagebbox dup 3 get mappapersize 3 exch put
    end
} bind def

%%EndProlog
%%BeginSetup
mark
/resolution 720 def
setup
2 setdecoding
%%EndSetup
%%Page: 1 1
/saveobj save def
mark
1 pagesetup
12 B f
(C Traps and Pitfalls*)3 1099 1 2330 1140 t
10 I f
(Andrew Koenig)1 625 1 2567 1380 t
10 R f
(AT&T Bell Laboratories)2 993 1 2383 1560 t
(Murray Hill, New Jersey 07974)4 1267 1 2246 1680 t
(ABSTRACT)2618 1920 w
( sharp, and extremely useful in)5 1331(The C language is like a carving knife: simple,)8 2019 2 1330 2316 t
( tool, C can injure people who don't know how to handle it.)12 2399( any sharp)2 414( Like)1 235(skilled hands.)1 552 4 1080 2436 t
(This paper shows some of the ways C can injure the unwary, and how to avoid injury.)16 3434 1 1080 2556 t
10 B f
(0. Introduction)1 670 1 720 2796 t
10 R f
( lan-)1 187( The)1 212(The C language and its typical implementations are designed to be used easily by experts.)14 3671 3 970 2952 t
( user who has)3 552( A)1 124( few restrictions to keep the user from blundering.)8 2015( are)1 149( There)1 285(guage is terse and expressive.)4 1195 6 720 3072 t
(blundered is often rewarded by an effect that is not obviously related to the cause.)14 3268 1 720 3192 t
( they are unexpected, it)4 980( Because)1 394(In this paper, we will look at some of these unexpected rewards.)11 2696 3 970 3348 t
( we have made a rough effort to do so)9 1573( Nevertheless,)1 597( impossible to classify them completely.)5 1646(may well be)2 504 4 720 3468 t
( passing)1 329( assume the reader has at least a)7 1292( We)1 191(by looking at what has to happen in order to run a C program.)13 2508 4 720 3588 t
(acquaintance with the C language.)4 1372 1 720 3708 t
( 2 fol-)2 252( Section)1 354(Section 1 looks at problems that occur while the program is being broken into tokens.)14 3464 3 970 3864 t
( Section)1 354( tokens into declarations, expressions, and statements.)6 2180(lows the program as the compiler groups its)7 1786 3 720 3984 t
( that a C program is often made out of several parts that are compiled separately and bound)17 3810(3 recognizes)1 510 2 720 4104 t
( while the program is actually)5 1190( 4 deals with misconceptions of meaning: things that happen)9 2427(together. Section)1 703 3 720 4224 t
( In)1 142( our programs and the library routines they use.)8 1966( 5 examines the relationship between)5 1518(running. Section)1 694 4 720 4344 t
( that the program we write is not really the program we run; the preprocessor has gotten at)17 3618(section 6 we note)3 702 2 720 4464 t
( might run on one implementa-)5 1261( section 7 discusses portability problems: reasons a program)8 2427( Finally,)1 362(it first.)1 270 4 720 4584 t
(tion and not another.)3 827 1 720 4704 t
10 B f
( Pitfalls)1 325(1. Lexical)1 436 2 720 4944 t
10 R f
(The first part of a compiler is usually called a)9 1834 1 970 5100 t
10 I f
(lexical analyzer)1 637 1 2831 5100 t
10 R f
( the sequence of charac-)4 974( looks at)2 343(. This)1 255 3 3468 5100 t
( up into)2 312(ters that make up the program and breaks them)8 1906 2 720 5220 t
10 I f
(tokens)2966 5220 w
10 R f
( token is a sequence of one or more char-)9 1669(. A)1 150 2 3221 5220 t
( C, for instance, the)4 842( In)1 147(acters that have a \(relatively\) uniform meaning in the language being compiled.)11 3331 3 720 5340 t
(token)720 5460 w
10 CW f
(->)969 5460 w
10 R f
( meaning that is quite distinct from that of either of the characters that make it up, and that is)19 3720(has a)1 204 2 1116 5460 t
(independent of the context in which the)6 1581 1 720 5580 t
10 CW f
(->)2326 5580 w
10 R f
(appears.)2471 5580 w
(For another example, consider the statement:)5 1797 1 970 5736 t
10 CW f
(if \(x > big\) big = x;)6 1260 1 1080 5916 t
10 R f
( character in this statement is a separate token, except for the keyword)12 2981(Each non-blank)1 643 2 720 6096 t
10 CW f
(if)4384 6096 w
10 R f
(and the two)2 496 1 4544 6096 t
(instances of the identifier)3 1012 1 720 6216 t
10 CW f
(big)1757 6216 w
10 R f
(.)1937 6216 w
( must)1 227( It)1 117( the preprocessor reads the program.)5 1481( First)1 240(In fact, C programs are broken into tokens twice.)8 2005 5 970 6372 t
( must then)2 425( It)1 115( represent macros.)2 736(tokenize the program so that it can find the identifiers, some of which may)13 3044 4 720 6492 t
( the result of the macro)5 1013( Finally,)1 377( evaluating that macro.)3 969(replace each macro invocation by the result of)7 1961 4 720 6612 t
( compiler)1 390( The)1 215( the compiler proper.)3 867(replacement is reassembled into a character stream which is given to)10 2848 4 720 6732 t
(then breaks the stream into tokens a second time.)8 1961 1 720 6852 t
8 S1 f
(__________________)720 6952 w
8 R f
( book)1 209( paper, greatly expanded, is the basis for the)8 1623(* This)1 250 3 720 7052 t
8 I f
(C Traps and Pitfalls)3 738 1 2851 7052 t
8 R f
(\(Addison-Wesley, 1989, ISBN)2 1042 1 3638 7052 t
(0\261201\26117928\2618\); interested readers may wish to refer there as well.)9 2163 1 720 7152 t
cleartomark
showpage
saveobj restore
%%EndPage: 1 1
%%Page: 0 2
/saveobj save def
mark
2 pagesetup
10 R f
( of tokens and)3 584(In this section, we will explore some common misunderstandings about the meanings)11 3486 2 970 840 t
( will talk about the preprocessor)5 1299( We)1 191( and the characters that make them up.)7 1556(the relationship between tokens)3 1274 4 720 960 t
(later.)720 1080 w
10 B f
(1.1.)720 1320 w
10 CW f
(=)920 1320 w
10 B f
(is not)1 231 1 1005 1320 t
10 CW f
(==)1261 1320 w
10 R f
(Programming languages derived from Algol, such as Pascal and Ada, use)10 3014 1 970 1476 t
10 CW f
(:=)4017 1476 w
10 R f
(for assignment and)2 776 1 4170 1476 t
10 CW f
(=)4980 1476 w
10 R f
( on the other hand, uses)5 1023( C,)1 158(for comparison.)1 648 3 720 1596 t
10 CW f
(=)2590 1596 w
10 R f
(for assignment and)2 792 1 2691 1596 t
10 CW f
(==)3524 1596 w
10 R f
( because)1 355( is)1 108( This)1 244(for comparison.)1 648 4 3685 1596 t
(assignment is more frequent than comparison, so the more common meaning is given to the shorter symbol.)16 4313 1 720 1716 t
( so that multiple assignments \(such as)6 1547(Moreover, C treats assignment as an operator,)6 1865 2 970 1872 t
10 CW f
(a=b=c)4413 1872 w
10 R f
(\) can be)2 327 1 4713 1872 t
(written easily and assignments can be embedded in larger expressions.)9 2817 1 720 1992 t
( write an assignment where one)5 1306(This convenience causes a potential problem: one can inadvertently)8 2764 2 970 2148 t
( this statement, which looks like it is checking whether)9 2193( Thus,)1 275(intended a comparison.)2 929 3 720 2268 t
10 CW f
(x)4142 2268 w
10 R f
(is equal to)2 411 1 4227 2268 t
10 CW f
(y)4663 2268 w
10 R f
(:)4723 2268 w
10 CW f
(if \(x = y\))3 600 1 1080 2448 t
(foo\(\);)1330 2568 w
10 R f
(actually sets)1 495 1 720 2748 t
10 CW f
(x)1245 2748 w
10 R f
(to the value of)3 589 1 1335 2748 t
10 CW f
(y)1954 2748 w
10 R f
( consider the following)3 939( Or)1 160(and then checks whether that value is nonzero.)7 1897 3 2044 2748 t
(loop that is intended to skip blanks, tabs, and newlines in a file:)12 2538 1 720 2868 t
10 CW f
(while \(c == ' ' || c = '\\t' || c == '\\n'\))12 2460 1 1080 3048 t
(c = getc \(f\);)3 780 1 1330 3168 t
10 R f
(The programmer mistakenly used)3 1354 1 720 3348 t
10 CW f
(=)2102 3348 w
10 R f
(instead of)1 394 1 2190 3348 t
10 CW f
(==)2612 3348 w
10 R f
(in the comparison with)3 928 1 2760 3348 t
10 CW f
('\\t')3716 3348 w
10 R f
(. This ``comparison'' actu-)3 1084 1 3956 3348 t
(ally assigns)1 468 1 720 3468 t
10 CW f
('\\t')1217 3468 w
10 R f
(to)1486 3468 w
10 CW f
(c)1593 3468 w
10 R f
(and compares the \(new\) value of)5 1319 1 1681 3468 t
10 CW f
(c)3028 3468 w
10 R f
( Since)1 275(to zero.)1 302 2 3116 3468 t
10 CW f
('\\t')3721 3468 w
10 R f
(is not zero, the ``compari-)4 1051 1 3989 3468 t
( it does after that depends on whether the)8 1657( What)1 268( the entire file.)3 588(son'' will always be true, so the loop will eat)9 1807 4 720 3588 t
( it does, the)3 479( If)1 122( a program to keep reading after it has reached end of file.)12 2384(particular implementation allows)2 1335 4 720 3708 t
(loop will run forever.)3 854 1 720 3828 t
( form)1 225(Some C compilers try to help the user by giving a warning message for conditions of the)16 3621 2 970 3984 t
10 I f
(e1 =)1 193 1 4847 3984 t
(e2)720 4104 w
10 R f
( to a variable and)4 719( avoid warning messages from such compilers, when you want to assign a value)13 3312(. To)1 195 3 814 4104 t
( other words, instead)3 853( In)1 140( making the comparison explicit.)4 1341(then check whether the variable is zero, consider)7 1986 4 720 4224 t
(of:)720 4344 w
10 CW f
(if \(x = y\))3 600 1 1080 4524 t
(foo\(\);)1330 4644 w
10 R f
(write:)720 4824 w
10 CW f
(if \(\(x = y\) != 0\))5 1020 1 1080 5004 t
(foo\(\);)1330 5124 w
10 R f
(This will also help make your intentions plain.)7 1861 1 720 5304 t
10 B f
(1.2.)720 5544 w
10 CW f
(&)920 5544 w
10 B f
(and)1005 5544 w
10 CW f
(|)1192 5544 w
10 B f
(are not)1 302 1 1277 5544 t
10 CW f
(&&)1604 5544 w
10 B f
(or)1749 5544 w
10 CW f
(||)1868 5544 w
10 R f
(It is easy to miss an inadvertent substitution of)8 1901 1 970 5700 t
10 CW f
(=)2901 5700 w
10 R f
(for)2991 5700 w
10 CW f
(==)3137 5700 w
10 R f
( languages use)2 594(because so many other)3 921 2 3287 5700 t
10 CW f
(=)4833 5700 w
10 R f
(for)4924 5700 w
( easy to interchange)3 801( is also)2 284(comparison. It)1 605 3 720 5820 t
10 CW f
(&)2437 5820 w
10 R f
(and)2524 5820 w
10 CW f
(&&)2695 5820 w
10 R f
(, or)1 135 1 2815 5820 t
10 CW f
(|)2977 5820 w
10 R f
(and)3064 5820 w
10 CW f
(||)3235 5820 w
10 R f
(, especially because the)3 942 1 3355 5820 t
10 CW f
(&)4324 5820 w
10 R f
(and)4411 5820 w
10 CW f
(|)4582 5820 w
10 R f
(operators)4669 5820 w
( will look at these operators more)6 1391( We)1 197( counterparts in some other languages.)5 1581(in C are different from their)5 1151 4 720 5940 t
(closely in section 4.)3 794 1 720 6060 t
10 B f
( Tokens)1 337(1.3. Multi-character)1 881 2 720 6300 t
10 R f
(Some C tokens, such as)4 947 1 970 6456 t
10 CW f
(/)1942 6456 w
10 R f
(,)2002 6456 w
10 CW f
(*)2052 6456 w
10 R f
(, and)1 194 1 2112 6456 t
10 CW f
(=)2331 6456 w
10 R f
( C tokens, such as)4 723( Other)1 278( long.)1 229(, are only one character)4 932 4 2391 6456 t
10 CW f
(/*)4579 6456 w
10 R f
(and)4725 6456 w
10 CW f
(==)4895 6456 w
10 R f
(,)5015 6456 w
( the C compiler encounters a)5 1160( When)1 291(and identifiers, are several characters long.)5 1728 3 720 6576 t
10 CW f
(/)3927 6576 w
10 R f
(followed by an)2 605 1 4015 6576 t
10 CW f
(*)4648 6576 w
10 R f
(, it must)2 332 1 4708 6576 t
( C)1 94( The)1 207( one single token.)3 711(be able to decide whether to treat these two characters as two separate tokens or as)15 3308 4 720 6696 t
( given charac-)2 564(reference manual tells how to decide: ``If the input stream has been parsed into tokens up to a)17 3756 2 720 6816 t
( token is taken to include the longest string of characters which could possibly constitute a)15 3820(ter, the next)2 500 2 720 6936 t
( if a)2 175(token.'' Thus,)1 598 2 720 7056 t
10 CW f
(/)1528 7056 w
10 R f
( character of a token, and the)6 1208(is the first)2 420 2 1623 7056 t
10 CW f
(/)3285 7056 w
10 R f
(is immediately followed by a)4 1202 1 3379 7056 t
10 CW f
(*)4615 7056 w
10 R f
(, the two)2 365 1 4675 7056 t
(characters begin a comment,)3 1141 1 720 7176 t
10 I f
(regardless)1886 7176 w
10 R f
(of any other context.)3 826 1 2333 7176 t
cleartomark
showpage
saveobj restore
%%EndPage: 0 2
%%Page: 1 3
/saveobj save def
mark
3 pagesetup
10 R f
(The following statement looks like it sets)6 1650 1 970 840 t
10 CW f
(y)2645 840 w
10 R f
(to the value of)3 574 1 2730 840 t
10 CW f
(x)3329 840 w
10 R f
(divided by the value pointed to by)6 1366 1 3414 840 t
10 CW f
(p)4805 840 w
10 R f
(:)4865 840 w
10 CW f
( p points at the divisor */;)6 1680( /*)1 360(y = x/*p)2 480 3 1080 1020 t
10 R f
(In fact,)1 283 1 720 1200 t
10 CW f
(/*)1029 1200 w
10 R f
( comment, so the compiler will simply gobble up the program text until the)13 3031(begins a)1 331 2 1175 1200 t
10 CW f
(*/)4564 1200 w
10 R f
(appears.)4711 1200 w
(In other words, the statement just sets)6 1537 1 720 1320 t
10 CW f
(y)2287 1320 w
10 R f
(to the value of)3 589 1 2377 1320 t
10 CW f
(x)2996 1320 w
10 R f
(and doesn't even look at)4 992 1 3086 1320 t
10 CW f
(p)4107 1320 w
10 R f
(. Rewriting this state-)3 873 1 4167 1320 t
(ment as)1 308 1 720 1440 t
10 CW f
( p points at the divisor */;)6 1680( /*)1 360(y = x / *p)4 600 3 1080 1620 t
10 R f
(or even)1 296 1 720 1800 t
10 CW f
( p points at the divisor */;)6 1680( /*)1 360(y = x/\(*p\))2 600 3 1080 1980 t
10 R f
(would cause it to do the division the comment suggests.)9 2233 1 720 2160 t
( C use)2 254( example, older versions of)4 1088( For)1 190(This sort of near-ambiguity can cause trouble in other contexts.)9 2538 4 970 2316 t
10 CW f
(=+)720 2436 w
10 R f
(to mean what present versions mean by)6 1575 1 865 2436 t
10 CW f
(+=)2465 2436 w
10 R f
(. Such a compiler will treat)5 1082 1 2585 2436 t
10 CW f
(a=-1;)1080 2616 w
10 R f
(as meaning the same thing as)5 1168 1 720 2796 t
10 CW f
(a =- 1;)2 420 1 1080 2976 t
10 R f
(or)720 3156 w
10 CW f
(a = a - 1;)4 600 1 1080 3336 t
10 R f
(This will surprise a programmer who intended)6 1853 1 720 3516 t
10 CW f
(a = -1;)2 420 1 1080 3696 t
10 R f
(On the other hand, compilers for these older versions of C would interpret)12 2959 1 970 3912 t
10 CW f
(a=/*b;)1080 4092 w
10 R f
(as)720 4272 w
10 CW f
(a =/ * b ;)4 600 1 1080 4452 t
10 R f
(even though the)2 638 1 720 4632 t
10 CW f
(/*)1383 4632 w
10 R f
(looks like a comment.)3 883 1 1528 4632 t
10 B f
(1.4. Exceptions)1 667 1 720 4908 t
10 R f
(Compound assignment operators such as)4 1632 1 970 5064 t
10 CW f
(+=)2627 5064 w
10 R f
( Thus,)1 275(are really multiple tokens.)3 1043 2 2772 5064 t
10 CW f
(a + /* strange */ = 1)6 1260 1 1080 5244 t
10 R f
(means the same as)3 740 1 720 5424 t
10 CW f
(a += 1)2 360 1 1080 5604 t
10 R f
( In)1 134( look like single tokens are really multiple tokens.)8 2004(These operators are the only cases in which things that)9 2182 3 720 5784 t
(particular,)720 5904 w
10 CW f
(p - > a)3 420 1 1080 6084 t
10 R f
( is)1 92( It)1 111(is illegal.)1 367 3 720 6264 t
10 I f
(not)1315 6264 w
10 R f
(a synonym for)2 577 1 1468 6264 t
10 CW f
(p -> a)2 360 1 1080 6444 t
10 R f
(As another example, the)3 970 1 720 6624 t
10 CW f
(>>)1715 6624 w
10 R f
(operator is a single token, so)5 1143 1 1860 6624 t
10 CW f
(>>=)3028 6624 w
10 R f
(is made up of two tokens, not three.)7 1429 1 3233 6624 t
(On the other hand, those older compilers that still accept)9 2267 1 970 6780 t
10 CW f
(=+)3263 6780 w
10 R f
(as a synonym for)3 688 1 3409 6780 t
10 CW f
(+=)4123 6780 w
10 R f
(treat)4269 6780 w
10 CW f
(=+)4473 6780 w
10 R f
(as a single)2 420 1 4620 6780 t
(token.)720 6900 w
cleartomark
showpage
saveobj restore
%%EndPage: 1 3
%%Page: 2 4
/saveobj save def
mark
4 pagesetup
10 R f
(- 2 -)2 166 1 2797 480 t
10 B f
( and Characters)2 688(1.5. Strings)1 506 2 720 840 t
10 R f
( in which con-)3 586(Single and double quotes mean very different things in C, and there are some contexts)14 3484 2 970 996 t
(fusing them will result in surprises rather than error messages.)9 2485 1 720 1116 t
( integer is the one)4 718( The)1 206(A character enclosed in single quotes is just another way of writing an integer.)13 3146 3 970 1272 t
( in an)2 262( Thus,)1 295( to the given character in the implementation's collating sequence.)9 2831(that corresponds)1 678 4 720 1392 t
8 R f
(ASCII)4831 1392 w
10 R f
(implementation,)720 1512 w
10 CW f
('a')1403 1512 w
10 R f
( string enclosed in double quotes, on)6 1493( A)1 127(means exactly the same thing as 0141 or 97.)8 1807 3 1613 1512 t
( a pointer to a nameless array that has been initialized with the)12 2490(the other hand, is a short-hand way of writing)8 1830 2 720 1632 t
(characters between the quotes and an extra character whose binary value is zero.)12 3208 1 720 1752 t
(The following two program fragments are equivalent:)6 2146 1 970 1908 t
10 CW f
(printf \("Hello world\\n"\);)2 1500 1 1080 2088 t
(char hello[] = {'H', 'e', 'l', 'l', 'o', ' ',)9 2700 1 1080 2328 t
('w', 'o', 'r', 'l', 'd', '\\n', 0};)6 2040 1 1330 2448 t
(printf \(hello\);)1 900 1 1080 2568 t
10 R f
( will often cause a warning message, so using)8 1918(Using a pointer instead of an integer \(or vice versa\))9 2152 2 970 2784 t
( major exception is in function)5 1234( The)1 207(double quotes instead of single quotes \(or vice versa\) is usually caught.)11 2879 3 720 2904 t
( saying)1 286( Thus,)1 275(calls, where most compilers do not check argument types.)8 2313 3 720 3024 t
10 CW f
(printf\('\\n'\);)1080 3204 w
10 R f
(instead of)1 391 1 720 3384 t
10 CW f
(printf \("\\n"\);)1 840 1 1080 3564 t
10 R f
(will usually result in a surprise at run time.)8 1713 1 720 3744 t
( enough to hold several characters, some C compilers permit mul-)10 2672(Because an integer is usually large)5 1398 2 970 3900 t
( means that writing)3 805( This)1 240(tiple characters in a character constant.)5 1610 3 720 4020 t
10 CW f
('yes')3412 4020 w
10 R f
(instead of)1 403 1 3749 4020 t
10 CW f
("yes")4188 4020 w
10 R f
(may well go)2 516 1 4524 4020 t
( latter means ``the address of the first of four consecutive memory locations containing)13 3539(undetected. The)1 666 2 720 4140 t
10 CW f
(y)4955 4140 w
10 R f
(,)5015 4140 w
10 CW f
(e)720 4260 w
10 R f
(,)780 4260 w
10 CW f
(s)835 4260 w
10 R f
( of the values of)4 659( former means ``an integer that is composed)7 1795( The)1 210(, and a null character, respectively.'')5 1481 4 895 4260 t
(the characters)1 561 1 720 4380 t
10 CW f
(y)1317 4380 w
10 R f
(,)1377 4380 w
10 CW f
(e)1438 4380 w
10 R f
(, and)1 205 1 1498 4380 t
10 CW f
(s)1740 4380 w
10 R f
( similarity between these two)4 1219( Any)1 234(in some implementation-defined manner.'')3 1750 3 1837 4380 t
(quantities is purely coincidental.)3 1299 1 720 4500 t
10 B f
( Pitfalls)1 325(2. Syntactic)1 519 2 720 4740 t
10 R f
( must also)2 410( One)1 218( it is not enough to understand the tokens that make it up.)12 2315(To understand a C program,)4 1127 4 970 4896 t
( While)1 306( expressions, statements, and programs.)4 1632(understand how the tokens combine to form declarations,)7 2382 3 720 5016 t
(these combinations are usually well-defined, the definitions are sometimes counter-intuitive or confusing.)11 4225 1 720 5136 t
(In this section, we look at some syntactic constructions that are less than obvious.)13 3261 1 970 5292 t
10 B f
( Declarations)1 563(2.1. Understanding)1 840 2 720 5532 t
10 R f
( writing a C program that was going to run stand-alone in a small)13 2671(I once talked to someone who was)6 1399 2 970 5688 t
( was switched on, the hardware would call the subroutine whose)10 2790( this machine)2 579(microprocessor. When)1 951 3 720 5808 t
(address was stored in location 0.)5 1298 1 720 5928 t
( on, we had to devise a C statement that would call this subroutine)13 2687(In order to simulate turning power)5 1383 2 970 6084 t
( some thought, we came up with the following:)8 1885(explicitly. After)1 663 2 720 6204 t
10 CW f
(\(*\(void\(*\)\(\)\)0\)\(\);)1080 6384 w
10 R f
( though, because)2 676( needn't,)1 352( They)1 258(Expressions like these strike terror into the hearts of C programmers.)10 2784 4 970 6600 t
( with the help of a single, simple rule:)8 1513(they can usually be constructed quite easily)6 1742 2 720 6720 t
10 I f
(declare it the way you use)5 1040 1 4000 6720 t
(it.)720 6840 w
10 R f
( a list of stylized expressions that are expected)8 1884(Every C variable declaration has two parts: a type and)9 2186 2 970 6996 t
( simplest such expression is a variable:)6 1554( The)1 205(to evaluate to that type.)4 935 3 720 7116 t
cleartomark
showpage
saveobj restore
%%EndPage: 2 4
%%Page: 3 5
/saveobj save def
mark
5 pagesetup
10 R f
(- 3 -)2 166 1 2797 480 t
10 CW f
(float f, g;)2 660 1 1080 900 t
10 R f
(indicates that the expressions)3 1177 1 720 1080 t
10 CW f
(f)1925 1080 w
10 R f
(and)2013 1080 w
10 CW f
(g)2186 1080 w
10 R f
(, when evaluated, will be of type)6 1327 1 2246 1080 t
10 CW f
(float)3602 1080 w
10 R f
(. Because the thing declared)4 1138 1 3902 1080 t
(is an expression, parentheses may be used freely:)7 1962 1 720 1200 t
10 CW f
(float \(\(f\)\);)1 720 1 1080 1380 t
10 R f
(means that)1 430 1 720 1560 t
10 CW f
(\(\(f\)\))1175 1560 w
10 R f
(evaluates to a)2 543 1 1500 1560 t
10 CW f
(float)2068 1560 w
10 R f
(and therefore, by inference, that)4 1273 1 2393 1560 t
10 CW f
(f)3691 1560 w
10 R f
(is also a)2 322 1 3776 1560 t
10 CW f
(float)4123 1560 w
10 R f
(.)4423 1560 w
( example,)1 388( For)1 189(Similar logic applies to function and pointer types.)7 2027 3 970 1716 t
10 CW f
(float ff\(\);)1 660 1 1080 1896 t
10 R f
(means that the expression)3 1065 1 720 2076 t
10 CW f
(ff\(\))1822 2076 w
10 R f
(is a)1 148 1 2099 2076 t
10 CW f
(float)2284 2076 w
10 R f
(, and therefore that)3 792 1 2584 2076 t
10 CW f
(ff)3414 2076 w
10 R f
(is a function that returns a)5 1105 1 3572 2076 t
10 CW f
(float)4715 2076 w
10 R f
(.)5015 2076 w
(Analogously,)720 2196 w
10 CW f
(float *pf;)1 600 1 1080 2376 t
10 R f
(means that)1 430 1 720 2556 t
10 CW f
(*pf)1175 2556 w
10 R f
(is a)1 136 1 1380 2556 t
10 CW f
(float)1541 2556 w
10 R f
(and therefore that)2 703 1 1866 2556 t
10 CW f
(pf)2594 2556 w
10 R f
(is a pointer to a)4 616 1 2739 2556 t
10 CW f
(float)3380 2556 w
10 R f
(.)3680 2556 w
( Thus)1 250(These forms combine in declarations the same way they do in expressions.)11 2984 2 970 2712 t
10 CW f
(float *g\(\), \(*h\)\(\);)2 1140 1 1080 2892 t
10 R f
(says that)1 355 1 720 3072 t
10 CW f
(*g\(\))1108 3072 w
10 R f
(and)1381 3072 w
10 CW f
(\(*h\)\(\))1558 3072 w
10 R f
(are)1951 3072 w
10 CW f
(float)2106 3072 w
10 R f
(expressions. Since)1 772 1 2440 3072 t
10 CW f
(\(\))3246 3072 w
10 R f
(binds more tightly than)3 958 1 3400 3072 t
10 CW f
(*)4392 3072 w
10 R f
(,)4452 3072 w
10 CW f
(*g\(\))4511 3072 w
10 R f
(means)4785 3072 w
(the same thing as)3 706 1 720 3192 t
10 CW f
(*\(g\(\)\))1456 3192 w
10 R f
(:)1816 3192 w
10 CW f
(g)1874 3192 w
10 R f
(is a function that returns a pointer to a)8 1560 1 1964 3192 t
10 CW f
(float)3553 3192 w
10 R f
(, and)1 198 1 3853 3192 t
10 CW f
(h)4080 3192 w
10 R f
(is a pointer to a func-)5 871 1 4169 3192 t
(tion that returns a)3 702 1 720 3312 t
10 CW f
(float)1447 3312 w
10 R f
(.)1747 3312 w
( know how to declare a variable of a given type, it is easy to write a cast for that type: just)21 3714(Once we)1 356 2 970 3468 t
( and enclose the whole thing in parenthe-)7 1665(remove the variable name and the semicolon from the declaration)9 2655 2 720 3588 t
( since)1 230(ses. Thus,)1 422 2 720 3708 t
10 CW f
(float *g\(\);)1 660 1 1080 3888 t
10 R f
(declares)720 4068 w
10 CW f
(g)1071 4068 w
10 R f
(to be a function returning a pointer to a)8 1564 1 1156 4068 t
10 CW f
(float)2745 4068 w
10 R f
(,)3045 4068 w
10 CW f
(\(float *\(\)\))1 660 1 3095 4068 t
10 R f
(is a cast to that type.)5 816 1 3780 4068 t
(Armed with this knowledge, we are now prepared to tackle)9 2442 1 970 4224 t
10 CW f
(\(*\(void\(*\)\(\)\)0\)\(\))3446 4224 w
10 R f
(. We can ana-)3 574 1 4466 4224 t
( suppose that we have a variable)6 1321( First,)1 264( statement in two parts.)4 950(lyze this)1 342 4 720 4344 t
10 CW f
(fp)3627 4344 w
10 R f
(that contains a function pointer)4 1263 1 3777 4344 t
(and we want to call the function to which)8 1653 1 720 4464 t
10 CW f
(fp)2398 4464 w
10 R f
( is done this way:)4 700(points. That)1 503 2 2543 4464 t
10 CW f
(\(*fp\)\(\);)1080 4644 w
10 R f
(If)720 4824 w
10 CW f
(fp)813 4824 w
10 R f
(is a pointer to a function,)5 1009 1 960 4824 t
10 CW f
(*fp)1996 4824 w
10 R f
( so)1 117(is the function itself,)3 828 2 2203 4824 t
10 CW f
(\(*fp\)\(\))3176 4824 w
10 R f
( paren-)1 282( The)1 208(is the way to invoke it.)5 926 3 3624 4824 t
(theses in)1 361 1 720 4944 t
10 CW f
(\(*fp\))1120 4944 w
10 R f
( would otherwise be interpreted as)5 1437(are essential because the expression)4 1485 2 1459 4944 t
10 CW f
(*\(fp\(\)\))4419 4944 w
10 R f
(. We)1 201 1 4839 4944 t
(have now reduced the problem to that of finding an appropriate expression to replace)13 3395 1 720 5064 t
10 CW f
(fp)4140 5064 w
10 R f
(.)4260 5064 w
( mind about types, we could)5 1197( C could read our)4 745( If)1 129(This problem is the second part of our analysis.)8 1999 4 970 5220 t
(write:)720 5340 w
10 CW f
(\(*0\)\(\);)1080 5520 w
10 R f
( the)1 164(This doesn't work because)3 1115 2 720 5700 t
10 CW f
(*)2041 5700 w
10 R f
( the)1 164( Furthermore,)1 591(operator insists on having a pointer as its operand.)8 2142 3 2143 5700 t
(operand must be a pointer to a function so that the result of)12 2394 1 720 5820 t
10 CW f
(*)3142 5820 w
10 R f
( into a)2 254( we need to cast 0)5 727( Thus,)1 278(can be called.)2 551 4 3230 5820 t
(type loosely described as ``pointer to function returning void.'')8 2521 1 720 5940 t
(If)970 6096 w
10 CW f
(fp)1072 6096 w
10 R f
( to a function returning void, then)6 1418(is a pointer)2 466 2 1228 6096 t
10 CW f
(\(*fp\)\(\))3149 6096 w
10 R f
(is a void value, and its declaration)6 1434 1 3606 6096 t
(would look like this:)3 826 1 720 6216 t
10 CW f
(void \(*fp\)\(\);)1 780 1 1080 6396 t
10 R f
(Thus, we could write:)3 871 1 720 6576 t
10 CW f
(void \(*fp\)\(\);)1 780 1 1080 6756 t
(\(*fp\)\(\);)1080 6876 w
10 R f
( the variable, we know how to)6 1218( once we know how to declare)6 1219( But)1 196(at the cost of declaring a dummy variable.)7 1687 4 720 7056 t
( we cast 0 to a ``pointer)6 954( Thus,)1 277( the name from the variable declaration.)6 1605(cast a constant to that type: just drop)7 1484 4 720 7176 t
(to function returning void'' by saying:)5 1535 1 720 7296 t
cleartomark
showpage
saveobj restore
%%EndPage: 3 5
%%Page: 4 6
/saveobj save def
mark
6 pagesetup
10 R f
(- 4 -)2 166 1 2797 480 t
10 CW f
(\(void\(*\)\(\)\)0)1080 900 w
10 R f
(and we can now replace)4 957 1 720 1080 t
10 CW f
(fp)1702 1080 w
10 R f
(by)1847 1080 w
10 CW f
(\(void\(*\)\(\)\)0)1972 1080 w
10 R f
(:)2692 1080 w
10 CW f
(\(*\(void\(*\)\(\)\)0\)\(\);)1080 1260 w
10 R f
(The semicolon on the end turns the expression into a statement.)10 2539 1 720 1440 t
( time we tackled this problem, there was no such thing as a)12 2451(At the)1 254 2 970 1596 t
10 CW f
(typedef)3708 1596 w
10 R f
( it,)1 114(declaration. Using)1 765 2 4161 1596 t
(we could have solved the problem more clearly:)7 1921 1 720 1716 t
10 CW f
(typedef void \(*funcptr\)\(\);)2 1560 1 1080 1896 t
(\(* \(funcptr\) 0\)\(\);)2 1080 1 1080 2016 t
10 B f
( Don't Always Have the Precedence You Want)7 1983(2.2. Operators)1 638 2 720 2316 t
10 R f
(Suppose that the manifest constant)4 1428 1 970 2472 t
10 CW f
(FLAG)2433 2472 w
10 R f
( integer with exactly one bit turned on in its binary)10 2136(is an)1 196 2 2708 2472 t
( the integer variable)3 813(representation \(in other words, a power of two\), and you want to test whether)13 3176 2 720 2592 t
10 CW f
(flags)4740 2592 w
10 R f
( usual way to write this is:)6 1050( The)1 205(has that bit turned on.)4 869 3 720 2712 t
10 CW f
(if \(flags & FLAG\) ...)4 1260 1 1080 2892 t
10 R f
(The meaning of this is plain to most C programmers: an)10 2308 1 720 3072 t
10 CW f
(if)3060 3072 w
10 R f
( whether the expression in the)5 1235(statement tests)1 593 2 3212 3072 t
( might be nice to make this test more explicit for documentation pur-)12 2837( It)1 118(parentheses evaluates to 0 or not.)5 1365 3 720 3192 t
(poses:)720 3312 w
10 CW f
(if \(flags & FLAG != 0\) ...)6 1560 1 1080 3492 t
10 R f
( is also wrong, because)4 947( It)1 117(The statement is now easier to understand.)6 1736 3 720 3672 t
10 CW f
(!=)3551 3672 w
10 R f
(binds more tightly than)3 949 1 3702 3672 t
10 CW f
(&)4682 3672 w
10 R f
(, so the)2 298 1 4742 3672 t
(interpretation is now:)2 855 1 720 3792 t
10 CW f
(if \(flags & \(FLAG != 0\)\) ...)6 1680 1 1080 3972 t
10 R f
(This will work \(by coincidence\) if)5 1367 1 720 4152 t
10 CW f
(FLAG)2112 4152 w
10 R f
(is 1 or 0 \(!\), but not for any other power of two.*)12 1952 1 2377 4152 t
(Suppose you have two integer variables,)5 1624 1 970 4308 t
10 CW f
(h)2621 4308 w
10 R f
(and)2709 4308 w
10 CW f
(l,)2881 4308 w
10 R f
(whose values are between 0 and 15 inclusive, and)8 2011 1 3029 4308 t
(you want to set)3 617 1 720 4428 t
10 CW f
(r)1365 4428 w
10 R f
(to an 8-bit value whose low-order bits are those of)9 2037 1 1453 4428 t
10 CW f
(l)3518 4428 w
10 R f
( bits are those)3 558(and whose high-order)2 876 2 3606 4428 t
(of)720 4548 w
10 CW f
(h.)828 4548 w
10 R f
(The natural way to do this is to write:)8 1499 1 973 4548 t
10 CW f
(r = h<<4 + l;)4 780 1 1080 4728 t
10 R f
( binds more tightly than shifting, so this example is equivalent to)11 2595( Addition)1 406(Unfortunately, this is wrong.)3 1152 3 720 4908 t
10 CW f
(r = h << \(4 + l\);)6 1020 1 1080 5088 t
10 R f
(Here are two ways to get it right:)7 1317 1 720 5268 t
10 CW f
(r = \(h << 4\) + l;)6 1020 1 1080 5448 t
(r = h << 4 | l;)6 900 1 1080 5568 t
10 R f
(One way to avoid these problems is to parenthesize everything, but expressions with too many paren-)15 4070 1 970 5784 t
(theses are hard to understand, so it is probably useful to try to remember the precedence levels in C.)18 3987 1 720 5904 t
( can be made easier,)4 865( It)1 126(Unfortunately, there are fifteen of them, so this is not always easy to do.)13 3079 3 970 6060 t
(though, by classifying them into groups.)5 1614 1 720 6180 t
( that bind the most tightly are the ones that aren't really operators: subscripting, func-)14 3513(The operators)1 557 2 970 6336 t
( all associate to the left.)5 943( These)1 288(tion calls, and structure selection.)4 1337 3 720 6456 t
( highest precedence of any of the true operators.)8 2055( have the)2 392( These)1 304(Next come the unary operators.)4 1319 4 970 6612 t
(Because function calls bind more tightly than unary operators, you must write)11 3201 1 720 6732 t
10 CW f
(\(*p\)\(\))3953 6732 w
10 R f
(to call a function)3 695 1 4345 6732 t
(pointed to by)2 548 1 720 6852 t
10 CW f
(p)1303 6852 w
10 R f
(;)1363 6852 w
10 CW f
(*p\(\))1427 6852 w
10 R f
(implies that)1 481 1 1703 6852 t
10 CW f
(p)2220 6852 w
10 R f
( are unary operators and)4 1007( Casts)1 278(is a function that returns a pointer.)6 1439 3 2316 6852 t
( operators are right-associative, so)4 1398( Unary)1 307(have the same precedence as any other unary operator.)8 2242 3 720 6972 t
10 CW f
(*p++)4700 6972 w
10 R f
(is)4973 6972 w
8 S1 f
(__________________)720 7072 w
8 R f
(* Recall that the result of)5 801 1 720 7172 t
8 CW f
(!=)1541 7172 w
8 R f
(is always either 1 or 0.)5 720 1 1657 7172 t
cleartomark
showpage
saveobj restore
%%EndPage: 4 6
%%Page: 5 7
/saveobj save def
mark
7 pagesetup
10 R f
(- 5 -)2 166 1 2797 480 t
(interpreted as)1 540 1 720 840 t
10 CW f
(*\(p++\))1285 840 w
10 R f
(and not as)2 405 1 1670 840 t
10 CW f
(\(*p\)++)2100 840 w
10 R f
(.)2460 840 w
( the highest precedence, then the)5 1327( arithmetic operators have)3 1051( The)1 209(Next come the true binary operators.)5 1483 4 970 996 t
( finally the con-)3 641(shift operators, the relational operators, the logical operators, the assignment operators, and)11 3679 2 720 1116 t
( two most important things to keep in mind are:)9 1903( The)1 205(ditional operator.)1 688 3 720 1236 t
( logical operator has lower precedence than every relational operator.)9 2763(1. Every)1 488 2 720 1392 t
( less tightly than the arithmetic)5 1261( shift operators bind more tightly than the relational operators but)10 2654(2. The)1 405 3 720 1548 t
(operators.)970 1668 w
( and remainder)2 612( division,)1 379( Multiplication,)1 654(Within the various operator classes, there are few surprises.)8 2425 4 970 1824 t
(have the same precedence, addition and subtraction have the same precedence, and the two shift operators)15 4320 1 720 1944 t
(have the same precedence.)3 1062 1 720 2064 t
( that the six relational operators do not all have the same precedence:)12 2767(One small surprise is)3 841 2 970 2220 t
10 CW f
(==)4604 2220 w
10 R f
(and)4750 2220 w
10 CW f
(!=)4920 2220 w
10 R f
( allows us, for instance, to see if)7 1298( This)1 230(bind less tightly than the other relational operators.)7 2051 3 720 2340 t
10 CW f
(a)4325 2340 w
10 R f
(and)4411 2340 w
10 CW f
(b)4581 2340 w
10 R f
(are in the)2 373 1 4667 2340 t
(same relative order as)3 872 1 720 2460 t
10 CW f
(c)1617 2460 w
10 R f
(and)1702 2460 w
10 CW f
(d)1871 2460 w
10 R f
(by the expression)2 699 1 1956 2460 t
10 CW f
(a < b == c < d)6 840 1 1080 2640 t
10 R f
( bitwise operators all bind more)5 1293( The)1 210( precedence.)1 502(Within the logical operators, no two have the same)8 2065 4 970 2856 t
( sequential operators, each)3 1064(tightly than the)2 612 2 720 2976 t
10 I f
(and)2423 2976 w
10 R f
(operator binds more tightly than the corresponding)6 2043 1 2600 2976 t
10 I f
(or)4670 2976 w
10 R f
(opera-)4786 2976 w
(tor, and the bitwise)3 766 1 720 3096 t
10 I f
(exclusive or)1 479 1 1511 3096 t
10 R f
(operator \()1 390 1 2015 3096 t
10 CW f
(\303)2405 3096 w
10 R f
(\) falls between bitwise)3 901 1 2465 3096 t
10 I f
(and)3391 3096 w
10 R f
(and bitwise)1 458 1 3566 3096 t
10 I f
(or)4049 3096 w
10 R f
(.)4138 3096 w
( per-)1 188( This)1 231( precedence than any we have mentioned so far.)8 1937(The ternary conditional operator has lower)5 1714 4 970 3252 t
(mits the selection expression to contain logical combinations of relational operators, as in)12 3577 1 720 3372 t
10 CW f
(z = a < b && b < c ? d : e)12 1560 1 1080 3552 t
10 R f
( than the con-)3 552(This example also shows that it makes sense for assignment to have a lower precedence)14 3518 2 970 3768 t
( have the same precedence and they all)7 1560( all the compound assignment operators)5 1600( Moreover,)1 470(ditional operator.)1 690 4 720 3888 t
(group right to left, so that)5 1022 1 720 4008 t
10 CW f
(a = b = c)4 540 1 1080 4188 t
10 R f
(means the same as)3 740 1 720 4368 t
10 CW f
(b = c; a = b;)5 780 1 1080 4548 t
10 R f
( is easy to remember because the comma is often used as a)12 2363( This)1 230( is the comma operator.)4 948(Lowest of all)2 529 4 970 4764 t
(substitute for the semicolon when an expression is required instead of a statement.)12 3287 1 720 4884 t
( for example, the fol-)4 849( Consider,)1 437(Assignment is another operator often involved in precedence mixups.)8 2784 3 970 5040 t
(lowing loop intended to copy one file to another:)8 1954 1 720 5160 t
10 CW f
(while \(c=getc\(in\) != EOF\))3 1500 1 1080 5340 t
(putc\(c,out\);)1330 5460 w
10 R f
(The way the expression in the)5 1195 1 720 5640 t
10 CW f
(while)1940 5640 w
10 R f
(statement is written makes it look like)6 1522 1 2265 5640 t
10 CW f
(c)3812 5640 w
10 R f
(should be assigned the value)4 1143 1 3897 5640 t
(of)720 5760 w
10 CW f
(getc\(in\))829 5760 w
10 R f
(and then compared with)3 965 1 1335 5760 t
10 CW f
(EOF)2325 5760 w
10 R f
( assignment has lower prece-)4 1158( Unhappily,)1 497(to terminate the loop.)3 855 3 2530 5760 t
(dence than any comparison operator, so the value of)8 2185 1 720 5880 t
10 CW f
(c)2943 5880 w
10 R f
(will be the result of comparing)5 1294 1 3041 5880 t
10 CW f
(getc\(in\))4374 5880 w
10 R f
(, the)1 186 1 4854 5880 t
(value of which is then discarded, and)6 1531 1 720 6000 t
10 CW f
(EOF)2284 6000 w
10 R f
( of the file will consist of a stream of bytes)10 1779(. Thus, the ``copy'')3 797 2 2464 6000 t
(whose value is 1.)3 688 1 720 6120 t
(It is not too hard to see that the example above should be written:)13 2611 1 970 6276 t
10 CW f
(while \(\(c=getc\(in\)\) != EOF\))3 1620 1 1080 6456 t
(putc\(c,out\);)1330 6576 w
10 R f
( example, several ver-)3 883( For)1 190( to spot in more complicated expressions.)6 1663(However, errors of this sort can be hard)7 1584 4 720 6756 t
(sions of the)2 461 1 720 6876 t
10 I f
(lint)1206 6876 w
10 R f
(program distributed with the)3 1141 1 1365 6876 t
9 R f
(UNIX)2529 6876 w
10 S f
(\322)2754 6876 w
10 R f
(system have the following erroneous line:)5 1673 1 2858 6876 t
10 CW f
(if\( \(t=BTYPE\(pt1->aty\)==STRTY\) || t==UNIONTY \){)4 2820 1 1330 7056 t
10 R f
( to)1 115(This was intended to assign a value)6 1481 2 720 7236 t
10 CW f
(t)2353 7236 w
10 R f
(and then see if)3 615 1 2450 7236 t
10 CW f
(t)3102 7236 w
10 R f
(is equal to)2 435 1 3199 7236 t
10 CW f
(STRTY)3671 7236 w
10 R f
(or)4008 7236 w
10 CW f
(UNIONTY)4128 7236 w
10 R f
(. The actual)2 492 1 4548 7236 t
cleartomark
showpage
saveobj restore
%%EndPage: 5 7
%%Page: 6 8
/saveobj save def
mark
8 pagesetup
10 R f
(- 6 -)2 166 1 2797 480 t
(effect is quite different.*)3 986 1 720 840 t
( the predecessor of)3 771( B,)1 149( for historical reasons.)3 908(The precedence of the C logical operators comes about)8 2242 4 970 996 t
(C, had logical operators that corresponded rougly to C's)8 2268 1 720 1116 t
10 CW f
(&)3016 1116 w
10 R f
(and)3104 1116 w
10 CW f
(|)3276 1116 w
10 R f
( they were defined to)4 850(operators. Although)1 826 2 3364 1116 t
(act on bits, the compiler would treat them as)8 1837 1 720 1236 t
10 CW f
(&&)2590 1236 w
10 R f
(and)2743 1236 w
10 CW f
(||)2920 1236 w
10 R f
( the)1 156( When)1 296(if they were in a conditional context.)6 1515 3 3073 1236 t
(two usages were split apart in C, it was deemed too dangerous to change the precedence much.**)16 3886 1 720 1356 t
10 B f
( Those Semicolons!)2 817(2.3. Watch)1 483 2 720 1596 t
10 R f
( difference: either it is a null statement, which)8 1849(An extra semicolon in a C program usually makes little)9 2221 2 970 1752 t
( One)1 226( easy to remove.)3 684(has no effect, or it elicits a diagnostic message from the compiler, which makes it)14 3410 3 720 1872 t
(important exception is after an)4 1268 1 720 1992 t
10 CW f
(if)2025 1992 w
10 R f
(or)2182 1992 w
10 CW f
(while)2302 1992 w
10 R f
( exactly one statement.)3 954(clause, which must be followed by)5 1447 2 2639 1992 t
(Consider this example:)2 922 1 720 2112 t
10 CW f
(if \(x[i] > big\);)3 960 1 1080 2292 t
(big = x[i];)2 660 1 1330 2412 t
10 R f
( this program fragment means something)5 1700(The semicolon on the first line will not upset the compiler, but)11 2620 2 720 2592 t
(quite different from:)2 815 1 720 2712 t
10 CW f
(if \(x[i] > big\))3 900 1 1080 2892 t
(big = x[i];)2 660 1 1330 3012 t
10 R f
(The first one is equivalent to:)5 1174 1 720 3192 t
10 CW f
( })1 120( {)1 180(if \(x[i] > big\))3 900 3 1080 3372 t
(big = x[i];)2 660 1 1080 3492 t
10 R f
(which is, of course, equivalent to:)5 1351 1 720 3672 t
10 CW f
(big = x[i];)2 660 1 1080 3852 t
10 R f
(\(unless)720 4032 w
10 CW f
(x)1028 4032 w
10 R f
(,)1088 4032 w
10 CW f
(i)1138 4032 w
10 R f
(, or)1 133 1 1198 4032 t
10 CW f
(big)1356 4032 w
10 R f
(is a macro with side effects\).)5 1147 1 1561 4032 t
( difference is at the end of a declaration just before a)11 2162(Another place that a semicolon can make a big)8 1908 2 970 4188 t
( the following fragment:)3 974( Consider)1 411(function definition.)1 772 3 720 4308 t
10 CW f
(struct foo {)2 720 1 1080 4488 t
(int x;)1 360 1 1330 4608 t
(})1080 4728 w
(f\(\))1080 4968 w
({)1080 5088 w
(. . .)2 300 1 1330 5208 t
(})1080 5328 w
10 R f
(There is a semicolon missing between the first)7 1863 1 720 5508 t
10 CW f
(})2609 5508 w
10 R f
(and the)1 292 1 2695 5508 t
10 CW f
(f)3013 5508 w
10 R f
( is)1 94( effect of this)3 532( The)1 206(that immediately follows it.)3 1109 4 3099 5508 t
(to declare that the function)4 1102 1 720 5628 t
10 CW f
(f)1854 5628 w
10 R f
(returns a)1 353 1 1946 5628 t
10 CW f
(struct foo)1 607 1 2331 5628 t
10 R f
( the)1 154( If)1 123(, which is defined as part of this declaration.)8 1825 3 2938 5628 t
(semicolon were present,)2 967 1 720 5748 t
10 CW f
(f)1712 5748 w
10 R f
(would be defined by default as returning an integer.\262)8 2115 1 1797 5748 t
10 B f
( Switch Statement)2 771(2.4. The)1 367 2 720 5988 t
10 R f
( in that the cases in its)6 895(C is unusual)2 497 2 970 6144 t
10 CW f
(switch)2389 6144 w
10 R f
( for exam-)2 419( Consider,)1 438(statement can flow into each other.)5 1407 3 2776 6144 t
(ple, the following program fragments in C and Pascal:)8 2167 1 720 6264 t
8 S1 f
(__________________)720 6780 w
8 R f
(* Thanks to Guy Harris for pointing this out to me.)10 1627 1 720 6880 t
(** Dennis Ritchie and Steve Johnson both pointed this out to me.)11 2087 1 720 6980 t
(\262 Thanks to an anonymous benefactor for this one.)8 1616 1 720 7080 t
cleartomark
showpage
saveobj restore
%%EndPage: 6 8
%%Page: 7 9
/saveobj save def
mark
9 pagesetup
10 R f
(- 7 -)2 166 1 2797 480 t
10 CW f
(switch \(color\) {)2 960 1 1080 900 t
(case 1: printf \("red"\);)3 1380 1 1080 1020 t
(break;)1560 1140 w
(case 2: printf \("yellow"\);)3 1560 1 1080 1260 t
(break;)1560 1380 w
(case 3: printf \("blue"\);)3 1440 1 1080 1500 t
(break;)1560 1620 w
(})1080 1740 w
(case color of)2 780 1 1080 1980 t
( \('red'\);)1 540(1: write)1 550 2 1080 2100 t
( \('yellow'\);)1 720(2: write)1 550 2 1080 2220 t
( \('blue'\))1 540(3: write)1 550 2 1080 2340 t
(end)1080 2460 w
10 R f
(Both these program fragments do the same thing: print)8 2243 1 970 2676 t
10 CW f
(red)3246 2676 w
10 R f
(,)3426 2676 w
10 CW f
(yellow)3484 2676 w
10 R f
(, or)1 141 1 3844 2676 t
10 CW f
(blue)4018 2676 w
10 R f
(\(without starting a)2 749 1 4291 2676 t
(new line\), depending on whether the variable)6 1810 1 720 2796 t
10 CW f
(color)2556 2796 w
10 R f
( exactly anal-)2 537( program fragments are)3 936( The)1 206(is 1, 2, or 3.)4 479 4 2882 2796 t
( does not have any part that corresponds to the C)10 2057(ogous, with one exception: the Pascal program)6 1927 2 720 2916 t
10 CW f
(break)4740 2916 w
10 R f
( unimpeded)1 477( reason for that is that case labels in C behave as true labels: control can flow)16 3221(statement. The)1 622 3 720 3036 t
(right through a case label.)4 1034 1 720 3156 t
(Looking at it another way, suppose the C fragment looked more like the Pascal fragment:)14 3570 1 970 3312 t
10 CW f
(switch \(color\) {)2 960 1 1080 3492 t
( \("red"\);)1 540( printf)1 440(case 1:)1 420 3 1080 3612 t
( \("yellow"\);)1 720( printf)1 440(case 2:)1 420 3 1080 3732 t
( \("blue"\);)1 600( printf)1 440(case 3:)1 420 3 1080 3852 t
(})1080 3972 w
10 R f
(and suppose further that)3 980 1 720 4152 t
10 CW f
(color)1732 4152 w
10 R f
( the program would print)4 1027( Then,)1 287(were equal to 2.)3 658 3 2064 4152 t
10 CW f
(yellowblue)4068 4152 w
10 R f
(, because)1 372 1 4668 4152 t
(control would pass naturally from the second)6 1803 1 720 4272 t
10 CW f
(printf)2548 4272 w
10 R f
(call to the statement after it.)5 1115 1 2933 4272 t
(This is both a strength and a weakness of C)9 1806 1 970 4428 t
10 CW f
(switch)2809 4428 w
10 R f
( is a weakness because leaving)5 1272(statements. It)1 566 2 3202 4428 t
(out a)1 201 1 720 4548 t
10 CW f
(break)950 4548 w
10 R f
( strength)1 350( is a)2 169( It)1 115(statement is easy to do, and often gives rise to obscure program misbehavior.)12 3127 4 1279 4548 t
( a)1 76(because by leaving out)3 930 2 720 4668 t
10 CW f
(break)1758 4668 w
10 R f
(statement deliberately, one can readily express a control structure that is)10 2950 1 2090 4668 t
( in large)2 347( Specifically,)1 562(inconvenient to implement otherwise.)3 1540 3 720 4788 t
10 CW f
(switch)3204 4788 w
10 R f
( that the)2 340(statements, one often finds)3 1101 2 3599 4788 t
( of the cases reduces to some other case after a relatively small amount of special han-)16 3569(processing for one)2 751 2 720 4908 t
(dling.)720 5028 w
( a)1 76( Such)1 256(For example, consider a program that is an interpreter for some kind of imaginary machine.)14 3738 3 970 5184 t
(program might contain a)3 1030 1 720 5304 t
10 CW f
(switch)1790 5304 w
10 R f
( a)1 83( such)1 223( On)1 187(statement to handle each of the various operation codes.)8 2357 4 2190 5304 t
( to an add operation after the sign of the second)10 1916(machine, it is often true that a subtract operation is identical)10 2404 2 720 5424 t
( it is nice to be able to write something like this:)11 1925( Thus,)1 275(operand has been inverted.)3 1069 3 720 5544 t
10 CW f
(case SUBTRACT:)1 840 1 1330 5724 t
(opnd2 = -opnd2;)2 900 1 1580 5844 t
(/* no break */)3 840 1 1580 5964 t
(case ADD:)1 540 1 1330 6084 t
(. . .)2 300 1 1580 6204 t
10 R f
( white space while looking for a token.)7 1574(As another example, consider the part of a compiler that skips)10 2496 2 970 6420 t
( treat spaces, tabs, and newlines identically except that a newline should cause a)13 3300(Here, one would want to)4 1020 2 720 6540 t
(line counter to be incremented:)4 1242 1 720 6660 t
cleartomark
showpage
saveobj restore
%%EndPage: 7 9
%%Page: 8 10
/saveobj save def
mark
10 pagesetup
10 R f
(- 8 -)2 166 1 2797 480 t
10 CW f
(case '\\n':)1 600 1 1330 900 t
(linecount++;)1580 1020 w
(/* no break */)3 840 1 1580 1140 t
(case '\\t':)1 600 1 1330 1260 t
(case ' ':)2 540 1 1330 1380 t
(. . .)2 300 1 1580 1500 t
10 B f
( Functions)1 448(2.5. Calling)1 512 2 720 1800 t
10 R f
( languages, C requires a function call to have an argument list, even)12 2754(Unlike some other programming)3 1316 2 970 1956 t
( if)1 86( Thus,)1 275(if there are no arguments.)4 1022 3 720 2076 t
10 CW f
(f)2128 2076 w
10 R f
(is a function,)2 519 1 2213 2076 t
10 CW f
(f\(\);)1080 2256 w
10 R f
(is a statement that calls the function, but)7 1610 1 720 2436 t
10 CW f
(f;)1080 2616 w
10 R f
( precisely, it evaluates the address of the function, but does not call it.*)13 2835( More)1 266(does nothing at all.)3 761 3 720 2796 t
10 B f
( Dangling)1 415(2.6. The)1 367 2 720 3036 t
10 CW f
(else)1527 3036 w
10 B f
(Problem)1792 3036 w
10 R f
(We would be remiss in leaving any discussion of syntactic pitfalls without mentioning this one.)14 4070 1 970 3192 t
(Although it is not unique to C, it has bitten C programmers with many years of experience.)16 3636 1 720 3312 t
(Consider the following program fragment:)4 1698 1 970 3468 t
10 CW f
(if \(x == 0\))3 660 1 1080 3648 t
(if \(y == 0\) error\(\);)4 1200 1 1330 3768 t
(else {)1 360 1 1080 3888 t
(z = x + y;)4 600 1 1330 4008 t
(f \(&z\);)1 420 1 1330 4128 t
(})1080 4248 w
10 R f
( main cases:)2 500(The programmer's intention for this fragment is that there should be two)11 2954 2 970 4464 t
10 I f
(x)4455 4464 w
10 S f
(=)4523 4464 w
10 R f
(0 and)1 225 1 4594 4464 t
10 I f
(x)4850 4464 w
10 S f
(\271)4902 4464 w
10 R f
(0.)4965 4464 w
(In the first case, the fragment should do nothing at all unless)11 2469 1 720 4584 t
10 I f
(y)3219 4584 w
10 S f
(=)3287 4584 w
10 R f
(0, in which case it should call)6 1215 1 3358 4584 t
10 CW f
(error)4603 4584 w
10 R f
(. In)1 137 1 4903 4584 t
(the second case, the program should set)6 1583 1 720 4704 t
10 I f
(z)2328 4704 w
10 S f
(=)2391 4704 w
10 I f
(x)2462 4704 w
10 S f
(+)2530 4704 w
10 I f
(y)2601 4704 w
10 R f
(and then call)2 510 1 2670 4704 t
10 I f
(f)3205 4704 w
10 R f
(with the address of)3 757 1 3258 4704 t
10 I f
(z)4040 4704 w
10 R f
(as its argument.)2 630 1 4104 4704 t
( reason is the rule that)5 914( The)1 212( quite different.)2 632(However, the program fragment actually does something)6 2312 4 970 4860 t
(an)720 4980 w
10 CW f
(else)842 4980 w
10 R f
(is always associated with the closest unmatched)6 1932 1 1110 4980 t
10 CW f
(if)3070 4980 w
10 R f
( to indent this fragment the way it)7 1366(. If we were)3 484 2 3190 4980 t
(is actually executed, it would look like this:)7 1744 1 720 5100 t
10 CW f
(if \(x == 0\) {)4 780 1 1080 5280 t
(if \(y == 0\))3 660 1 1330 5400 t
(error\(\);)1580 5520 w
(else {)1 360 1 1330 5640 t
(z = x + y;)4 600 1 1580 5760 t
(f \(&z\);)1 420 1 1580 5880 t
(})1330 6000 w
(})1080 6120 w
10 R f
( words, nothing at all will happen if)7 1455(In other)1 316 2 720 6300 t
10 I f
(x)2520 6300 w
10 S f
(\271)2572 6300 w
10 R f
( get the effect implied by the indentation of the origi-)10 2165(0. To)1 240 2 2635 6300 t
(nal example, write:)2 768 1 720 6420 t
8 S1 f
(__________________)720 6980 w
8 R f
(* Thanks to Richard Stevens for pointing this out.)8 1592 1 720 7080 t
cleartomark
showpage
saveobj restore
%%EndPage: 8 10
%%Page: 9 11
/saveobj save def
mark
11 pagesetup
10 R f
(- 9 -)2 166 1 2797 480 t
10 CW f
(if \(x == 0\) {)4 780 1 1080 900 t
(if \(y == 0\))3 660 1 1330 1020 t
(error\(\);)1580 1140 w
(} else {)2 480 1 1080 1260 t
(z = x + y;)4 600 1 1330 1380 t
(f \(&z\);)1 420 1 1330 1500 t
(})1080 1620 w
10 B f
(3. Linkage)1 476 1 720 1920 t
10 R f
( separately and then bound together by a)7 1651(A C program may consist of several parts that are compiled)10 2419 2 970 2076 t
(program usually called a)3 987 1 720 2196 t
10 I f
(linker)1733 2196 w
10 R f
(,)1966 2196 w
10 I f
(linkage editor)1 559 1 2017 2196 t
10 R f
(, or)1 134 1 2576 2196 t
10 I f
(loader)2736 2196 w
10 R f
( compiler normally sees only one file)6 1487( the)1 148(. Because)1 408 3 2997 2196 t
( detect errors whose recognition would require knowledge of several source program)11 3531(at a time, it cannot)4 789 2 720 2316 t
(files at once.)2 507 1 720 2436 t
( C implementations, but not all, have a pro-)8 1754( Some)1 280( type.)1 224(In this section, we look at some errors of that)9 1812 4 970 2592 t
(gram called)1 482 1 720 2712 t
10 I f
(lint)1241 2712 w
10 R f
( is impossible to overemphasize the importance of)7 2097( It)1 124(that catches many of these errors.)5 1405 3 1414 2712 t
(using such a program if it is available.)7 1526 1 720 2832 t
10 B f
( Must Check External Types Yourself)5 1613(3.1. You)1 378 2 720 3072 t
10 R f
( file contains the declaration:)4 1159( One)1 216(Suppose you have a C program divided into two files.)9 2154 3 970 3228 t
10 CW f
(int n;)1 360 1 1080 3408 t
10 R f
(and the other contains the declaration:)5 1522 1 720 3588 t
10 CW f
(long n;)1 420 1 1080 3768 t
10 R f
(This is not a valid C program, because the same external name is declared with two different types in the)19 4320 1 720 3948 t
( because the compiler does not know)6 1505( many implementations will not detect this error,)7 1991( However,)1 446(two files.)1 378 4 720 4068 t
( the job of checking type)5 1015( Thus,)1 280( the other.)2 412(about the contents of either of the two files while it is compiling)12 2613 4 720 4188 t
(consistency can only be done by the linker \(or some utility program like)12 2916 1 720 4308 t
10 I f
(lint)3664 4308 w
10 R f
( has a)2 231(\); if the operating system)4 1011 2 3798 4308 t
(linker that doesn't know about data types, there is little the C compiler can do to force it.)17 3543 1 720 4428 t
( are many possibilities:)3 925( There)1 282(What actually happens when this program is run?)7 1977 3 970 4584 t
( then expect to see a diag-)6 1072( would)1 280( One)1 221( implementation is clever enough to detect the type clash.)9 2342(1. The)1 405 5 720 4740 t
(nostic message explaining that the type of)6 1676 1 970 4860 t
10 CW f
(n)2671 4860 w
10 R f
(was given differently in two different files.)6 1716 1 2756 4860 t
( are using an implementation in which)6 1544(2. You)1 422 2 720 5016 t
10 CW f
(int)2713 5016 w
10 R f
(and)2920 5016 w
10 CW f
(long)3091 5016 w
10 R f
( typically)1 378( is)1 94( This)1 230(are really the same type.)4 980 4 3358 5016 t
( this case, your program will)5 1193( In)1 143( naturally.)1 415(true of machines in which 32-bit arithmetic comes most)8 2319 4 970 5136 t
(probably work as if you had said)6 1309 1 970 5256 t
10 CW f
(long)2304 5256 w
10 R f
(\(or)2569 5256 w
10 CW f
(int)2710 5256 w
10 R f
( would be a good example)5 1051( This)1 228(\) in both declarations.)3 871 3 2890 5256 t
(of a program that works only by coincidence.)7 1813 1 970 5376 t
( two instances of)3 680(3. The)1 405 2 720 5532 t
10 CW f
(n)1832 5532 w
10 R f
(require different amounts of storage, but they happen to share storage in such)12 3121 1 1919 5532 t
( for example, if the)4 778( might happen,)2 607( This)1 233(a way that the values assigned to one are valid for the other.)12 2452 4 970 5652 t
(linker arranged for the)3 897 1 970 5772 t
10 CW f
(int)1893 5772 w
10 R f
( the low-order part of the)5 1010(to share storage with)3 832 2 2099 5772 t
10 CW f
(long)3968 5772 w
10 R f
(. Whether or not this)4 832 1 4208 5772 t
( is an even better example of a program)8 1627( This)1 234( system-dependent.)1 777(happens is obviously machine- and)4 1432 4 970 5892 t
(that works only by coincidence.)4 1273 1 970 6012 t
( two instances of)3 704(4. The)1 405 2 720 6168 t
10 CW f
(n)1864 6168 w
10 R f
( such a way that assigning a value to one has the effect of)13 2435(share storage in)2 646 2 1959 6168 t
( this case, the program will probably fail.)7 1645( In)1 133(apparently assigning a different value to the other.)7 2007 3 970 6288 t
( file of a program will contain)6 1222( One)1 220(Another example of this sort of thing happens surprisingly often.)9 2628 3 970 6444 t
(a declaration like:)2 715 1 720 6564 t
10 CW f
(char filename[] = "/etc/passwd";)3 1920 1 1080 6744 t
10 R f
(and another will contain this declaration:)5 1634 1 720 6924 t
10 CW f
(char *filename;)1 900 1 1080 7104 t
cleartomark
showpage
saveobj restore
%%EndPage: 9 11
%%Page: 10 12
/saveobj save def
mark
12 pagesetup
10 R f
(- 10 -)2 216 1 2772 480 t
(Although arrays and pointers behave very similarly in some contexts,)9 2837 1 970 840 t
10 I f
(they are not the same.)4 907 1 3839 840 t
10 R f
(In the)1 237 1 4803 840 t
(first declaration,)1 658 1 720 960 t
10 CW f
(filename)1407 960 w
10 R f
( generate)1 365( using the name will)4 827( Although)1 432(is the name of an array of characters.)7 1500 4 1916 960 t
( generated as needed and not actually kept around.)8 2046(a pointer to the first element of that array, that pointer is)11 2274 2 720 1080 t
(In the second declaration,)3 1115 1 720 1200 t
10 CW f
(filename)1890 1200 w
10 R f
( points wherever the)3 899( pointer)1 338( That)1 263(is the name of a pointer.)5 1115 4 2425 1200 t
( it a value, it will have a zero \(null\) value by)11 1857( the programmer doesn't give)4 1209( If)1 123(programmer makes it point.)3 1131 4 720 1320 t
(default.)720 1440 w
(The two declarations of)3 945 1 970 1596 t
10 CW f
(filename)1940 1596 w
10 R f
(use storage in different ways; they cannot coexist.)7 1996 1 2445 1596 t
( a tool like)3 428(One way to avoid type clashes of this sort is to use)11 2023 2 970 1752 t
10 I f
(lint)3447 1752 w
10 R f
( order to be able)4 652( In)1 134(if it is available.)3 647 3 3607 1752 t
( to)1 110(to check for type clashes between separately compiled parts of a program, some program must be able)16 4210 2 720 1872 t
( typical compiler does not do this, but)7 1511( The)1 205(see all the parts at once.)5 953 3 720 1992 t
10 I f
(lint)3414 1992 w
10 R f
(does.)3573 1992 w
(Another way to avoid these problems is to put external declarations into)11 2932 1 970 2148 t
10 CW f
(include)3932 2148 w
10 R f
( way,)1 222(files. That)1 436 2 4382 2148 t
(the type of an external object only appears once.*)8 1981 1 720 2268 t
10 B f
( Pitfalls)1 325(4. Semantic)1 519 2 720 2508 t
10 R f
( perfectly spelled and written with impeccable grammar and still be meaningless.)11 3327(A sentence can be)3 743 2 970 2664 t
( we will look at ways of writing programs that look like they mean one thing but actually)17 3714(In this section,)2 606 2 720 2784 t
(mean something quite different.)3 1276 1 720 2904 t
(We will also discuss contexts in which things that look reasonable on the surface actually give unde-)16 4070 1 970 3060 t
( not guaranteed to work on any C implementa-)8 1888( will limit ourselves here to things that are)8 1720( We)1 192(fined results.)1 520 4 720 3180 t
( 7, which)2 385( will leave those that might work on some implementations but not others until section)14 3559(tion. We)1 376 3 720 3300 t
(looks at portability problems.)3 1178 1 720 3420 t
10 B f
( Evaluation Sequence)2 918(4.1. Expression)1 673 2 720 3660 t
10 R f
( Con-)1 260( don't.)1 271( Others)1 326( their operands in a known, specified order.)7 1802(Some C operators always evaluate)4 1411 5 970 3816 t
(sider, for instance, the following expression:)5 1778 1 720 3936 t
10 CW f
(a < b && c < d)6 840 1 1080 4116 t
10 R f
(The language definition states that)4 1376 1 720 4296 t
10 CW f
(a<b)2121 4296 w
10 R f
( If)1 117(will be evaluated first.)3 896 2 2326 4296 t
10 CW f
(a)3365 4296 w
10 R f
(is indeed less than)3 733 1 3451 4296 t
10 CW f
(b)4210 4296 w
10 R f
(,)4270 4296 w
10 CW f
(c<d)4321 4296 w
10 R f
(must then be)2 513 1 4527 4296 t
( the other hand, if)4 711( On)1 173(evaluated to determine the value of the whole expression.)8 2306 3 720 4416 t
10 CW f
(a)3936 4416 w
10 R f
( than or equal to)4 649(is greater)1 369 2 4022 4416 t
10 CW f
(b)720 4536 w
10 R f
(, then)1 222 1 780 4536 t
10 CW f
(c<d)1027 4536 w
10 R f
(is not evaluated at all.)4 874 1 1232 4536 t
(To evaluate)1 491 1 970 4692 t
10 CW f
(a<b)1509 4692 w
10 R f
(, on the other hand, the compiler may evaluate either)9 2311 1 1689 4692 t
10 CW f
(a)4048 4692 w
10 R f
(or)4157 4692 w
10 CW f
(b)4289 4692 w
10 R f
( some)1 260(first. On)1 382 2 4398 4692 t
(machines, it may even evaluate them in parallel.)7 1927 1 720 4812 t
(Only the four C operators)4 1034 1 970 4968 t
10 CW f
(&&)2031 4968 w
10 R f
(,)2151 4968 w
10 CW f
(||)2203 4968 w
10 R f
(,)2323 4968 w
10 CW f
(?:)2376 4968 w
10 R f
(, and)1 197 1 2496 4968 t
10 CW f
(,)2721 4968 w
10 R f
(specify an order of evaluation.)4 1228 1 2809 4968 t
10 CW f
(&&)4090 4968 w
10 R f
(and)4238 4968 w
10 CW f
(||)4410 4968 w
10 R f
(evaluate the)1 482 1 4558 4968 t
( The)1 213( first, and the right operand only if necessary.)8 1877(left operand)1 488 3 720 5088 t
10 CW f
(?:)3331 5088 w
10 R f
(operator takes three operands:)3 1223 1 3484 5088 t
10 CW f
(a?b:c)4740 5088 w
10 R f
(evaluates)720 5208 w
10 CW f
(a)1117 5208 w
10 R f
(first, and then evaluates either)4 1204 1 1203 5208 t
10 CW f
(b)2433 5208 w
10 R f
(or)2519 5208 w
10 CW f
(c)2628 5208 w
10 R f
(, depending on the value of)5 1092 1 2688 5208 t
10 CW f
(a)3807 5208 w
10 R f
(. The)1 207 1 3867 5208 t
10 CW f
(,)4101 5208 w
10 R f
(operator evaluates its)2 852 1 4188 5208 t
(left operand and discards its value, then evaluates its right operand.\262)10 2734 1 720 5328 t
( particular, the assignment opera-)4 1341( In)1 135( undefined order.)2 688(All other C operators evaluate their operands in)7 1906 4 970 5484 t
(tors do not make any guarantees about evaluation order.)8 2237 1 720 5604 t
(For this reason, the following way of copying the first)9 2159 1 970 5760 t
10 CW f
(n)3154 5760 w
10 R f
(elements of array)2 692 1 3239 5760 t
10 CW f
(x)3956 5760 w
10 R f
(to array)1 307 1 4041 5760 t
10 CW f
(y)4373 5760 w
10 R f
(doesn't work:)1 552 1 4458 5760 t
10 CW f
(i = 0;)2 360 1 1080 5940 t
(while \(i < n\))3 780 1 1080 6060 t
(y[i] = x[i++];)2 840 1 1330 6180 t
10 R f
( that the address of)4 758(The trouble is that there is no guarantee)7 1583 2 720 6360 t
10 CW f
(y[i])3087 6360 w
10 R f
(will be evaluated before)3 964 1 3353 6360 t
10 CW f
(i)4343 6360 w
10 R f
(is incremented.)1 611 1 4429 6360 t
8 S1 f
(__________________)720 6460 w
8 R f
( an external object, although there may be many dec-)9 1713(* Some C compilers insist that there must be exactly one definition of)12 2247 2 720 6560 t
( put a declaration in an)5 741( using such a compiler, it may be easiest to)9 1401(larations. When)1 527 3 720 6660 t
8 CW f
(include)3412 6660 w
8 R f
(file and a definition in some)5 909 1 3771 6660 t
( than having it appear more)5 912( means that the type of each external object appears twice, but that is better)14 2482( This)1 189(other place.)1 377 4 720 6760 t
(than two times.)2 489 1 720 6860 t
( example,)1 310( For)1 152(\262 Commas that separate function arguments are not comma operators.)9 2243 3 720 6960 t
8 CW f
(x)3446 6960 w
8 R f
(and)3515 6960 w
8 CW f
(y)3651 6960 w
8 R f
(are fetched in undefined order)4 960 1 3720 6960 t
(in)720 7060 w
8 CW f
(f\(x,y\))802 7060 w
8 R f
(, but not in)3 346 1 1090 7060 t
8 CW f
(g\(\(x,y\)\))1456 7060 w
8 R f
(. In the latter example,)4 714 1 1840 7060 t
8 CW f
(g)2574 7060 w
8 R f
( of that argument is determined by)6 1101( value)1 192( The)1 164(has one argument.)2 581 4 2642 7060 t
(evaluating)720 7160 w
8 CW f
(x)1071 7160 w
8 R f
(, discarding its value, and then evaluating)6 1321 1 1119 7160 t
8 CW f
(y)2460 7160 w
8 R f
(.)2508 7160 w
cleartomark
showpage
saveobj restore
%%EndPage: 10 12
%%Page: 11 13
/saveobj save def
mark
13 pagesetup
10 R f
(- 11 -)2 216 1 2772 480 t
( similar version fails for the same reason:)7 1650( This)1 228(On some implementations, it will; on others, it won't.)8 2148 3 720 840 t
10 CW f
(i = 0;)2 360 1 1080 1020 t
(while \(i < n\))3 780 1 1080 1140 t
(y[i++] = x[i];)2 840 1 1330 1260 t
10 R f
(On the other hand, this one will work fine:)8 1701 1 720 1440 t
10 CW f
(i = 0;)2 360 1 1080 1620 t
(while \(i < n\) {)4 900 1 1080 1740 t
(y[i] = x[i];)2 720 1 1330 1860 t
(i++;)1330 1980 w
(})1080 2100 w
10 R f
(This can, of course, be abbreviated:)5 1421 1 720 2280 t
10 CW f
(for \(i = 0; i < n; i++\))7 1380 1 1080 2460 t
(y[i] = x[i];)2 720 1 1330 2580 t
10 B f
(4.2. The)1 367 1 720 2880 t
10 CW f
(&&)1112 2880 w
10 B f
(,)1232 2880 w
10 CW f
(||)1282 2880 w
10 B f
(, and)1 212 1 1402 2880 t
10 CW f
(!)1639 2880 w
10 B f
(Operators)1724 2880 w
10 R f
(C has two classes of logical operators that are occasionally interchangeable: the bitwise operators)13 3954 1 970 3036 t
10 CW f
(&)4955 3036 w
10 R f
(,)5015 3036 w
10 CW f
(|)720 3156 w
10 R f
(, and)1 196 1 780 3156 t
10 CW f
(\304)1003 3156 w
10 R f
( operators)1 397(, and the logical)3 644 2 1063 3156 t
10 CW f
(&&)2130 3156 w
10 R f
(,)2250 3156 w
10 CW f
(||)2301 3156 w
10 R f
(, and)1 195 1 2421 3156 t
10 CW f
(!)2642 3156 w
10 R f
(. A programmer who substitutes one of these operators for)9 2338 1 2702 3156 t
( may be in for a surprise: the program may appear to work)12 2400(the corresponding operator from the other class)6 1920 2 720 3276 t
(correctly after such an interchange but may actually be working only by coincidence.)12 3400 1 720 3396 t
(The)970 3552 w
10 CW f
(&)1154 3552 w
10 R f
(,)1214 3552 w
10 CW f
(|)1268 3552 w
10 R f
(, and)1 198 1 1328 3552 t
10 CW f
(\304)1555 3552 w
10 R f
( their operands as a sequence of bits and work on each bit separately.)13 2819(operators treat)1 577 2 1644 3552 t
(For example,)1 543 1 720 3672 t
10 CW f
(10&12)1304 3672 w
10 R f
(is 8 \()2 232 1 1645 3672 t
10 CW f
(1000)1877 3672 w
10 R f
(\), because)1 414 1 2117 3672 t
10 CW f
(&)2572 3672 w
10 R f
( of 10 \()3 336(looks at the binary representations)4 1434 2 2673 3672 t
10 CW f
(1010)4443 3672 w
10 R f
(\) and 12)2 357 1 4683 3672 t
(\()720 3792 w
10 CW f
(1100)753 3792 w
10 R f
( position in both)3 684(\) and produces a result that has a bit turned on for each bit that is on in the same)19 3363 2 993 3792 t
(operands. Similarly,)1 834 1 720 3912 t
10 CW f
(10|12)1580 3912 w
10 R f
(is 14 \()2 252 1 1906 3912 t
10 CW f
(1110\))2158 3912 w
10 R f
(and)2483 3912 w
10 CW f
(\30410)2652 3912 w
10 R f
(is \26111 \()2 300 1 2857 3912 t
10 CW f
(11...110101)3157 3912 w
10 R f
(\), at least on a 2's complement)6 1223 1 3817 3912 t
(machine.)720 4032 w
(The)970 4188 w
10 CW f
(&&)1154 4188 w
10 R f
(,)1274 4188 w
10 CW f
(||)1329 4188 w
10 R f
(, and)1 199 1 1449 4188 t
10 CW f
(!)1678 4188 w
10 R f
(operators, on the other hand, treat their arguments as if they are either ``true'' or)14 3272 1 1768 4188 t
( oper-)1 237( These)1 290( 0 represents ``false'' and any other value represents ``true.'')9 2442(``false,'' with the convention that)4 1351 4 720 4308 t
(ators return 1 for ``true'' and 0 for ``false,'' and the)10 2106 1 720 4428 t
10 CW f
(&&)2856 4428 w
10 R f
(and)3006 4428 w
10 CW f
(||)3180 4428 w
10 R f
(operators do not even evaluate their right-)6 1710 1 3330 4428 t
(hand operands if their results can be determined from their left-hand operands.)11 3137 1 720 4548 t
(Thus)970 4704 w
10 CW f
(!10)1206 4704 w
10 R f
(is zero, because 10 is nonzero,)5 1271 1 1422 4704 t
10 CW f
(10&&12)2729 4704 w
10 R f
( and)1 181(is 1, because both 10 and 12 are nonzero,)8 1734 2 3125 4704 t
10 CW f
(10||12)720 4824 w
10 R f
( is not even evaluated in the latter expression, nor)9 2016( 12)1 130( Moreover,)1 473(is also 1, because 10 is nonzero.)6 1311 4 1110 4824 t
(is)720 4944 w
10 CW f
(f\(\))812 4944 w
10 R f
(in)1017 4944 w
10 CW f
(10||f\(\))1120 4944 w
10 R f
(.)1540 4944 w
(Consider the following program fragment to look for a particular element in a table:)13 3353 1 970 5100 t
10 CW f
(i = 0;)2 360 1 1080 5280 t
(while \(i < tabsize && tab[i] != x\))7 2040 1 1080 5400 t
(i++;)1330 5520 w
10 R f
( is that if)3 392(The idea behind this loop)4 1064 2 720 5700 t
10 CW f
(i)2214 5700 w
10 R f
(is equal to)2 437 1 2312 5700 t
10 CW f
(tabsize)2787 5700 w
10 R f
(when the loop terminates, then the element)6 1795 1 3245 5700 t
( Otherwise,)1 485(sought was not found.)3 883 2 720 5820 t
10 CW f
(i)2113 5820 w
10 R f
(contains the element's index.)3 1165 1 2198 5820 t
(Suppose that the)2 661 1 970 5976 t
10 CW f
(&&)1656 5976 w
10 R f
(were inadvertently replaced by)3 1232 1 1801 5976 t
10 CW f
(&)3058 5976 w
10 R f
( probably)1 381( the loop would)3 625( Then)1 255(in this example.)2 636 4 3143 5976 t
(still appear to work, but would do so only because of two lucky breaks.)13 2849 1 720 6096 t
( sort that yield 0 if the condition is false and)10 1770(The first is that both comparisons in this example are of a)11 2300 2 970 6252 t
( long as)2 327( As)1 169(1 if the condition is true.)5 1023 3 720 6372 t
10 CW f
(x)2272 6372 w
10 R f
(and)2365 6372 w
10 CW f
(y)2542 6372 w
10 R f
( or 0,)2 222(are both 1)2 415 2 2635 6372 t
10 CW f
(x&y)3304 6372 w
10 R f
(and)3516 6372 w
10 CW f
(x&&y)3692 6372 w
10 R f
(will always have the same)4 1076 1 3964 6372 t
( replaced by one that uses some non-zero value other)9 2141( if one of the comparisons were to be)8 1496(value. However,)1 683 3 720 6492 t
(than 1 to represent ``true,'' then the loop would stop working.)10 2474 1 720 6612 t
( element off the end of an array is usually harmless,)10 2135(The second lucky break is that looking just one)8 1935 2 970 6768 t
( modified program looks past the end of the)8 1828( The)1 215(provided that the program doesn't change that element.)7 2277 3 720 6888 t
(array because)1 556 1 720 7008 t
10 CW f
(&)1313 7008 w
10 R f
(, unlike)1 312 1 1373 7008 t
10 CW f
(&&)1722 7008 w
10 R f
( in the last iteration of the)6 1105( Thus)1 263( operands.)1 423(, must always evaluate both of its)6 1407 4 1842 7008 t
(loop, the value of)3 702 1 720 7128 t
10 CW f
(tab[i])1448 7128 w
10 R f
( fetched even though)3 834(will be)1 276 2 1834 7128 t
10 CW f
(i)2969 7128 w
10 R f
(is equal to)2 411 1 3054 7128 t
10 CW f
(tabsize)3490 7128 w
10 R f
(. If)1 116 1 3910 7128 t
10 CW f
(tabsize)4051 7128 w
10 R f
(is the number)2 544 1 4496 7128 t
(of elements in)2 566 1 720 7248 t
10 CW f
(tab)1311 7248 w
10 R f
(, this will fetch a non-existent element of)7 1637 1 1491 7248 t
10 CW f
(tab)3153 7248 w
10 R f
(.)3333 7248 w
cleartomark
showpage
saveobj restore
%%EndPage: 11 13
%%Page: 12 14
/saveobj save def
mark
14 pagesetup
10 R f
(- 12 -)2 216 1 2772 480 t
10 B f
( Start from Zero)3 706(4.3. Subscripts)1 651 2 720 840 t
10 R f
(In most languages, an array with)5 1333 1 970 996 t
10 CW f
(n)2334 996 w
10 R f
( has those elements numbered with subscripts)6 1868(elements normally)1 747 2 2425 996 t
(ranging from)1 524 1 720 1116 t
10 CW f
(1)1269 1116 w
10 R f
(to)1354 1116 w
10 CW f
(n)1457 1116 w
10 R f
( so in C.)3 334(inclusive. Not)1 586 2 1542 1116 t
(A C array with)3 608 1 970 1272 t
10 CW f
(n)1607 1272 w
10 R f
(elements does not have an element with a subscript of)9 2191 1 1696 1272 t
10 CW f
(n)3916 1272 w
10 R f
( are num-)2 392(, as the elements)3 672 2 3976 1272 t
(bered from)1 444 1 720 1392 t
10 CW f
(0)1193 1392 w
10 R f
(through)1282 1392 w
10 CW f
(n-1)1622 1392 w
10 R f
( this, programmers coming from other languages must be especially)9 2740(. Because of)2 498 2 1802 1392 t
(careful when using arrays:)3 1055 1 720 1512 t
10 CW f
(int i, a[10];)2 780 1 1080 1692 t
(for \(i=1; i<=10; i++\))3 1260 1 1080 1812 t
(a[i] = 0;)2 540 1 1330 1932 t
10 R f
(This example, intended to set the elements of)7 1851 1 720 2112 t
10 CW f
(a)2602 2112 w
10 R f
( the com-)2 391( Because)1 389( unexpected side effect.)3 962(to zero, had an)3 605 4 2693 2112 t
(parison in the)2 558 1 720 2232 t
10 CW f
(for)1310 2232 w
10 R f
(statement was)1 570 1 1522 2232 t
10 CW f
(i<=10)2124 2232 w
10 R f
(instead of)1 398 1 2456 2232 t
10 CW f
(i<10)2886 2232 w
10 R f
(, the non-existent element number 10 of)6 1637 1 3126 2232 t
10 CW f
(a)4794 2232 w
10 R f
(was)4885 2232 w
(set to zero, thus clobbering the word that followed)8 2027 1 720 2352 t
10 CW f
(a)2775 2352 w
10 R f
( compiler on which this program was)6 1505( The)1 208(in memory.)1 464 3 2863 2352 t
( memory locations, so the word after)6 1473(run allocates memory for users' variables in decreasing)7 2222 2 720 2472 t
10 CW f
(a)4441 2472 w
10 R f
(turned out to)2 513 1 4527 2472 t
(be)720 2592 w
10 CW f
(i)839 2592 w
10 R f
(. Setting)1 334 1 899 2592 t
10 CW f
(i)1258 2592 w
10 R f
(to zero made the loop into an infinite loop.)8 1707 1 1343 2592 t
10 B f
( Doesn't Always Cast Actual Parameters)5 1732(4.4. C)1 272 2 720 2832 t
10 R f
(The following simple program fragment fails for two reasons:)8 2474 1 970 2988 t
10 CW f
(double s;)1 540 1 1080 3168 t
(s = sqrt \(2\);)3 780 1 1080 3288 t
(printf \("%g\\n", s\);)2 1140 1 1080 3408 t
10 R f
(The first reason is that)4 917 1 970 3624 t
10 CW f
(sqrt)1918 3624 w
10 R f
(expects a)1 374 1 2189 3624 t
10 CW f
(double)2594 3624 w
10 R f
( The)1 211(value as its argument and it isn't getting one.)8 1844 2 2985 3624 t
(second is that it returns a)5 996 1 720 3744 t
10 CW f
(double)1741 3744 w
10 R f
( way to correct it is:)5 796( One)1 216(result but isn't declared as such.)5 1281 3 2126 3744 t
10 CW f
(double s, sqrt\(\);)2 1020 1 1080 3924 t
(s = sqrt \(2.0\);)3 900 1 1080 4044 t
(printf \("%g\\n", s\);)2 1140 1 1080 4164 t
10 R f
( that control conversion of function arguments: \(1\) integer values shorter than)11 3147(C has two simple rules)4 923 2 970 4380 t
(an)720 4500 w
10 CW f
(int)845 4500 w
10 R f
(are converted to)2 654 1 1056 4500 t
10 CW f
(int)1741 4500 w
10 R f
(; \(2\) floating-point values shorter than a)6 1628 1 1921 4500 t
10 CW f
(double)3580 4500 w
10 R f
(are converted to)2 654 1 3971 4500 t
10 CW f
(double)4655 4500 w
10 R f
(.)5015 4500 w
(All other values are left unconverted.)5 1495 1 720 4620 t
10 I f
( the programmer's responsibility to ensure that the arguments to)9 2618(It is)1 155 2 2267 4620 t
(a function are of the right type.)6 1247 1 720 4740 t
10 R f
(Therefore, a programmer who uses a function like)7 2077 1 970 4896 t
10 CW f
(sqrt)3083 4896 w
10 R f
(, whose parameter is a)4 933 1 3323 4896 t
10 CW f
(double)4292 4896 w
10 R f
(, must be)2 388 1 4652 4896 t
(careful to pass arguments that are of)6 1464 1 720 5016 t
10 CW f
(float)2212 5016 w
10 R f
(or)2540 5016 w
10 CW f
(double)2651 5016 w
10 R f
( constant)1 360( The)1 207(type only.)1 402 3 3038 5016 t
10 CW f
(2)4034 5016 w
10 R f
(is an)1 188 1 4121 5016 t
10 CW f
(int)4336 5016 w
10 R f
(and is there-)2 497 1 4543 5016 t
(fore of the wrong type.)4 917 1 720 5136 t
( value of a function is used in an expression, that value is automatically cast to an appropri-)17 3684(When the)1 386 2 970 5292 t
( know the actual type returned by the function in order to be able to)14 2765( the compiler must)3 765( However,)1 446(ate type.)1 344 4 720 5412 t
( without further declaration are assumed to return an)8 2135( used)1 212( Functions)1 449(do this.)1 299 4 720 5532 t
10 CW f
(int)3845 5532 w
10 R f
(, so declarations for such)4 1015 1 4025 5532 t
( However,)1 443(functions are unnecessary.)2 1061 2 720 5652 t
10 CW f
(sqrt)2252 5652 w
10 R f
(returns a)1 349 1 2520 5652 t
10 CW f
(double)2897 5652 w
10 R f
( be declared as such before it can)7 1334(, so it must)3 449 2 3257 5652 t
(be used successfully.)2 840 1 720 5772 t
( generally provide a file that can be brought in with an)11 2336(In practice, C implementations)3 1274 2 970 5928 t
10 CW f
(include)4620 5928 w
10 R f
(statement that contains declarations for library functions like)7 2469 1 720 6048 t
10 CW f
(sqrt)3220 6048 w
10 R f
( still neces-)2 465(, but writing declarations is)4 1115 2 3460 6048 t
( who write their own functions \261 in other words, for anyone who writes non-trivial C)15 3450(sary for programmers)2 870 2 720 6168 t
(programs.)720 6288 w
(Here is a more spectacular example:)5 1448 1 970 6444 t
cleartomark
showpage
saveobj restore
%%EndPage: 12 14
%%Page: 13 15
/saveobj save def
mark
15 pagesetup
10 R f
(- 13 -)2 216 1 2772 480 t
10 CW f
(main\(\))1080 900 w
({)1080 1020 w
(int i;)1 360 1 1330 1140 t
(char c;)1 420 1 1330 1260 t
(for \(i=0; i<5; i++\) {)4 1260 1 1330 1380 t
(scanf \("%d", &c\);)2 1020 1 1580 1500 t
(printf \("%d ", i\);)3 1080 1 1580 1620 t
(})1330 1740 w
(printf \("\\n"\);)1 840 1 1330 1860 t
(})1080 1980 w
10 R f
( standard input and writes)4 1060(Ostensibly, this program reads five numbers from its)7 2151 2 970 2196 t
10 CW f
(0 1 2 3 4)4 568 1 4213 2196 t
10 R f
(on its)1 227 1 4813 2196 t
( one compiler, for example, its output is)7 1631( On)1 177( doesn't always do that.)4 966( fact, it)2 292( In)1 139(standard output.)1 650 6 720 2316 t
10 CW f
(0 0 0 0)3 435 1 4605 2316 t
(0 1 2 3 4)4 540 1 720 2436 t
10 R f
(.)1260 2436 w
( key is the declaration of)5 999(Why? The)1 446 2 970 2592 t
10 CW f
(c)2443 2592 w
10 R f
(as a)1 155 1 2531 2592 t
10 CW f
(char)2714 2592 w
10 R f
(rather than as an)3 668 1 2983 2592 t
10 CW f
(int)3680 2592 w
10 R f
(. When you ask)3 633 1 3860 2592 t
10 I f
(scanf)4522 2592 w
10 R f
(to read)1 278 1 4762 2592 t
( a pointer to a character.)5 978( it gets in this case is)6 852( What)1 270(an integer, it expects a pointer to an integer.)8 1784 4 720 2712 t
10 I f
(Scanf)4657 2712 w
10 R f
(has)4907 2712 w
( that it didn't get what it expected: it treats its input as an integer pointer and stores an integer)19 3770(no way to tell)3 550 2 720 2832 t
( an integer takes up more memory than a character, this steps on some of the memory near)17 3609(there. Since)1 496 2 720 2952 t
10 CW f
(c)4850 2952 w
10 R f
(.)4910 2952 w
(Exactly what is near)3 815 1 970 3108 t
10 CW f
(c)1811 3108 w
10 R f
( in this case it turned out to be the low-order part of)12 2082(is the compiler's business;)3 1061 2 1897 3108 t
10 CW f
(i)720 3228 w
10 R f
( read for)2 341(. Therefore, each time a value was)6 1385 2 780 3228 t
10 CW f
(c)2533 3228 w
10 R f
(, it reset)2 323 1 2593 3228 t
10 CW f
(i)2943 3228 w
10 R f
( the program finally reached end of)6 1419( When)1 290(to zero.)1 301 3 3030 3228 t
(file,)720 3348 w
10 I f
(scanf)903 3348 w
10 R f
(stopped trying to put new values into)6 1483 1 1139 3348 t
10 CW f
(c)2647 3348 w
10 R f
(, so)1 139 1 2707 3348 t
10 CW f
(i)2871 3348 w
10 R f
(could be incremented normally to end the loop.)7 1892 1 2956 3348 t
10 B f
( are not Arrays)3 651(4.5. Pointers)1 555 2 720 3588 t
10 R f
( stored as an array of characters, fol-)7 1490(C programs often adopt the convention that a character string is)10 2580 2 970 3744 t
( we have two such strings)5 1069( Suppose)1 397( a null character.)3 688(lowed by)1 378 4 720 3864 t
10 CW f
(s)3285 3864 w
10 R f
(and)3378 3864 w
10 CW f
(t)3555 3864 w
10 R f
(, and we want to concatenate them)6 1425 1 3615 3864 t
(into a single string)3 745 1 720 3984 t
10 CW f
(r)1491 3984 w
10 R f
( we have the usual library functions)6 1437(. To do this,)3 484 2 1551 3984 t
10 CW f
(strcpy)3499 3984 w
10 R f
(and)3886 3984 w
10 CW f
(strcat)4057 3984 w
10 R f
(. The following)2 623 1 4417 3984 t
(obvious method doesn't work:)3 1219 1 720 4104 t
10 CW f
(char *r;)1 480 1 1080 4284 t
(strcpy \(r, s\);)2 840 1 1080 4404 t
(strcat \(r, t\);)2 840 1 1080 4524 t
10 R f
( doesn't work is that)4 832(The reason it)2 527 2 720 4704 t
10 CW f
(r)2108 4704 w
10 R f
( Although)1 432(is not initialized to point anywhere.)5 1436 2 2197 4704 t
10 CW f
(r)4094 4704 w
10 R f
(is potentially capable)2 857 1 4183 4704 t
(of identifying an area of memory,)5 1347 1 720 4824 t
10 I f
(that area doesn't exist until you allocate it.)7 1722 1 2092 4824 t
10 R f
(Let's try again, allocating some memory for)6 1761 1 970 4980 t
10 CW f
(r)2756 4980 w
10 R f
(:)2816 4980 w
10 CW f
(char r[100];)1 720 1 1080 5160 t
(strcpy \(r, s\);)2 840 1 1080 5280 t
(strcat \(r, t\);)2 840 1 1080 5400 t
10 R f
(This now works as long as the strings pointed to by)10 2075 1 720 5580 t
10 CW f
(s)2822 5580 w
10 R f
(and)2909 5580 w
10 CW f
(t)3080 5580 w
10 R f
( C requires us to)4 667( Unfortunately,)1 638( big.)1 181(aren't too)1 387 4 3167 5580 t
( is no way to be certain that)7 1108(state the size of an array as a constant, so there)10 1884 2 720 5700 t
10 CW f
(r)3738 5700 w
10 R f
( However,)1 441(will be big enough.)3 775 2 3824 5700 t
(most C implementations have a library function called)7 2215 1 720 5820 t
10 CW f
(malloc)2966 5820 w
10 R f
(that takes a number and allocates enough)6 1683 1 3357 5820 t
( is also usually a function called)6 1360( There)1 295( characters.)1 466(memory for that many)3 938 4 720 5940 t
10 CW f
(strlen)3817 5940 w
10 R f
(that tells how many)3 825 1 4215 5940 t
( might seem, therefore, that we could write:)7 1744( It)1 111(characters are in a string.)4 999 3 720 6060 t
10 CW f
(char *r, *malloc\(\);)2 1140 1 1080 6240 t
(r = malloc \(strlen\(s\) + strlen\(t\)\);)5 2100 1 1080 6360 t
(strcpy \(r, s\);)2 840 1 1080 6480 t
(strcat \(r, t\);)2 840 1 1080 6600 t
10 R f
( First,)1 259(This example, however, fails for two reasons.)6 1821 2 970 6816 t
10 CW f
(malloc)3076 6816 w
10 R f
(might run out of memory, an event that)7 1578 1 3462 6816 t
(it generally signals by quietly returning a null pointer.)8 2157 1 720 6936 t
( much more important, is that the call to)8 1610(Second, and)1 488 2 970 7092 t
10 CW f
(malloc)3094 7092 w
10 R f
(doesn't allocate quite enough memory.)4 1560 1 3480 7092 t
( The)1 219(Recall the convention that a string is terminated by a null character.)11 2855 2 720 7212 t
10 CW f
(strlen)3833 7212 w
10 R f
(function returns the)2 808 1 4232 7212 t
cleartomark
showpage
saveobj restore
%%EndPage: 13 15
%%Page: 14 16
/saveobj save def
mark
16 pagesetup
10 R f
(- 14 -)2 216 1 2772 480 t
(number of characters in the argument string,)6 1939 1 720 840 t
10 I f
(excluding)2712 840 w
10 R f
( if)1 115( Therefore,)1 496(the null character at the end.)5 1275 3 3154 840 t
10 CW f
(strlen\(s\))720 960 w
10 R f
(is)1286 960 w
10 I f
(n)1379 960 w
10 R f
(,)1429 960 w
10 CW f
(s)1480 960 w
10 R f
(really requires)1 574 1 1566 960 t
10 I f
(n+1)2166 960 w
10 R f
( must therefore allocate one extra char-)6 1561( We)1 188(characters to contain it.)3 931 3 2360 960 t
(acter for)1 334 1 720 1080 t
10 CW f
(r)1079 1080 w
10 R f
(. After doing this and checking that)6 1412 1 1139 1080 t
10 CW f
(malloc)2576 1080 w
10 R f
(worked, we get:)2 640 1 2961 1080 t
10 CW f
(char *r, *malloc\(\);)2 1140 1 1080 1260 t
(r = malloc \(strlen\(s\) + strlen\(t\) + 1\);)7 2340 1 1080 1380 t
(if \(!r\) {)2 540 1 1080 1500 t
(complain\(\);)1330 1620 w
(exit \(1\);)1 540 1 1330 1740 t
(})1080 1860 w
(strcpy \(r, s\);)2 840 1 1080 1980 t
(strcat \(r, t\);)2 840 1 1080 2100 t
10 B f
( Synecdoche)1 525(4.6. Eschew)1 522 2 720 2400 t
10 R f
(A synecdoche \(sin-)2 772 1 970 2556 t
8 R f
(ECK)1742 2556 w
10 R f
( in which,)2 401(-duh-key\) is a literary device, somewhat like a simile or a metaphor,)11 2736 2 1903 2556 t
( used for a less comprehensive)5 1232(according to the Oxford English Dictionary, ``a more comprehensive term is)10 3088 2 720 2676 t
(or vice versa; as whole for part or part for whole, genus for species or species for genus, etc.'')18 3747 1 720 2796 t
( confusing a pointer with the data to which it points.)10 2155(This exactly describes the common C pitfall of)7 1915 2 970 2952 t
( instance:)1 380( For)1 189(This is most common for character strings.)6 1712 3 720 3072 t
10 CW f
(char *p, *q;)2 720 1 1080 3252 t
(p = "xyz";)2 600 1 1080 3372 t
10 R f
( the value of)3 517(It is important to understand that while it is sometimes useful to think of)13 2970 2 720 3552 t
10 CW f
(p)4239 3552 w
10 R f
(as the string)2 497 1 4331 3552 t
10 CW f
(xyz)4860 3552 w
10 R f
( the value of)3 496( Instead,)1 363(after the assignment, this is not really true.)7 1701 3 720 3672 t
10 CW f
(p)3305 3672 w
10 R f
(is a)1 136 1 3390 3672 t
10 I f
(pointer)3551 3672 w
10 R f
(to the 0th element of an array)6 1175 1 3865 3672 t
(of four characters, whose values are)5 1433 1 720 3792 t
10 CW f
('x')2178 3792 w
10 R f
(,)2358 3792 w
10 CW f
('y')2408 3792 w
10 R f
(,)2588 3792 w
10 CW f
('z')2638 3792 w
10 R f
(, and)1 194 1 2818 3792 t
10 CW f
('\\0')3037 3792 w
10 R f
(. Thus, if we now execute)5 1028 1 3277 3792 t
10 CW f
(q = p;)2 360 1 1080 3972 t
(p)720 4152 w
10 R f
(and)818 4152 w
10 CW f
(q)1000 4152 w
10 R f
( characters in that memory did not get)7 1615( The)1 219(are now two pointers to the same part of memory.)9 2108 3 1098 4152 t
( situation now looks like this:)5 1182( The)1 205(copied by the assignment.)3 1038 3 720 4272 t
cleartomark
saveobj restore
%%BeginGlobal
%
% Version 3.3.1 drawing procedures for dpost. Automatically pulled in when
% needed.
%

/inpath false def
/savematrix matrix def

/Dl {
	inpath
		{pop pop neg lineto}
		{newpath neg moveto neg lineto stroke}
	ifelse
} bind def

/De {
	/y1 exch 2 div def
	/x1 exch 2 div def
	/savematrix savematrix currentmatrix def
	neg exch x1 add exch translate
	x1 y1 scale
	0 0 1 0 360
	inpath
		{1 0 moveto arc savematrix setmatrix}
		{newpath arc savematrix setmatrix stroke}
	ifelse
} bind def

/Da {
	/dy2 exch def
	/dx2 exch def
	/dy1 exch def
	/dx1 exch def
	dy1 add neg exch dx1 add exch
	dx1 dx1 mul dy1 dy1 mul add sqrt
	dy1 dx1 neg atan
	dy2 neg dx2 atan
	inpath
		{arc}
		{newpath arc stroke}
	ifelse
} bind def

/DA {
	/dy2 exch def
	/dx2 exch def
	/dy1 exch def
	/dx1 exch def
	dy1 add neg exch dx1 add exch
	dx1 dx1 mul dy1 dy1 mul add sqrt
	dy1 dx1 neg atan
	dy2 neg dx2 atan
	inpath
		{arcn}
		{newpath arcn stroke}
	ifelse
} bind def

/Ds {
	/y2 exch def
	/x2 exch def
	/y1 exch def
	/x1 exch def
	/y0 exch def
	/x0 exch def
	x0 5 x1 mul add 6 div
	y0 5 y1 mul add -6 div
	x2 5 x1 mul add 6 div
	y2 5 y1 mul add -6 div
	x1 x2 add 2 div
	y1 y2 add -2 div
	inpath
		{curveto}
		{newpath x0 x1 add 2 div y0 y1 add -2 div moveto curveto stroke}
	ifelse
} bind def
%%EndGlobal
/saveobj save def
mark
10 R f
3398 4664 3398 4664 3398 4894 Ds
3398 4664 3398 4894 2189 5009 Ds
3398 4894 2189 5009 2189 5124 Ds
2189 5009 2189 5124 2189 5124 Ds
2361 4664 2361 4664 2368 4894 Ds
2361 4664 2368 4894 2188 5009 Ds
2368 4894 2188 5009 2188 5124 Ds
2188 5009 2188 5124 2188 5124 Ds
2188 5125 2170 5053 Dl
2188 5125 2206 5053 Dl
2131 5125 2131 5240 Dl
2822 5125 2131 5125 Dl
2822 5240 2822 5125 Dl
2131 5240 2822 5240 Dl
2303 5067 1 1 De
2303 5240 2303 5125 Dl
2476 5240 2476 5125 Dl
2649 5240 2649 5125 Dl
( \\0)1 190( z)1 170(x y)1 223 3 2163 5202 t
(q p)1 -987 1 3373 4511 t
3167 4549 3167 4779 Dl
3627 4549 3167 4549 Dl
3628 4779 3628 4549 Dl
3168 4779 3628 4779 Dl
2131 4549 2131 4779 Dl
2591 4549 2131 4549 Dl
2591 4779 2591 4549 Dl
2131 4779 2591 4779 Dl
(The thing to remember is that)5 1185 1 970 5532 t
10 I f
(copying a pointer does not copy the thing it points to.)10 2136 1 2180 5532 t
10 R f
(Thus, if after this we were to execute)7 1479 1 970 5688 t
10 CW f
(q[1] = 'Y';)2 660 1 1080 5868 t
(q)720 6048 w
10 R f
(would point to memory containing the string)6 1801 1 807 6048 t
10 CW f
(xYz)2635 6048 w
10 R f
(. So would)2 437 1 2815 6048 t
10 CW f
(p)3280 6048 w
10 R f
(, because)1 368 1 3340 6048 t
10 CW f
(p)3736 6048 w
10 R f
(and)3824 6048 w
10 CW f
(q)3996 6048 w
10 R f
(point to the same mem-)4 956 1 4084 6048 t
(ory.)720 6168 w
10 B f
( Null Pointer is Not the Null String)7 1481(4.7. The)1 367 2 720 6408 t
10 R f
(The result of converting an integer to a pointer is implementation-dependent, with one important)13 4070 1 970 6564 t
( the constant)2 505( exception is)2 507(exception. That)1 647 3 720 6684 t
10 CW f
(0)2404 6684 w
10 R f
(, which is guaranteed to be converted to a pointer that is unequal)12 2576 1 2464 6684 t
( documentation, this value is often given symbolically:)7 2194( For)1 189(to any valid pointer.)3 805 3 720 6804 t
10 CW f
(#define NULL 0)2 840 1 1080 6984 t
10 R f
( important thing to remember about)5 1444( The)1 209(but the effect is the same.)5 1040 3 720 7164 t
10 CW f
(0)3442 7164 w
10 R f
( a pointer is that)4 664(when used as)2 540 2 3531 7164 t
10 I f
(it must)1 275 1 4765 7164 t
(never be dereferenced.)2 924 1 720 7284 t
10 R f
( when you have assigned)4 1022(In other words,)2 621 2 1701 7284 t
10 CW f
(0)3375 7284 w
10 R f
(to a pointer variable, you must not ask)7 1574 1 3466 7284 t
cleartomark
showpage
saveobj restore
%%EndPage: 14 16
%%Page: 15 17
/saveobj save def
mark
17 pagesetup
10 R f
(- 15 -)2 216 1 2772 480 t
( is valid to write:)4 678( It)1 111(what is in the memory it points to.)7 1373 3 720 840 t
10 CW f
(if \(p == \(char *\) 0\) ...)6 1440 1 1080 1020 t
10 R f
(but it is not valid to write:)6 1040 1 720 1200 t
10 CW f
(if \(strcmp \(p, \(char *\) 0\) == 0\) ...)8 2160 1 1080 1380 t
10 R f
(because)720 1560 w
10 CW f
(strcmp)1060 1560 w
10 R f
(always looks at the memory addressed by its arguments.)8 2250 1 1445 1560 t
(If)970 1716 w
10 CW f
(p)1061 1716 w
10 R f
(is a null pointer, it is not even valid to say:)10 1703 1 1146 1716 t
10 CW f
(printf \(p\);)1 660 1 1080 1896 t
10 R f
(or)720 2076 w
10 CW f
(printf \("%s", p\);)2 1020 1 1080 2256 t
10 B f
( Overflow)1 424(4.8. Integer)1 510 2 720 2556 t
10 R f
( when an integer operation overflows)5 1523(The C language definition is very specific about what happens)9 2547 2 970 2712 t
(or underflows.)1 582 1 720 2832 t
( 2)1 83(If either operand is unsigned, the result is unsigned, and is defined to be modulo)14 3300 2 970 2988 t
7 I f
(n)4358 2948 w
10 R f
(, where)1 301 1 4401 2988 t
10 I f
(n)4735 2988 w
10 R f
(is the)1 222 1 4818 2988 t
( both operands are signed, the result is)7 1531( If)1 116(word size.)1 410 3 720 3108 t
10 I f
(undefined)2802 3108 w
10 R f
(.)3196 3108 w
(Suppose, for example, that)3 1104 1 970 3264 t
10 CW f
(a)2111 3264 w
10 R f
(and)2208 3264 w
10 CW f
(b)2389 3264 w
10 R f
( be non-negative, and you)4 1086(are two integer variables, known to)5 1468 2 2486 3264 t
(want to test whether)3 807 1 720 3384 t
10 CW f
(a+b)1552 3384 w
10 R f
( obvious way to do it looks something like this:)9 1899( One)1 216(might overflow.)1 644 3 1757 3384 t
10 CW f
(if \(a + b < 0\))5 840 1 1080 3564 t
(complain\(\);)1330 3684 w
10 R f
(In general, this does not work.)5 1212 1 720 3864 t
(The point is that once)4 866 1 970 4020 t
10 CW f
(a+b)1861 4020 w
10 R f
( example,)1 389( For)1 190( all bets are off as to what the result will be.)11 1758(has overflowed,)1 637 4 2066 4020 t
( an internal register to one of four states: positive, negative,)10 2454(on some machines, an addition operation sets)6 1866 2 720 4140 t
( to implement the example)4 1126( such a machine, the compiler would have every right)9 2266( On)1 186(zero, or overflow.)2 742 4 720 4260 t
(given above by adding)3 913 1 720 4380 t
10 CW f
(a)1660 4380 w
10 R f
(and)1747 4380 w
10 CW f
(b)1918 4380 w
10 R f
( this internal register was in negative state afterwards.)8 2156(and checking whether)2 879 2 2005 4380 t
(If the operation overflowed, the register would be in overflow state, and the test would fail.)15 3643 1 720 4500 t
( particular test relies on the fact that unsigned arithmetic is well-defined)11 2881(One correct way of doing this)5 1189 2 970 4656 t
(for all values, as are the conversions between signed and unsigned values:)11 2955 1 720 4776 t
10 CW f
(if \(\(int\) \(\(unsigned\) a + \(unsigned\) b\) < 0\))8 2640 1 1080 4956 t
(complain\(\);)1330 5076 w
10 B f
( Operators)1 463(4.9. Shift)1 406 2 720 5376 t
10 R f
(Two questions seem to cause trouble for people who use shift operators:)11 2887 1 970 5532 t
( a right shift, are vacated bits filled with zeroes or copies of the sign bit?)15 2884(1. In)1 333 2 720 5688 t
( values are permitted for the shift count?)7 1616(2. What)1 466 2 720 5844 t
( the item)2 386( If)1 134( first question is simple but sometimes implementation-dependent.)7 2787(The answer to the)3 763 4 970 6000 t
( to fill)2 247( the item is signed, the implementation is permitted)8 2069( If)1 118(being shifted is unsigned, zeroes are shifted in.)7 1886 4 720 6120 t
( you care about vacated bits in a)7 1346( If)1 125( the sign bit.)3 522(vacated bit positions either with zeroes or with copies of)9 2327 4 720 6240 t
(right shift, declare the variable in question as)7 1811 1 720 6360 t
10 CW f
(unsigned)2558 6360 w
10 R f
( vacated bits)2 501(. You are then entitled to assume that)7 1501 2 3038 6360 t
(will be set to zero.)4 735 1 720 6480 t
(The answer to the second question is also simple: if the item being shifted is)14 3048 1 970 6636 t
10 I f
(n)4043 6636 w
10 R f
( the shift)2 352(bits long, then)2 570 2 4118 6636 t
( equal to zero and)4 733(count must be greater than or)5 1202 2 720 6756 t
10 I f
(strictly)2686 6756 w
10 R f
(less than)1 353 1 2995 6756 t
10 I f
(n)3379 6756 w
10 R f
( it is not possible to shift all the)8 1305(. Thus,)1 306 2 3429 6756 t
(bits out of a value in a single operation.)8 1579 1 720 6876 t
(For example, if an)3 762 1 970 7032 t
10 CW f
(int)1767 7032 w
10 R f
( bits, and)2 386(is 32)1 202 2 1982 7032 t
10 CW f
(n)2606 7032 w
10 R f
(is an)1 197 1 2702 7032 t
10 CW f
(int)2935 7032 w
10 R f
(, it is legal to write)5 805 1 3115 7032 t
10 CW f
(n<<31)3956 7032 w
10 R f
(and)4292 7032 w
10 CW f
(n<<0)4472 7032 w
10 R f
(but not)1 292 1 4748 7032 t
10 CW f
(n<<32)720 7152 w
10 R f
(or)1045 7152 w
10 CW f
(n<<-1)1153 7152 w
10 R f
(.)1453 7152 w
( equivalent to division by a power of two,)8 1732(Note that a right shift of a signed integer is generally not)11 2338 2 970 7308 t
cleartomark
showpage
saveobj restore
%%EndPage: 15 17
%%Page: 16 18
/saveobj save def
mark
18 pagesetup
10 R f
(- 16 -)2 216 1 2772 480 t
( prove this, consider that the value of)7 1530( To)1 168(even if the implementation copies the sign into vacated bits.)9 2470 3 720 840 t
10 CW f
(\(-)4920 840 w
(1\)>>1)720 960 w
10 R f
(cannot possibly be zero.)3 965 1 1045 960 t
10 B f
( Functions)1 448(5. Library)1 464 2 720 1200 t
10 R f
( use library functions, because there is no way of doing input or output)13 2876(Every useful C program must)4 1194 2 970 1356 t
( this section, we look at some cases where some widely-available library func-)12 3231( In)1 141( language.)1 418(built into the)2 530 4 720 1476 t
(tions behave in ways that the programmer might not expect.)9 2397 1 720 1596 t
10 B f
( Returns an Integer)3 835(5.1. Getc)1 399 2 720 1836 t
10 R f
(Consider the following program:)3 1313 1 970 1992 t
10 CW f
(#include <stdio.h>)1 1080 1 1080 2172 t
(main\(\))1080 2412 w
({)1080 2532 w
(char c;)1 420 1 1330 2652 t
(while \(\(c = getchar\(\)\) != EOF\))5 1800 1 1330 2892 t
(putchar \(c\);)1 720 1 1580 3012 t
(})1080 3132 w
10 R f
( fact, it doesn't)3 632( In)1 144( copy its standard input to its standard output.)8 1913(This program looks like it should)5 1381 4 970 3348 t
(quite do this.)2 520 1 720 3468 t
(The reason is that)3 710 1 970 3624 t
10 CW f
(c)1706 3624 w
10 R f
( that it is impossi-)4 726( means)1 281( This)1 229(is declared as a character rather than as an integer.)9 2012 4 1792 3624 t
(ble for)1 263 1 720 3744 t
10 CW f
(c)1008 3744 w
10 R f
(to hold every possible character as well as)7 1682 1 1093 3744 t
10 CW f
(EOF)2800 3744 w
10 R f
(.)2980 3744 w
( some legitimate input character will cause)6 1762( Either)1 303( are two possibilities.)3 877(Thus there)1 432 4 970 3900 t
10 CW f
(c)4378 3900 w
10 R f
(to take on the)3 568 1 4472 3900 t
(same value as)2 562 1 720 4020 t
10 CW f
(EOF)1310 4020 w
10 R f
(, or it will be impossible for)6 1132 1 1490 4020 t
10 CW f
(c)2650 4020 w
10 R f
(to have the value)3 688 1 2738 4020 t
10 CW f
(EOF)3454 4020 w
10 R f
( the former case, the pro-)5 1017( In)1 136(at all.)1 225 3 3662 4020 t
( the latter case, the program will go into an infinite)10 2058( In)1 136( middle of certain files.)4 941(gram will stop copying in the)5 1185 4 720 4140 t
(loop.)720 4260 w
( C Reference Manual)3 898( The)1 221( the program may work by coincidence.)6 1684(Actually, there is a third case:)5 1267 4 970 4416 t
(defines the result of the expression)5 1389 1 720 4536 t
10 CW f
(\(\(c = getchar\(\)\) != EOF\))4 1440 1 1080 4716 t
10 R f
( 6.1 states:)2 425( Section)1 350(quite rigorously.)1 661 3 720 4896 t
(When a longer integer is converted to a shorter or to a)11 2230 1 970 5052 t
10 CW f
(char)3232 5052 w
10 R f
( on the left; excess bits)5 953(, it is truncated)3 615 2 3472 5052 t
(are simply discarded.)2 851 1 970 5172 t
(Section 7.14 states:)2 775 1 720 5292 t
( require an lvalue as)4 811( All)1 180(There are a number of assignment operators, all of which group right-to-left.)11 3079 3 970 5448 t
( value is)2 347( The)1 212( the type of an assignment expression is that of its left operand.)12 2606(their left operand, and)3 905 4 970 5568 t
(the value stored in the left operand after the assignment has taken place.)12 2874 1 970 5688 t
( to require that the result of)6 1105(The combined effect of these two sections is)7 1791 2 720 5808 t
10 I f
(getchar)3644 5808 w
10 R f
(be truncated to a character)4 1063 1 3977 5808 t
( truncated value then be compared with)6 1574(value by discarding the high-order bits, and that this)8 2092 2 720 5928 t
10 CW f
(EOF)4411 5928 w
10 R f
(. As part of)3 449 1 4591 5928 t
(this comparison, the value of)4 1173 1 720 6048 t
10 CW f
(c)1922 6048 w
10 R f
( zero bits)2 376(must be extended to an integer, either by padding on the left with)12 2653 2 2011 6048 t
(or by sign extension, as appropriate.)5 1445 1 720 6168 t
( properly assign the)3 842( They)1 274(However, some compilers do not implement this expression correctly.)8 2954 3 970 6324 t
(low-order bits of the value of)5 1167 1 720 6444 t
10 I f
(getchar)1912 6444 w
10 R f
(to)2242 6444 w
10 CW f
(c)2345 6444 w
10 R f
(. However, instead of then comparing)5 1505 1 2405 6444 t
10 CW f
(c)3935 6444 w
10 R f
(to)4020 6444 w
10 CW f
(EOF)4123 6444 w
10 R f
(, they compare the)3 737 1 4303 6444 t
(entire value of)2 592 1 720 6564 t
10 I f
(getchar)1345 6564 w
10 R f
( make the sample program shown above appear to)8 2073( compiler that does this will)5 1154(! A)1 163 3 1650 6564 t
(work ``correctly.'')1 741 1 720 6684 t
cleartomark
showpage
saveobj restore
%%EndPage: 16 18
%%Page: 17 19
/saveobj save def
mark
19 pagesetup
10 R f
(- 17 -)2 216 1 2772 480 t
10 B f
( Output and Memory Allocation)4 1378(5.2. Buffered)1 577 2 720 840 t
10 R f
( to see that output imme-)5 1022(When a program produces output, how important is it that a human be able)13 3048 2 970 996 t
( depends on the program.)4 1012(diately? It)1 427 2 720 1116 t
( asking the person sitting at that terminal to)8 1805(For example, if the output is going to a terminal and is)11 2265 2 970 1272 t
( On)1 176( to be able to know what to type.)8 1339(answer a question, it is crucial that the person see the output in order)13 2805 3 720 1392 t
( is going to a file, from where it will eventually be sent to a line printer, it is)18 3175(the other hand, if the output)5 1145 2 720 1512 t
(only important that all the output get there eventually.)8 2157 1 720 1632 t
( to arrange for output to appear immediately than it is to save it up for a)16 3001(It is often more expensive)4 1069 2 970 1788 t
( this reason, C implementations typically afford the pro-)8 2317( For)1 198( in a large chunk.)4 726(while and write it later on)5 1079 4 720 1908 t
(grammer some control over how much output is to be produced before it is actually written.)15 3667 1 720 2028 t
(That control is often vested in a library function called)9 2177 1 970 2184 t
10 I f
(setbuf)3172 2184 w
10 R f
(. If)1 141 1 3411 2184 t
10 CW f
(buf)3577 2184 w
10 R f
(is a character array of appropri-)5 1258 1 3782 2184 t
(ate size, then)2 518 1 720 2304 t
10 CW f
(setbuf \(stdout, buf\);)2 1260 1 1080 2484 t
10 R f
(tells the I/O library that all output written to)8 1795 1 720 2664 t
10 I f
(stdout)2545 2664 w
10 R f
(should henceforth use)2 886 1 2820 2664 t
10 I f
(buf)3736 2664 w
10 R f
( that)1 181(as an output buffer, and)4 965 2 3894 2664 t
(output directed to)2 737 1 720 2784 t
10 I f
(stdout)1498 2784 w
10 R f
(should not actually be written until)5 1477 1 1784 2784 t
10 I f
(buf)3302 2784 w
10 R f
(becomes full or until the programmer)5 1570 1 3470 2784 t
( by calling)2 442(directs it to be written)4 913 2 720 2904 t
10 I f
(fflush)2110 2904 w
10 R f
( appropriate size for such a buffer is defined as)9 1959(. The)1 240 2 2333 2904 t
10 CW f
(BUFSIZ)4567 2904 w
10 R f
(in)4962 2904 w
10 S f
(<)720 3024 w
10 I f
(stdio.h)775 3024 w
10 S f
(>)1045 3024 w
10 R f
(.)1100 3024 w
(Thus, the following program illustrates the obvious way to use)9 2612 1 970 3180 t
10 I f
(setbuf)3619 3180 w
10 R f
(in a program that copies its)5 1145 1 3895 3180 t
(standard input to its standard output:)5 1464 1 720 3300 t
10 CW f
(#include <stdio.h>)1 1080 1 1080 3480 t
(main\(\))1080 3720 w
({)1080 3840 w
(int c;)1 360 1 1330 3960 t
(char buf[BUFSIZ];)1 1020 1 1330 4200 t
(setbuf \(stdout, buf\);)2 1260 1 1330 4320 t
(while \(\(c = getchar\(\)\) != EOF\))5 1800 1 1330 4560 t
(putchar \(c\);)1 720 1 1580 4680 t
(})1080 4800 w
10 R f
(Unfortunately, this program is wrong, for a subtle reason.)8 2299 1 970 5016 t
( after the main)3 585( Answer:)1 390(To see where the trouble lies, ask when the buffer is flushed for the last time.)15 3095 3 970 5172 t
( the cleaning up that the library does before handing control back to the)13 2996(program has finished, as part of)5 1324 2 720 5292 t
( by that time, the buffer has already been freed!)9 1894( But)1 195(operating system.)1 705 3 720 5412 t
(There are two ways to prevent this sort of trouble.)9 1996 1 970 5568 t
(First, make the buffer static, either by declaring it explicitly as static:)11 2755 1 970 5724 t
10 CW f
(static char buf[BUFSIZ];)2 1440 1 1330 5904 t
10 R f
(or by moving the declaration outside the main program entirely.)9 2558 1 720 6084 t
(Another possibility is to allocate the buffer dynamically and never free it:)11 2937 1 970 6240 t
10 CW f
(char *malloc\(\);)1 900 1 1330 6420 t
(setbuf \(stdout, malloc \(BUFSIZ\)\);)3 1980 1 1330 6540 t
10 R f
( unnecessary to check if)4 998(Note that in this latter case, it is)7 1329 2 720 6720 t
10 I f
(malloc)3082 6720 w
10 R f
(was successful, because if it fails it will)7 1651 1 3389 6720 t
( argument to)2 519( null pointer is an acceptable second)6 1495( A)1 130(return a null pointer.)3 845 4 720 6840 t
10 I f
(setbuf)3741 6840 w
10 R f
(; it requests that)3 657 1 3980 6840 t
10 I f
(stdout)4669 6840 w
10 R f
(be)4946 6840 w
( will work slowly, but it will work.)7 1398(unbuffered. This)1 690 2 720 6960 t
cleartomark
showpage
saveobj restore
%%EndPage: 17 19
%%Page: 18 20
/saveobj save def
mark
20 pagesetup
10 R f
(- 18 -)2 216 1 2772 480 t
10 B f
( Preprocessor)1 584(6. The)1 292 2 720 840 t
10 R f
( they are first transformed by the C preprocessor.)8 1967(The programs we run are not the programs we write:)9 2103 2 970 996 t
( reasons \(and several)3 836(The preprocessor gives us a way of abbreviating things that is important for two major)14 3484 2 720 1116 t
(minor ones\).)1 505 1 720 1236 t
( size of a)3 387(First, we may want to be able to change all instances of a particular quantity, such as the)17 3683 2 970 1392 t
(table, by changing one number and recompiling the program.*)8 2496 1 720 1512 t
( do not have the execution over-)6 1298(Second, we may want to define things that appear to be functions but)12 2772 2 970 1668 t
( example,)1 393( For)1 194(head normally associated with a function call.)6 1863 3 720 1788 t
10 I f
(getchar)3200 1788 w
10 R f
(and)3535 1788 w
10 I f
(putchar)3709 1788 w
10 R f
(are usually implemented)2 990 1 4050 1788 t
(as macros to avoid having to call a function for each character of input or output.)15 3232 1 720 1908 t
10 B f
( are not Functions)3 775(6.1. Macros)1 521 2 720 2148 t
10 R f
(Because macros can be made to appear almost as if they were functions, programmers are sometimes)15 4070 1 970 2304 t
( one sees things like this:)5 1003( Thus,)1 275(tempted to regard them as truly equivalent.)6 1717 3 720 2424 t
10 CW f
(#define max\(a,b\) \(\(a\)>\(b\)?\(a\):\(b\)\))2 2040 1 1080 2604 t
10 R f
( defend against the possibility that)5 1434( They)1 268(Notice all the parentheses in the macro body.)7 1893 3 720 2784 t
10 CW f
(a)4353 2784 w
10 R f
(or)4452 2784 w
10 CW f
(b)4574 2784 w
10 R f
(might be)1 367 1 4673 2784 t
(expressions that contain operators of lower precedence than)7 2385 1 720 2904 t
10 CW f
(>)3130 2904 w
10 R f
(.)3190 2904 w
(The main problem, though, with defining things like)7 2153 1 970 3060 t
10 CW f
(max)3156 3060 w
10 R f
( operand that is used)4 857(as macros is that an)4 814 2 3369 3060 t
( in this example, if)4 755( Thus,)1 277(twice may be evaluated twice.)4 1213 3 720 3180 t
10 CW f
(a)2992 3180 w
10 R f
(is greater than)2 569 1 3079 3180 t
10 CW f
(b)3675 3180 w
10 R f
(,)3735 3180 w
10 CW f
(a)3786 3180 w
10 R f
(will be evaluated twice: once)4 1168 1 3872 3180 t
(during the comparison, and again to calculate the value yielded by)10 2648 1 720 3300 t
10 CW f
(max)3393 3300 w
10 R f
(.)3573 3300 w
(Not only can this be inefficient, it can also be wrong:)10 2122 1 970 3456 t
10 CW f
(biggest = x[0];)2 900 1 1080 3636 t
(i = 1;)2 360 1 1080 3756 t
(while \(i < n\))3 780 1 1080 3876 t
(biggest = max \(biggest, x[i++]\);)4 1920 1 1330 3996 t
10 R f
(This would work fine if)4 961 1 720 4176 t
10 CW f
(max)1710 4176 w
10 R f
(were a true function, but fails with)6 1402 1 1919 4176 t
10 CW f
(max)3350 4176 w
10 R f
( for example, that)3 716( Suppose,)1 418(a macro.)1 347 3 3559 4176 t
10 CW f
(x[0])720 4296 w
10 R f
(is 2,)1 177 1 996 4296 t
10 CW f
(x[1])1208 4296 w
10 R f
(is 3, and)2 356 1 1483 4296 t
10 CW f
(x[2])1874 4296 w
10 R f
( The)1 215( at what happens during the first iteration of the loop.)10 2228( Look)1 271(is 1.)1 177 4 2149 4296 t
(assignment statement expands into:)3 1419 1 720 4416 t
10 CW f
(biggest = \(\(biggest\)>\(x[i++]\)?\(biggest\):\(x[i++]\)\);)2 3000 1 1080 4596 t
10 R f
(First,)720 4776 w
10 CW f
(biggest)954 4776 w
10 R f
(is compared to)2 590 1 1399 4776 t
10 CW f
(x[i++])2015 4776 w
10 R f
(. Since)1 273 1 2375 4776 t
10 CW f
(i)2674 4776 w
10 R f
(is 1 and)2 313 1 2760 4776 t
10 CW f
(x[1])3099 4776 w
10 R f
( a side effect,)3 534( As)1 162(is 3, the relation is false.)5 979 3 3365 4776 t
10 CW f
(i)720 4896 w
10 R f
(is incremented to 2.)3 788 1 805 4896 t
( the value of)3 502(Because the relation is false,)4 1143 2 970 5052 t
10 CW f
(x[i++])2642 5052 w
10 R f
(is now assigned to)3 742 1 3029 5052 t
10 CW f
(biggest)3798 5052 w
10 R f
(. However,)1 442 1 4218 5052 t
10 CW f
(i)4687 5052 w
10 R f
(is now)1 266 1 4774 5052 t
(2, so the value assigned to)5 1049 1 720 5172 t
10 CW f
(biggest)1794 5172 w
10 R f
(is the value of)3 563 1 2239 5172 t
10 CW f
(x[2])2827 5172 w
10 R f
(, which is 1.)3 486 1 3067 5172 t
(One way around these worries is to ensure that the arguments to the)12 2730 1 970 5328 t
10 CW f
(max)3727 5328 w
10 R f
( side)1 189(macro do not have any)4 917 2 3934 5328 t
(effects:)720 5448 w
10 CW f
(biggest = x[0];)2 900 1 1080 5628 t
(for \(i = 1; i < n; i++\))7 1380 1 1080 5748 t
(biggest = max \(biggest, x[i]\);)4 1800 1 1330 5868 t
10 R f
( is the definition of)4 793( This)1 236( macros.)1 346(Here is another example of the hazards of mixing side effects and)11 2695 4 970 6084 t
(the)720 6204 w
10 CW f
(putc)867 6204 w
10 R f
(macro from)1 468 1 1132 6204 t
10 CW f
(<stdio.h>)1625 6204 w
10 R f
(in the Eighth Edition of the Unix system:)7 1648 1 2190 6204 t
10 CW f
(#define putc\(x,p\) \(--\(p\)->_cnt>=0?\(*\(p\)->_ptr++=\(x\)\):_flsbuf\(x,p\)\))2 3960 1 1080 6384 t
10 R f
( to)1 114(The first argument)2 763 2 720 6564 t
10 CW f
(putc)1633 6564 w
10 R f
(is a character to be written to a file; the second argument is a pointer to an)16 3131 1 1909 6564 t
( the first argument, which could easily be some-)8 1986( that)1 184( Notice)1 325(internal data structure that describes the file.)6 1825 4 720 6684 t
(thing like)1 395 1 720 6804 t
10 CW f
(*z++)1154 6804 w
10 R f
( evaluated only once, even though it appears in two separate places in the)13 3122(, is carefully)2 524 2 1394 6804 t
( body,)1 256(macro body, while the second argument is evaluated twice \(in the macro)11 2971 2 720 6924 t
10 CW f
(x)3978 6924 w
10 R f
(appears twice, but since)3 971 1 4069 6924 t
8 S1 f
(__________________)720 7024 w
8 R f
(* The preprocessor also makes it easy to group such)9 1659 1 720 7124 t
8 I f
(manifest constants)1 591 1 2399 7124 t
8 R f
(together to make them easier to find.)6 1167 1 3010 7124 t
cleartomark
showpage
saveobj restore
%%EndPage: 18 20
%%Page: 19 21
/saveobj save def
mark
21 pagesetup
10 R f
(- 19 -)2 216 1 2772 480 t
(the two occurrences are on opposite sides of a)8 1850 1 720 840 t
10 CW f
(:)2597 840 w
10 R f
(operator, exactly one of them will be evaluated in any sin-)10 2356 1 2684 840 t
( of)1 122(gle instance)1 489 2 720 960 t
10 CW f
(putc)1370 960 w
10 R f
(\). Since it is unusual for the file argument to)9 1891 1 1610 960 t
10 I f
(putc)3540 960 w
10 R f
(to have side effects, this rarely)5 1289 1 3751 960 t
( a)1 88( it is documented in the user's manual: ``Because it is implemented as)12 3013( Nevertheless,)1 608(causes trouble.)1 611 4 720 1080 t
(macro,)720 1200 w
10 I f
(putc)1022 1200 w
10 R f
(treats a)1 288 1 1222 1200 t
10 I f
(stream)1538 1200 w
10 R f
( particular,)1 435( In)1 136(argument with side effects improperly.)4 1562 3 1838 1200 t
10 CW f
(putc\(c,*f++\))3999 1200 w
10 R f
(doesn't)4746 1200 w
( that)1 175( Notice)1 316(work sensibly.'')1 649 3 720 1320 t
10 CW f
(putc\(*c++,f\))1885 1320 w
10 R f
(works fine in this implementation.)4 1375 1 2630 1320 t
( instance, not everyone handles)4 1266( For)1 192(Some C implementations are less careful.)5 1674 3 970 1476 t
10 CW f
(putc\(*c++,f\))4131 1476 w
10 R f
(cor-)4880 1476 w
( another example, consider the)4 1246(rectly. As)1 419 2 720 1596 t
10 I f
(toupper)2416 1596 w
10 R f
( translates a)2 481( It)1 116(function that appears in many C libraries.)6 1686 3 2757 1596 t
( we)1 154( If)1 129( letter while leaving other characters unchanged.)6 2014(lower-case letter to the corresponding upper-case)5 2023 4 720 1716 t
( contiguous \(with a possible gap)5 1366(assume that all the lower-case letters and all the upper-case letters are)11 2954 2 720 1836 t
(between the cases\), we get the following function:)7 2007 1 720 1956 t
10 CW f
(toupper\(c\))1080 2136 w
({)1080 2256 w
(if \(c >= 'a' && c <= 'z'\))7 1500 1 1330 2376 t
(c += 'A' - 'a';)4 900 1 1580 2496 t
(return c;)1 540 1 1330 2616 t
(})1080 2736 w
10 R f
( is much longer than the actual calculations, so the)9 2041(In most C implementations, the subroutine call overhead)7 2279 2 720 2916 t
(implementor is tempted to make it a macro:)7 1746 1 720 3036 t
10 CW f
(#define toupper\(c\) \(\(c\)>='a' && \(c\)<='z'? \(c\)+\('A'-'a'\): \(c\)\))6 3660 1 1080 3216 t
10 R f
( it will cause a surprise for anyone who)8 1657( However,)1 451( in many cases.)3 643(This is indeed faster than the function)6 1569 4 720 3396 t
(tries to use)2 433 1 720 3516 t
10 CW f
(toupper\(*p++\))1178 3516 w
10 R f
(.)1958 3516 w
( generate very large expressions)4 1311(Another thing to watch out for when using macros is that they may)12 2759 2 970 3672 t
( example, look again at the definition of)7 1598(indeed. For)1 480 2 720 3792 t
10 CW f
(max)2823 3792 w
10 R f
(:)3003 3792 w
10 CW f
(#define max\(a,b\) \(\(a\)>\(b\)?\(a\):\(b\)\))2 2040 1 1080 3972 t
10 R f
(Suppose we want to use this definition to find the largest of)11 2379 1 720 4152 t
10 CW f
(a)3124 4152 w
10 R f
(,)3184 4152 w
10 CW f
(b)3234 4152 w
10 R f
(,)3294 4152 w
10 CW f
(c)3344 4152 w
10 R f
(, and)1 194 1 3404 4152 t
10 CW f
(d)3623 4152 w
10 R f
(. If we write the obvious:)5 1004 1 3683 4152 t
10 CW f
(max\(a,max\(b,max\(c,d\)\)\))1080 4332 w
10 R f
(this expands to:)2 628 1 720 4512 t
10 CW f
(\(\(a\)>\(\(\(b\)>\(\(\(c\)>\(d\)?\(c\):\(d\)\)\)?\(b\):\(\(\(c\)>\(d\)?\(c\):\(d\)\)\)\)\)?)1080 4692 w
(\(a\):\(\(\(b\)>\(\(\(c\)>\(d\)?\(c\):\(d\)\)\)?\(b\):\(\(\(c\)>\(d\)?\(c\):\(d\)\)\)\)\)\))1140 4812 w
10 R f
( can make it a little less large by balancing the operands:)11 2260( We)1 188(which is surprisingly large.)3 1088 3 720 4992 t
10 CW f
(max\(max\(a,b\),max\(c,d\)\))1080 5172 w
10 R f
(which gives:)1 508 1 720 5352 t
10 CW f
(\(\(\(\(a\)>\(b\)?\(a\):\(b\)\)\)>\(\(\(c\)>\(d\)?\(c\):\(d\)\)\)?)1080 5532 w
(\(\(\(a\)>\(b\)?\(a\):\(b\)\)\):\(\(\(c\)>\(d\)?\(c\):\(d\)\)\)\))1140 5652 w
10 R f
(Somehow, though, it seems easier to write:)6 1721 1 720 5832 t
10 CW f
(biggest = a;)2 720 1 1080 6012 t
(if \(biggest < b\) biggest = b;)6 1740 1 1080 6132 t
(if \(biggest < c\) biggest = c;)6 1740 1 1080 6252 t
(if \(biggest < d\) biggest = d;)6 1740 1 1080 6372 t
10 B f
( are not Type Definitions)4 1061(6.2. Macros)1 521 2 720 6672 t
10 R f
(One common use of macros is to permit several things in diverse places to be the same type:)17 3692 1 970 6828 t
cleartomark
showpage
saveobj restore
%%EndPage: 19 21
%%Page: 20 22
/saveobj save def
mark
22 pagesetup
10 R f
(- 20 -)2 216 1 2772 480 t
10 CW f
(#define FOOTYPE struct foo)3 1560 1 1080 900 t
(FOOTYPE a;)1 600 1 1080 1020 t
(FOOTYPE b, c;)2 780 1 1080 1140 t
10 R f
(This lets the programmer change the types of)7 1826 1 720 1320 t
10 CW f
(a)2574 1320 w
10 R f
(,)2634 1320 w
10 CW f
(b)2687 1320 w
10 R f
(, and)1 197 1 2747 1320 t
10 CW f
(c)2972 1320 w
10 R f
( line of the program, even if)6 1141(just by changing one)3 839 2 3060 1320 t
10 CW f
(a)720 1440 w
10 R f
(,)780 1440 w
10 CW f
(b)830 1440 w
10 R f
(, and)1 194 1 890 1440 t
10 CW f
(c)1109 1440 w
10 R f
(are declared in widely different places.)5 1550 1 1194 1440 t
( Most)1 259( supports it.)2 476(Using a macro definition for this has the advantage of portability \261 any C compiler)14 3335 3 970 1596 t
(C compilers also support another way of doing this:)8 2071 1 720 1716 t
10 CW f
(typedef struct foo FOOTYPE;)3 1620 1 1080 1896 t
10 R f
(This defines)1 491 1 720 2076 t
10 CW f
(FOOTYPE)1236 2076 w
10 R f
(as a new type that is equivalent to)7 1351 1 1681 2076 t
10 CW f
(struct foo)1 600 1 3057 2076 t
10 R f
(.)3657 2076 w
( type may appear to be equivalent, but the)8 1736(These two ways of naming a)5 1180 2 970 2232 t
10 CW f
(typedef)3919 2232 w
10 R f
(is more flexible.)2 668 1 4372 2232 t
(Consider, for example, the following:)4 1504 1 720 2352 t
10 CW f
(#define T1 struct foo *)4 1380 1 1080 2532 t
(typedef struct foo *T2;)3 1380 1 1080 2652 t
10 R f
(These definitions make)2 938 1 720 2832 t
10 CW f
(T1)1686 2832 w
10 R f
(and)1834 2832 w
10 CW f
(T2)2006 2832 w
10 R f
( pointer to a)3 492(conceptually equivalent to a)3 1132 2 2154 2832 t
10 CW f
(struct foo)1 604 1 3807 2832 t
10 R f
(. But look what)3 629 1 4411 2832 t
(happens when we try to use them with more than one variable:)11 2504 1 720 2952 t
10 CW f
(T1 a, b;)2 480 1 1080 3132 t
(T2 c, d;)2 480 1 1080 3252 t
10 R f
(The first declaration gets expanded to)5 1505 1 720 3432 t
10 CW f
(struct foo * a, b;)4 1080 1 1080 3612 t
10 R f
(This defines)1 492 1 720 3792 t
10 CW f
(a)1238 3792 w
10 R f
( defines)1 315(to be a pointer to a structure, but)7 1305 2 1324 3792 t
10 CW f
(b)2971 3792 w
10 R f
( second dec-)2 502( The)1 207(to be a structure \(not a pointer\).)6 1273 3 3058 3792 t
(laration, in contrast, defines both)4 1315 1 720 3912 t
10 CW f
(c)2060 3912 w
10 R f
(and)2145 3912 w
10 CW f
(d)2314 3912 w
10 R f
(as pointers to structures, because)4 1311 1 2399 3912 t
10 CW f
(T2)3735 3912 w
10 R f
(behaves as a true type.)4 900 1 3880 3912 t
10 B f
( Pitfalls)1 325(7. Portability)1 586 2 720 4152 t
10 R f
( one of the reasons to)5 901( Indeed,)1 356( people to run on many machines.)6 1411(C has been implemented by many)5 1402 4 970 4308 t
(write programs in C in the first place is that it is easy to move them from one programming environment to)20 4320 1 720 4428 t
(another.)720 4548 w
( dif-)1 171( Moreover,)1 470(However, because there are so many implementors, they do not all talk to each other.)14 3429 3 970 4704 t
( expect C implementations to differ slightly)6 1744(ferent systems have different requirements, so it is reasonable to)9 2576 2 720 4824 t
(between one machine and another.)4 1382 1 720 4944 t
( were associated with the)4 1043(Because so many of the early C implementations)7 2019 2 970 5100 t
9 R f
(UNIX)4065 5100 w
10 R f
(operating system,)1 715 1 4325 5100 t
( people started implementing C)4 1299( When)1 299( functions was shaped by that system.)6 1573(the nature of many of these)5 1149 4 720 5220 t
( tried to make the library behave in ways that would be familiar to programmers)14 3284(under other systems, they)3 1036 2 720 5340 t
(used to the)2 433 1 720 5460 t
9 R f
(UNIX)1176 5460 w
10 R f
(system.)1426 5460 w
( the world started)3 726( is more, as more people in different parts of)9 1855( What)1 275(They did not always succeed.)4 1214 4 970 5616 t
(working on different versions of the)5 1554 1 720 5736 t
9 R f
(UNIX)2320 5736 w
10 R f
( of some of the library functions)6 1419(system, the exact nature)3 1028 2 2593 5736 t
( in someone else's envi-)4 981( a C programmer who wishes to write programs useful)9 2204( Today,)1 333(inevitably diverged.)1 802 4 720 5856 t
(ronment must know about many of these subtle differences.)8 2388 1 720 5976 t
10 B f
( in a Name?)3 508(7.1. What's)1 511 2 720 6216 t
10 R f
( ignore charac-)2 607( Others)1 321( the characters of an identifier as being significant.)8 2060(Some C compilers treat all)4 1082 4 970 6372 t
( compilers usually produce object programs that must then)8 2382( C)1 122(ters past some limit when storing identifiers.)6 1816 3 720 6492 t
( in turn, often impose)4 902( Loaders,)1 407( be able to access library subroutines.)6 1560(be processed by loaders in order to)6 1451 4 720 6612 t
(their own restrictions on the kinds of names they can handle.)10 2427 1 720 6732 t
( When)1 298( in external names must be in upper case only.)9 1937(One common loader restriction is that letters)6 1835 3 970 6888 t
( to force all external names to upper case.)8 1675(faced with such a restriction, it is reasonable for a C implementor)11 2645 2 720 7008 t
(Restrictions of this sort are blessed by section 2.1 the C reference manual:)12 2959 1 720 7128 t
( underscore)1 466( The)1 209(An identifier is a sequence of letters and digits; the first character must be a letter.)15 3316 3 970 7284 t
10 S f
(_)4990 7284 w
cleartomark
showpage
saveobj restore
%%EndPage: 20 22
%%Page: 21 23
/saveobj save def
mark
23 pagesetup
10 R f
(- 21 -)2 216 1 2772 480 t
( charac-)1 320( more than the first eight)5 1005( No)1 176( and lower case letters are different.)6 1449( Upper)1 303(counts as as a letter.)4 817 6 970 840 t
( identifiers, which are used by various)6 1630( External)1 405(ters are significant, although more may be used.)7 2035 3 970 960 t
(assemblers and loaders, are more restricted:)5 1745 1 970 1080 t
( external)1 348(Here, the reference manual goes on to give examples of various implementations that restrict)13 3722 2 970 1236 t
(identifiers to a single case, or to fewer than eight characters, or both.)12 2735 1 720 1356 t
( is important to be careful when choosing identifiers in programs intended to be)13 3221(Because of all this, it)4 849 2 970 1512 t
( two subroutines named, say)4 1159(portable. Having)1 702 2 720 1632 t
10 CW f
(print_fields)2612 1632 w
10 R f
(and)3363 1632 w
10 CW f
(print_float)3537 1632 w
10 R f
(would not be a very)4 813 1 4227 1632 t
(good idea.)1 416 1 720 1752 t
(As a striking example, consider the following function:)7 2209 1 970 1908 t
10 CW f
(char *)1 360 1 1080 2088 t
(Malloc \(n\))1 600 1 1080 2208 t
(unsigned n;)1 660 1 1330 2328 t
({)1080 2448 w
(char *p, *malloc\(\);)2 1140 1 1330 2568 t
(p = malloc \(n\);)3 900 1 1330 2688 t
(if \(p == NULL\))3 840 1 1330 2808 t
(panic \("out of memory"\);)3 1440 1 1580 2928 t
(return p;)1 540 1 1330 3048 t
(})1080 3168 w
10 R f
( The)1 214(This function is a simple way of ensuring that running out of memory will not go undetected.)16 3856 2 970 3384 t
( program to allocate memory by calling)6 1611(idea is for the)3 564 2 720 3504 t
10 CW f
(Malloc)2925 3504 w
10 R f
(instead of)1 396 1 3315 3504 t
10 CW f
(malloc)3741 3504 w
10 R f
(. If)1 121 1 4101 3504 t
10 CW f
(malloc)4252 3504 w
10 R f
(ever fails,)1 398 1 4642 3504 t
(the result will be to call)5 996 1 720 3624 t
10 CW f
(panic)1752 3624 w
10 R f
( the program with an appropriate error)6 1601(which will presumably terminate)3 1351 2 2088 3624 t
(message.)720 3744 w
( that ignores case distinc-)4 1055(Consider, however, what happens when this function is used on a system)11 3015 2 970 3900 t
( effect, the names)3 763( In)1 153( identifiers.)1 475(tions in external)2 686 4 720 4020 t
10 CW f
(malloc)2842 4020 w
10 R f
(and)3247 4020 w
10 CW f
(Malloc)3436 4020 w
10 R f
( other)1 250( In)1 153(become equivalent.)1 796 3 3841 4020 t
(words, the library function)3 1077 1 720 4140 t
10 CW f
(malloc)1826 4140 w
10 R f
(is effectively replaced by the)4 1172 1 2216 4140 t
10 CW f
(Malloc)3418 4140 w
10 R f
(function above, which when it)4 1232 1 3808 4140 t
(calls)720 4260 w
10 CW f
(malloc)943 4260 w
10 R f
( of course, is that the first attempt to allocate memory)10 2279( result,)1 287( The)1 220(is really calling itself.)3 911 4 1343 4260 t
( loop and consequent mayhem, even though the function will work on an implementa-)13 3485(results in a recursion)3 835 2 720 4380 t
(tion that preserves case distinctions.)4 1440 1 720 4500 t
10 B f
( Big is an Integer?)4 778(7.2. How)1 400 2 720 4740 t
10 R f
( with charac-)2 538(C provides the programmer with three sizes of integers: ordinary, short, and long, and)13 3532 2 970 4896 t
( language definition does not guarantee much about)7 2101( The)1 210( integers.)1 371(ters, which behave as if they were small)7 1638 4 720 5016 t
(the relative sizes of the various kinds of integer:)8 1919 1 720 5136 t
( four sizes of integers are non-decreasing.)6 1664(1. The)1 405 2 720 5292 t
( ordinary integer is large enough to contain any array subscript.)10 2531(2. An)1 372 2 720 5448 t
( size of a character is natural for the particular hardware.)10 2255(3. The)1 405 2 720 5604 t
( few have 7- or 9-bit characters, so characters)8 1852(Most modern machines have 8-bit characters, though a)7 2218 2 970 5760 t
(are usually 7, 8, or 9 bits.)6 1013 1 720 5880 t
( at least 32 bits long, so that a long integer can be used to represent the size)17 3052(Long integers are usually)3 1018 2 970 6036 t
(of a file.)2 335 1 720 6156 t
( of)1 109(Ordinary integers are usually at least 16 bits long, because shorter integers would impose too much)15 3961 2 970 6312 t
(a restriction on the maximum size of an array.)8 1838 1 720 6432 t
(Short integers are almost always exactly 16 bits long.)8 2134 1 970 6588 t
( cannot count on having)4 1000( most important thing is that one)6 1355( The)1 214(What does this all mean in practice?)6 1501 4 970 6744 t
( can probably expect 16 bits for a short or an ordinary)11 2247( one)1 179( Informally,)1 512(any particular precision available.)3 1382 4 720 6864 t
( ordi-)1 224( can certainly use)3 707( One)1 220(integer, and 32 bits for a long integer, but not even those sizes are guaranteed.)14 3169 4 720 6984 t
( hold val-)2 391(nary integers to express table sizes and subscripts, but what about a variable that must be able to)17 3929 2 720 7104 t
(ues up to ten million?)4 867 1 720 7224 t
cleartomark
showpage
saveobj restore
%%EndPage: 21 23
%%Page: 22 24
/saveobj save def
mark
24 pagesetup
10 R f
(- 22 -)2 216 1 2772 480 t
(The most portable way to do that is probably to define a ``new'' type:)13 2787 1 970 840 t
10 CW f
(typedef long tenmil;)2 1200 1 1080 1020 t
10 R f
(Now one can use this type to declare a variable of that width and know that, at worst, one will have to)21 4320 1 720 1200 t
(change a single type definition to get all those variables to be the right type.)14 3027 1 720 1320 t
10 B f
( Characters Signed or Unsigned?)4 1411(7.3. Are)1 360 2 720 1560 t
10 R f
( 8-bit characters, so most modern C compilers implement characters)9 2759(Most modern computers support)3 1311 2 970 1716 t
( not all compilers interpret those 8-bit quantities the same way.)10 2517( However,)1 440(as 8-bit integers.)2 663 3 720 1836 t
( when converting a)3 786(The issue becomes important only)4 1399 2 970 1992 t
10 CW f
(char)3188 1992 w
10 R f
( the)1 155( Going)1 308(quantity to a larger integer.)4 1116 3 3461 1992 t
( a compiler converting a)4 970( But)1 195( results are well-defined: excess bits are simply discarded.)8 2320(other way, the)2 570 4 720 2112 t
10 CW f
(char)4800 2112 w
10 R f
(to an)1 209 1 720 2232 t
10 CW f
(int)966 2232 w
10 R f
(has a choice: should it treat the)6 1309 1 1183 2232 t
10 CW f
(char)2529 2232 w
10 R f
( it)1 94( the former,)2 492( If)1 128(as a signed or an unsigned quantity?)6 1520 4 2806 2232 t
(should expand the)2 737 1 720 2352 t
10 CW f
(char)1487 2352 w
10 R f
(to an)1 202 1 1757 2352 t
10 CW f
(int)1989 2352 w
10 R f
( the latter, it should fill the extra bit posi-)9 1680(by replicating the sign bit; if)5 1161 2 2199 2352 t
(tions with zeroes.)2 702 1 720 2472 t
( with characters with their)4 1094(The results of this decision are important to virtually anyone who deals)11 2976 2 970 2628 t
( be considered to range from)5 1169( determines whether 8-bit characters are going to)7 1995( It)1 117(high-order bits turned on.)3 1039 4 720 2748 t
( design things)2 566( in turn, affects the way the programmer will)8 1820( This,)1 257(\261128 through 127 or from 0 through 255.)7 1677 4 720 2868 t
(like hash tables and translate tables.)5 1431 1 720 2988 t
( with the high-order bit on is treated as a negative number, you)12 2570(If you care whether a character value)6 1500 2 970 3144 t
(should probably declare it as)4 1200 1 720 3264 t
10 CW f
(unsigned char)1 793 1 1958 3264 t
10 R f
( guaranteed to be zero-extended when)5 1574(. Such values are)3 715 2 2751 3264 t
(converted to integer, whereas ordinary)4 1549 1 720 3384 t
10 CW f
(char)2297 3384 w
10 R f
(variables may be signed in one implementation and unsigned)8 2474 1 2566 3384 t
(in another.)1 427 1 720 3504 t
(Incidentally, it is a common misconception that if)7 2128 1 970 3660 t
10 CW f
(c)3143 3660 w
10 R f
( obtain the)2 464(is a character variable, one can)5 1328 2 3248 3660 t
(unsigned integer equivalent of)3 1233 1 720 3780 t
10 CW f
(c)1984 3780 w
10 R f
(by writing)1 420 1 2075 3780 t
10 CW f
(\(unsigned\) c)1 720 1 2526 3780 t
10 R f
(. This fails because a)4 858 1 3246 3780 t
10 CW f
(char)4135 3780 w
10 R f
(quantity is con-)2 634 1 4406 3780 t
(verted to)1 354 1 720 3900 t
10 CW f
(int)1101 3900 w
10 R f
(before any operator is applied to it,)6 1412 1 1308 3900 t
10 I f
(even a cast)2 449 1 2748 3900 t
10 R f
(. Thus)1 278 1 3197 3900 t
10 CW f
(c)3503 3900 w
10 R f
(is converted first to a signed integer)6 1449 1 3591 3900 t
(and then to an unsigned integer, with possibly unexpected results.)9 2628 1 720 4020 t
(The right way to do it is)6 961 1 970 4176 t
10 CW f
(\(unsigned char\) c)2 1020 1 1956 4176 t
10 R f
(.)2976 4176 w
10 B f
( Right Shifts Signed or Unsigned?)5 1444(7.4. Are)1 360 2 720 4416 t
10 R f
( declare the quantities being)4 1136(This bears repeating: a program that cares how shifts are done had better)12 2934 2 970 4572 t
(shifted as unsigned.)2 791 1 720 4692 t
10 B f
( Does Division Truncate?)3 1075(7.5. How)1 400 2 720 4932 t
10 R f
(Suppose we divide)2 755 1 970 5088 t
10 CW f
(a)1750 5088 w
10 R f
(by)1835 5088 w
10 CW f
(b)1960 5088 w
10 R f
(to give a quotient)3 697 1 2045 5088 t
10 CW f
(q)2767 5088 w
10 R f
(and remainder)1 573 1 2852 5088 t
10 CW f
(r)3450 5088 w
10 R f
(:)3510 5088 w
10 CW f
(q = a / b;)4 600 1 1080 5268 t
(r = a % b;)4 600 1 1080 5388 t
10 R f
(For the moment, suppose also that)5 1372 1 720 5568 t
10 CW f
(b>0)2117 5568 w
10 R f
(.)2297 5568 w
(What relationships might we want to hold between)7 2034 1 970 5724 t
10 CW f
(a)3029 5724 w
10 R f
(,)3089 5724 w
10 CW f
(b)3139 5724 w
10 R f
(,)3199 5724 w
10 CW f
(p)3249 5724 w
10 R f
(, and)1 194 1 3309 5724 t
10 CW f
(q)3528 5724 w
10 R f
(?)3588 5724 w
( important, we want)3 799(1. Most)1 456 2 720 5880 t
10 CW f
(q*b + r == a)4 720 1 2000 5880 t
10 R f
(, because this is the relation that defines the remainder.)9 2193 1 2720 5880 t
( we change the sign of)5 895(2. If)1 316 2 720 6036 t
10 CW f
(a)1956 6036 w
10 R f
(, we want that to change the sign of)8 1417 1 2016 6036 t
10 CW f
(q)3458 6036 w
10 R f
(, but not the absolute value.)5 1102 1 3518 6036 t
( want to ensure that)4 798(3. We)1 388 2 720 6192 t
10 CW f
(r>=0)1935 6192 w
10 R f
(and)2204 6192 w
10 CW f
(r<b)2377 6192 w
10 R f
( the remainder is being used as an index to a)10 1819(. For instance, if)3 664 2 2557 6192 t
(hash table, it is important to be able to know that it will always be a valid index.)17 3201 1 970 6312 t
( Unfortu-)1 408( for integer division and remainder operations.)6 1909(These three properties are clearly desirable)5 1753 3 970 6468 t
(nately,)720 6588 w
10 I f
(they cannot all be true at once)6 1215 1 1014 6588 t
10 R f
(.)2229 6588 w
(Consider 3)1 438 1 970 6744 t
10 I f
(/)1416 6744 w
10 R f
( should be)2 417( What)1 269( satisfies property 1.)3 816( This)1 230(2, giving a quotient of 1 and a remainder of 1.)10 1856 5 1452 6744 t
(the value of)2 485 1 720 6864 t
10 S f
(-)1237 6864 w
10 R f
(3)1308 6864 w
10 I f
(/)1366 6864 w
10 R f
( 2 suggests that it should be)6 1148(2? Property)1 495 2 1402 6864 t
10 S f
(-)3077 6864 w
10 R f
( the remainder must)3 814(1, but if that is so,)5 755 2 3148 6864 t
10 I f
(also)4748 6864 w
10 R f
(be)4946 6864 w
10 S f
(-)720 6984 w
10 R f
( 3 by making the remainder 1, in)7 1416( we can satisfy property)4 1013( Alternatively,)1 617(1, which violates property 3.)4 1203 4 791 6984 t
(which case property 1 demands that the quotient be)8 2052 1 720 7104 t
10 S f
(-)2797 7104 w
10 R f
( violates property 2.)3 799(2. This)1 303 2 2868 7104 t
( give up at least one of)6 958(Thus C, and any language that implements truncating integer division, must)10 3112 2 970 7260 t
cleartomark
showpage
saveobj restore
%%EndPage: 22 24
%%Page: 23 25
/saveobj save def
mark
25 pagesetup
10 R f
(- 23 -)2 216 1 2772 480 t
(these three principles.)2 873 1 720 840 t
( same sign)2 430(Most programming languages give up number 3, saying instead that the remainder has the)13 3640 2 970 996 t
( C implementations do this in)5 1202( Most)1 260( possible to preserve properties 1 and 2.)7 1614( makes it)2 371( This)1 233(as the dividend.)2 640 6 720 1116 t
(practice, also.)1 551 1 720 1236 t
( with the property that)4 908(However, the C language definition only guarantees property 1, along)9 2828 2 970 1392 t
10 S f
(\357)4736 1409 w
10 I f
(r)4784 1392 w
10 S f
(\357)4823 1409 w
(<)4871 1392 w
(\357)4926 1409 w
10 I f
(b)4974 1392 w
10 S f
(\357)5024 1409 w
10 R f
(and that)1 322 1 720 1512 t
10 I f
(r)1070 1512 w
10 S f
(\263)1117 1512 w
10 R f
(0 whenever)1 465 1 1180 1512 t
10 I f
(a)1673 1512 w
10 S f
(\263)1731 1512 w
10 R f
(0 and)1 222 1 1794 1512 t
10 I f
(b)2044 1512 w
10 S f
(>)2118 1512 w
10 R f
( property is less restrictive than either property 2 or property 3,)11 2545(0. This)1 306 2 2189 1512 t
( rather strange implementations that would be unlikely to occur in practice \(such)12 3265(and actually permits some)3 1055 2 720 1632 t
(as an implementation that always truncates the quotient)7 2217 1 720 1752 t
10 I f
(away)2962 1752 w
10 R f
(from zero\).)1 448 1 3198 1752 t
( unwanted flexibility, the C definition is enough that we can usually make inte-)13 3190(Despite its sometimes)2 880 2 970 1908 t
( for example, that we have a)6 1133( Suppose,)1 415( what we want.)3 607(ger division do what we want, provided that we know)9 2165 4 720 2028 t
(number)720 2148 w
10 CW f
(n)1059 2148 w
10 R f
( identifier, and we want to use division to)8 1737(that represents some function of the characters in an)8 2150 2 1153 2148 t
(obtain a hash table entry)4 1016 1 720 2268 t
10 CW f
(h)1771 2268 w
10 R f
(such that 0)2 453 1 1866 2268 t
10 S f
(\243)2327 2268 w
10 I f
(h)2390 2268 w
10 S f
(<)2464 2268 w
10 I f
(HASHSIZE)2535 2268 w
10 R f
( we know that)3 593(. If)1 151 2 2990 2268 t
10 CW f
(n)3769 2268 w
10 R f
(is never negative, we simply)4 1176 1 3864 2268 t
(write)720 2388 w
10 CW f
(h = n % HASHSIZE;)4 1020 1 1080 2568 t
10 R f
(However, if)1 478 1 720 2748 t
10 CW f
(n)1225 2748 w
10 R f
(might be negative, this is not good enough, because)8 2089 1 1312 2748 t
10 CW f
(h)3429 2748 w
10 R f
( we)1 144( However,)1 443(might also be negative.)3 936 3 3517 2748 t
(know that)1 397 1 720 2868 t
10 I f
(h)1142 2868 w
10 S f
(> -)1 126 1 1216 2868 t
10 I f
(HASHSIZE)1358 2868 w
10 R f
(, so we can write:)4 701 1 1813 2868 t
10 CW f
(h = n % HASHSIZE;)4 1020 1 1080 3048 t
(if \(h < 0\))3 600 1 1080 3168 t
(h += HASHSIZE;)2 840 1 1330 3288 t
10 R f
(Better yet, declare n as)4 911 1 970 3504 t
10 CW f
(unsigned)1906 3504 w
10 R f
(.)2386 3504 w
10 B f
( Big is a Random Number?)5 1159(7.6. How)1 400 2 720 3744 t
10 R f
( the only C implementation ran on the)7 1547( When)1 292(This size ambiguity has affected library design as well.)8 2231 3 970 3900 t
9 R f
(PDP-11)720 4020 w
10 R f
( a function called)3 699(\263 computer, there was)3 887 2 1005 4020 t
10 I f
(rand)2619 4020 w
10 R f
(that returned a \(pseudo-\) random non-negative integer.)6 2204 1 2836 4020 t
9 R f
(PDP-11)720 4140 w
10 R f
(integers were 16 bits long, including the sign, so)8 2082 1 1048 4140 t
10 I f
(rand)3172 4140 w
10 R f
(would return an integer between 0 and)6 1637 1 3403 4140 t
(2)720 4260 w
7 R f
(15)775 4220 w
10 S f
(-)869 4260 w
10 R f
(1.)940 4260 w
( the)1 159(When C was implemented on)4 1226 2 970 4416 t
9 R f
(VAX-11)2392 4416 w
10 R f
( was the range of the)5 888( What)1 278(, integers were 32 bits long.)5 1167 3 2707 4416 t
10 I f
(rand)720 4536 w
10 R f
(function on the)2 605 1 934 4536 t
9 R f
(VAX-11)1564 4536 w
10 R f
(?)1879 4536 w
( of California took the view that)6 1336(For their system, the people at the University)7 1860 2 970 4692 t
10 I f
(rand)4200 4692 w
10 R f
(should return a)2 617 1 4423 4692 t
( version of)2 427(value that ranges over all possible non-negative integers, so their)9 2599 2 720 4812 t
10 I f
(rand)3771 4812 w
10 R f
(returns an integer between)3 1055 1 3985 4812 t
(0 and 2)2 294 1 720 4932 t
7 R f
(31)1019 4892 w
10 S f
(-)1113 4932 w
10 R f
(1.)1184 4932 w
( AT&T, on the other hand, decided that a)8 1719(The people at)2 559 2 970 5088 t
9 R f
(PDP-11)3282 5088 w
10 R f
(program that expected the result of)5 1439 1 3601 5088 t
10 I f
(rand)720 5208 w
10 R f
(to be less than 2)4 696 1 948 5208 t
7 R f
(15)1649 5168 w
10 R f
(would be easier to transport to a)6 1359 1 1765 5208 t
9 R f
(VAX-11)3162 5208 w
10 R f
(if the)1 221 1 3515 5208 t
10 I f
(rand)3774 5208 w
10 R f
(function returned a value)3 1039 1 4001 5208 t
(between 0 and 2)3 651 1 720 5328 t
7 R f
(15)1376 5288 w
10 R f
(there, too.)1 402 1 1479 5328 t
( that uses)2 380(As a result, it is now difficult to write a program)10 1964 2 970 5484 t
10 I f
(rand)3343 5484 w
10 R f
(without tailoring it to the implemen-)5 1479 1 3561 5484 t
(tation.)720 5604 w
10 B f
( Conversion)1 514(7.7. Case)1 405 2 720 5844 t
10 R f
(The)970 6000 w
10 I f
(toupper)1150 6000 w
10 R f
(and)1486 6000 w
10 I f
(tolower)1655 6000 w
10 R f
( were originally written as macros:)5 1389( They)1 255(functions have a similar history.)4 1285 3 1986 6000 t
10 CW f
(#define toupper\(c\) \(\(c\)+'A'-'a'\))2 1920 1 1080 6180 t
(#define tolower\(c\) \(\(c\)+'a'-'A'\))2 1920 1 1080 6300 t
10 R f
( letter as input)3 590(When given a lower-case)3 1028 2 720 6480 t
10 I f
(toupper)2370 6480 w
10 R f
(yields the corresponding upper-case letter.)4 1721 1 2713 6480 t
10 I f
(Tolower)4491 6480 w
10 R f
(does)4857 6480 w
( to the extent that they)5 986( these macros depend on the implementation's character set)8 2538( Both)1 265(the opposite.)1 531 4 720 6600 t
( lower-case letter be the)4 1020(demand that the difference between an upper-case letter and the corresponding)10 3300 2 720 6720 t
( assumption is valid for both the)6 1361( This)1 240(same constant for all letters.)4 1171 3 720 6840 t
8 R f
(ASCII)3524 6840 w
10 R f
(and)3770 6840 w
8 R f
(EBCDIC)3951 6840 w
10 R f
(character sets, and)2 757 1 4283 6840 t
( definitions can be encapsulated in)5 1398(probably isn't too dangerous, because the non-portability of these macro)9 2922 2 720 6960 t
8 S1 f
(__________________)720 7060 w
8 R f
(\263 PDP-11 and VAX-11 are Trademarks of Digital Equipment Corporation.)9 2386 1 720 7160 t
cleartomark
showpage
saveobj restore
%%EndPage: 23 25
%%Page: 24 26
/saveobj save def
mark
26 pagesetup
10 R f
(- 24 -)2 216 1 2772 480 t
(the single file that contains them.)5 1327 1 720 840 t
( have one disadvantage, though: when given something that is not a letter of the)14 3370(These macros do)2 700 2 970 996 t
( the following innocent program fragment to convert a file to)10 2505( Thus,)1 283(appropriate case, they return garbage.)4 1532 3 720 1116 t
(lower case doesn't work with these macros:)6 1746 1 720 1236 t
10 CW f
(int c;)1 360 1 1080 1416 t
(while \(\(c = getchar\(\)\) != EOF\))5 1800 1 1080 1536 t
(putchar \(tolower \(c\)\);)2 1320 1 1330 1656 t
10 R f
(Instead, one must write:)3 960 1 720 1836 t
10 CW f
(int c;)1 360 1 1080 2016 t
(while \(\(c = getchar\(\)\) != EOF\))5 1800 1 1080 2136 t
(putchar \(isupper \(c\)? tolower \(c\): c\);)5 2280 1 1330 2256 t
10 R f
( soul in the)3 478(At one point, some enterprising)4 1307 2 970 2472 t
9 R f
(UNIX)2790 2472 w
10 R f
(development organization at AT&T noticed that)5 1988 1 3052 2472 t
(most uses of)2 516 1 720 2592 t
10 I f
(toupper)1269 2592 w
10 R f
(and)1613 2592 w
10 I f
(tolower)1790 2592 w
10 R f
( tests to ensure that their arguments were appropriate.)8 2193(were preceded by)2 718 2 2129 2592 t
(He considered rewriting the macros this way:)6 1813 1 720 2712 t
10 CW f
(#define toupper\(c\) \(\(c\) >= 'a' && \(c\) <= 'z'? \(c\) + 'A' - 'a': \(c\)\))14 4020 1 1080 2892 t
(#define tolower\(c\) \(\(c\) >= 'A' && \(c\) <= 'Z'? \(c\) + 'a' - 'A': \(c\)\))14 4020 1 1080 3012 t
10 R f
(but realized that this would cause)5 1374 1 720 3192 t
10 CW f
(c)2128 3192 w
10 R f
(to be evaluated anywhere between one and three times for each call,)11 2818 1 2222 3192 t
(which would play havoc with expressions like)6 1896 1 720 3312 t
10 CW f
(toupper\(*p++\))2649 3312 w
10 R f
(. Instead, he decided to rewrite)5 1267 1 3429 3312 t
10 I f
(toupper)4729 3312 w
10 R f
(and)720 3432 w
10 I f
(tolower)889 3432 w
10 R f
(as functions.)1 505 1 1220 3432 t
10 I f
(Toupper)1775 3432 w
10 R f
(now looked something like this:)4 1284 1 2139 3432 t
10 CW f
(int toupper \(c\))2 900 1 1080 3612 t
(int c;)1 360 1 1330 3732 t
({)1080 3852 w
(if \(c >= 'a' && c <= 'z'\))7 1500 1 1330 3972 t
(return c + 'A' - 'a';)5 1260 1 1580 4092 t
(return c;)1 540 1 1330 4212 t
(})1080 4332 w
10 R f
(and)720 4512 w
10 I f
(tolower)889 4512 w
10 R f
(looked similar.)1 600 1 1220 4512 t
( call overhead into)3 770(This change had the advantage of robustness, at the cost of introducing function)12 3300 2 970 4668 t
( some people might not be willing to pay the cost of this)12 2274( hero realized that)3 726( Our)1 208(each use of these functions.)4 1112 4 720 4788 t
(overhead, so he re-introduced the macros with new names:)8 2347 1 720 4908 t
10 CW f
(#define _toupper\(c\) \(\(c\)+'A'-'a'\))2 1980 1 1080 5088 t
(#define _tolower\(c\) \(\(c\)+'a'-'A'\))2 1980 1 1080 5208 t
10 R f
(This gave users a choice of convenience or speed.)8 1991 1 720 5388 t
( just one problem in all this: the people at Berkeley never followed suit, nor did some other)17 3658(There was)1 412 2 970 5544 t
( that uses)2 376( means that a program written on an AT&T system)9 2066( This)1 231(C implementors.)1 670 4 720 5664 t
10 I f
(toupper)4090 5664 w
10 R f
(or)4428 5664 w
10 I f
(tolower)4538 5664 w
10 R f
(, and)1 196 1 4844 5664 t
( a letter of the appropriate case, may stop work-)9 1951(assumes that it will be able to pass an argument that is not)12 2369 2 720 5784 t
(ing on some other C implementation.)5 1489 1 720 5904 t
(This sort of failure is very hard to trace for someone who does not know this bit of history.)18 3626 1 970 6060 t
10 B f
( First, then Reallocate)3 937(7.8. Free)1 393 2 720 6300 t
10 R f
(Most C implementations provide users with three memory allocation functions called)10 3414 1 970 6456 t
10 I f
(malloc)4409 6456 w
10 R f
(,)4681 6456 w
10 I f
(realloc)4732 6456 w
10 R f
(,)5015 6456 w
(and)720 6576 w
10 I f
(free)898 6576 w
10 R f
(. Calling)1 379 1 1053 6576 t
10 CW f
(malloc\(n\))1466 6576 w
10 R f
(returns a pointer to)3 784 1 2040 6576 t
10 I f
(n)2858 6576 w
10 R f
(characters of newly-allocated memory that the pro-)6 2098 1 2942 6576 t
( Giving)1 343(grammer can use.)2 736 2 720 6696 t
10 I f
(free)1839 6696 w
10 R f
(a pointer to memory previously returned by)6 1832 1 2034 6696 t
10 I f
(malloc)3907 6696 w
10 R f
(makes that memory)2 820 1 4220 6696 t
( Calling)1 350( re-use.)1 298(available for)1 507 3 720 6816 t
10 I f
(realloc)1905 6816 w
10 R f
(with a pointer to an allocated area and a new size stretches or shrinks)13 2822 1 2218 6816 t
(the memory to the new size, possibly copying it in the process.)11 2512 1 720 6936 t
( is an excerpt from the)5 965( Here)1 257( subtle.)1 303( truth is actually somewhat more)5 1372( The)1 218(Or so one might think.)4 955 6 970 7092 t
(description of)1 552 1 720 7212 t
10 I f
(realloc)1297 7212 w
10 R f
(that appears in the System V Interface Definition:)7 1988 1 1605 7212 t
cleartomark
showpage
saveobj restore
%%EndPage: 24 26
%%Page: 25 27
/saveobj save def
mark
27 pagesetup
10 R f
(- 25 -)2 216 1 2772 480 t
10 I f
(Realloc)970 840 w
10 R f
( to by)2 234(changes the size of the block pointed)6 1487 2 1302 840 t
10 I f
(ptr)3051 840 w
10 R f
(to)3196 840 w
10 I f
(size)3302 840 w
10 R f
(bytes and returns a pointer to the \(pos-)7 1560 1 3480 840 t
( contents will be unchanged up to the lesser of the new and old sizes.)14 2754( The)1 205(sibly moved\) block.)2 797 3 970 960 t
( the)1 154(The Seventh Edition of the reference manual for)7 1973 2 720 1116 t
9 R f
(UNIX)2877 1116 w
10 R f
(system contains a copy of the same paragraph.)7 1906 1 3134 1116 t
(In addition, it contains a second paragraph describing)7 2135 1 720 1236 t
10 I f
(realloc)2880 1236 w
10 R f
(:)3163 1236 w
10 I f
(Realloc)970 1392 w
10 R f
(also works if)2 526 1 1304 1392 t
10 I f
(ptr)1860 1392 w
10 R f
(points to a block freed since the last call of)9 1756 1 2007 1392 t
10 I f
(malloc)3793 1392 w
10 R f
(,)4065 1392 w
10 I f
(realloc)4120 1392 w
10 R f
(, or)1 138 1 4403 1392 t
10 I f
(calloc)4571 1392 w
10 R f
(; thus)1 225 1 4815 1392 t
(sequences of)1 523 1 970 1512 t
10 I f
(free)1529 1512 w
10 R f
(,)1684 1512 w
10 I f
(malloc)1745 1512 w
10 R f
(and)2053 1512 w
10 I f
(realloc)2233 1512 w
10 R f
( of)1 118(can exploit the search strategy)4 1252 2 2552 1512 t
10 I f
(malloc)3957 1512 w
10 R f
(to do storage com-)3 776 1 4264 1512 t
(paction.)970 1632 w
(Thus, the following is legal under the Seventh Edition:)8 2191 1 720 1788 t
10 CW f
(free \(p\);)1 540 1 1080 1968 t
(p = realloc \(p, newsize\);)4 1500 1 1080 2088 t
10 R f
( storage)1 314(This idiosyncrasy remains in systems derived from the Seventh Edition: it is possible to free a)15 3756 2 970 2304 t
( memory on these systems is guaranteed not to change)9 2235( implication, freeing)2 829( By)1 175(area and then reallocate it.)4 1081 4 720 2424 t
( elements)1 386( on these systems, one can free all the)8 1545( Thus,)1 280(its contents until the next time memory is allocated.)8 2109 4 720 2544 t
(of a list by the following curious means:)7 1613 1 720 2664 t
10 CW f
(for \(p = head; p != NULL; p = p->next\))9 2280 1 1080 2844 t
(free \(\(char *\) p\);)3 1080 1 1330 2964 t
10 R f
(without worrying that the call to)5 1291 1 720 3144 t
10 I f
(free)2036 3144 w
10 R f
(might invalidate)1 653 1 2216 3144 t
10 CW f
(p->next)2894 3144 w
10 R f
(.)3314 3144 w
(Needless to say, this technique is not recommended, if only because not all C implementations pre-)15 4070 1 970 3300 t
( the Seventh Edition manual leaves one thing)7 1835( However,)1 444( been freed.)2 475(serve memory long enough after it has)6 1566 4 720 3420 t
(unstated: the original implementation of)4 1605 1 720 3540 t
10 I f
(realloc)2350 3540 w
10 R f
(actually required that the area given to it for reallocation be)10 2382 1 2658 3540 t
( free memory first and then real-)6 1314( this reason, there are many C programs floating around that)10 2444( For)1 193(free first.)1 369 4 720 3660 t
(locate it, and this is something to watch out for when moving a C program to another implementation.)17 4078 1 720 3780 t
10 B f
( Example of Portability Problems)4 1427(7.9. An)1 328 2 720 4020 t
10 R f
( following pro-)2 613( The)1 209( many people.)2 571(Let's take a look at a problem that has been solved many times by)13 2677 4 970 4176 t
( converts the integer to decimal)5 1291( It)1 118( a \(pointer to a\) function.)5 1033(gram takes two arguments: a long integer and)7 1878 4 720 4296 t
(and calls the given function with each character of the decimal representation.)11 3114 1 720 4416 t
10 CW f
(void)1080 4596 w
(printnum \(n, p\))2 900 1 1080 4716 t
(long n;)1 420 1 1330 4836 t
(void \(*p\)\(\);)1 720 1 1330 4956 t
({)1080 5076 w
(if \(n < 0\) {)4 720 1 1330 5196 t
(\(*p\) \('-'\);)1 660 1 1580 5316 t
(n = -n;)2 420 1 1580 5436 t
(})1330 5556 w
(if \(n >= 10\))3 720 1 1330 5676 t
(printnum \(n/10, p\);)2 1140 1 1580 5796 t
(\(*p\) \(n % 10 + '0'\);)5 1200 1 1330 5916 t
(})1080 6036 w
10 R f
( we check if)3 523( First)1 247(This program is fairly straightforward.)4 1591 3 970 6252 t
10 CW f
(n)3369 6252 w
10 R f
( and)1 183(is negative; if so, we print a sign)7 1390 2 3467 6252 t
(make)720 6372 w
10 CW f
(n)964 6372 w
10 R f
( we test if)3 400(positive. Next,)1 614 2 1052 6372 t
10 I f
(n)2094 6372 w
10 S f
(\263)2152 6372 w
10 R f
( so, its decimal representation has two or more digits, so we call)12 2582(10. If)1 243 2 2215 6372 t
10 I f
(printnum)720 6492 w
10 R f
( we print the last digit.)5 900( Finally,)1 359(recursively to print all but the last digit.)7 1583 3 1112 6492 t
( uses to)2 302( first is the method it)5 831( The)1 205(This program, for all its simplicity, has several portability problems.)9 2732 4 970 6648 t
( of)1 114(convert the low-order decimal digit)4 1442 2 720 6768 t
10 CW f
(n)2307 6768 w
10 R f
( Using)1 295(to character form.)2 723 2 2398 6768 t
10 CW f
(n%10)3447 6768 w
10 R f
(to get the value of the low-order)6 1322 1 3718 6768 t
(digit is fine, but adding)4 971 1 720 6888 t
10 CW f
('0')1726 6888 w
10 R f
( addition)1 364( This)1 239( corresponding character representation is not.)5 1900(to it to get the)4 596 4 1941 6888 t
( that)1 176(assumes that the machine collating sequence has all the digits in sequence with no gaps, so)15 3659 2 720 7008 t
10 CW f
('0'+5)4581 7008 w
10 R f
(has)4907 7008 w
(the same value as)3 737 1 720 7128 t
10 CW f
('5')1494 7128 w
10 R f
( assumption, while true of the)5 1253( This)1 241( on.)1 163(, and so)2 332 4 1674 7128 t
8 R f
(ASCII)3701 7128 w
10 R f
(and)3948 7128 w
8 R f
(EBCDIC)4130 7128 w
10 R f
(character sets,)1 577 1 4463 7128 t
( way to avoid that problem is to use a table:)10 1743( The)1 205(might not be true for some machines.)6 1490 3 720 7248 t
cleartomark
showpage
saveobj restore
%%EndPage: 25 27
%%Page: 26 28
/saveobj save def
mark
28 pagesetup
10 R f
(- 26 -)2 216 1 2772 480 t
10 CW f
(void)1080 900 w
(printnum \(n, p\))2 900 1 1080 1020 t
(long n;)1 420 1 1330 1140 t
(void \(*p\)\(\);)1 720 1 1330 1260 t
({)1080 1380 w
(if \(n < 0\) {)4 720 1 1330 1500 t
(\(*p\) \('-'\);)1 660 1 1580 1620 t
(n = -n;)2 420 1 1580 1740 t
(})1330 1860 w
(if \(n >= 10\))3 720 1 1330 1980 t
(printnum \(n/10, p\);)2 1140 1 1580 2100 t
(\(*p\) \("0123456789"[n % 10]\);)3 1680 1 1330 2220 t
(})1080 2340 w
10 R f
( involves what happens if)4 1049(The next problem)2 722 2 970 2556 t
10 I f
(n)2773 2556 w
10 S f
(<)2847 2556 w
10 R f
( program prints a negative sign and sets)7 1633(0. The)1 287 2 2918 2556 t
10 CW f
(n)4870 2556 w
10 R f
(to)4962 2556 w
10 CW f
(-n)720 2676 w
10 R f
( because 2's complement machines generally allow more negative val-)9 2824(. This assignment might overflow,)4 1376 2 840 2676 t
( \(long\) integer is)3 666( particular, if a)3 587( In)1 133(ues than positive values to be represented.)6 1683 4 720 2796 t
10 I f
(k)3815 2796 w
10 R f
(bits plus one extra bit for the)6 1155 1 3885 2796 t
(sign,)720 2916 w
10 S f
(-)937 2916 w
10 R f
(2)1008 2916 w
7 I f
(k)1063 2876 w
10 R f
(can be represented but 2)4 969 1 1127 2916 t
7 I f
(k)2101 2876 w
10 R f
(cannot.)2165 2916 w
( most obvious one is to assign)6 1231( The)1 210( around this problem.)3 870(There are several ways)3 927 4 970 3072 t
10 CW f
(n)4238 3072 w
10 R f
(to an)1 202 1 4328 3072 t
10 CW f
(unsigned)4560 3072 w
(long)720 3192 w
10 R f
( some C compilers do not implement)6 1490( However,)1 442( be done with it.)4 655(value and)1 388 4 988 3192 t
10 CW f
(unsigned long)1 782 1 3990 3192 t
10 R f
(, so let)2 268 1 4772 3192 t
(us see how we can get along without it.)8 1573 1 720 3312 t
( machines, changing the sign of a)6 1424(In both 1's complement and 2's complement)6 1871 2 970 3468 t
10 I f
(positive)4305 3468 w
10 R f
(integer is)1 384 1 4656 3468 t
( only trouble comes when changing the sign of a)9 2011( The)1 213( to overflow.)2 529(guaranteed not)1 599 4 720 3588 t
10 I f
(negative)4105 3588 w
10 R f
(value. There-)1 564 1 4476 3588 t
(fore, we can avoid trouble by making sure we do not attempt to make)13 2773 1 720 3708 t
10 CW f
(n)3518 3708 w
10 R f
(positive.)3603 3708 w
(Of course, once we have printed the sign of a negative value, we would like to be able to treat nega-)20 4070 1 970 3864 t
( way to do that is to force)7 1053( The)1 210( the same way.)3 608(tive and positive numbers)3 1048 4 720 3984 t
10 CW f
(n)3669 3984 w
10 R f
(to be negative after printing the)5 1281 1 3759 3984 t
( we do this, we will have to ensure that the part of)12 2018( If)1 118( negative values.)2 672(sign, and to do all our arithmetic with)7 1512 4 720 4104 t
( to split the program into)5 997(the program that prints the sign is executed only once; the easiest way to do that is)16 3323 2 720 4224 t
(two functions:)1 575 1 720 4344 t
10 CW f
(void)1080 4524 w
(printnum \(n, p\))2 900 1 1080 4644 t
(long n;)1 420 1 1330 4764 t
(void \(*p\)\(\);)1 720 1 1330 4884 t
({)1080 5004 w
(void printneg\(\);)1 960 1 1330 5124 t
(if \(n < 0\) {)4 720 1 1330 5244 t
(\(*p\) \('-'\);)1 660 1 1580 5364 t
(printneg \(n, p\);)2 960 1 1580 5484 t
(} else)1 360 1 1330 5604 t
(printneg \(-n, p\);)2 1020 1 1580 5724 t
(})1080 5844 w
(void)1080 6084 w
(printneg \(n, p\))2 900 1 1080 6204 t
(long n;)1 420 1 1330 6324 t
(void \(*p\)\(\);)1 720 1 1330 6444 t
({)1080 6564 w
(if \(n <= -10\))3 780 1 1330 6684 t
(printneg \(n/10, p\);)2 1140 1 1580 6804 t
(\(*p\) \("0123456789"[-\(n % 10\)]\);)3 1860 1 1330 6924 t
(})1080 7044 w
10 I f
(Printnum)970 7260 w
10 R f
( In)1 140( a negative sign.)3 670(now just checks if the number being printed is negative; if so it prints)13 2851 3 1379 7260 t
cleartomark
showpage
saveobj restore
%%EndPage: 26 28
%%Page: 27 29
/saveobj save def
mark
29 pagesetup
10 R f
(- 27 -)2 216 1 2772 480 t
(either case, it calls)3 779 1 720 840 t
10 I f
(printneg)1538 840 w
10 R f
(with the negative absolute value of)5 1465 1 1916 840 t
10 I f
(n)3420 840 w
10 R f
( have also modified the body of)6 1343(. We)1 227 2 3470 840 t
10 I f
(printneg)720 960 w
10 R f
(to cater to the fact that)5 895 1 1084 960 t
10 I f
(n)2004 960 w
10 R f
(will always be a negative number or zero.)7 1668 1 2079 960 t
( have used)2 423( We)1 189(Or have we?)2 505 3 970 1116 t
10 CW f
(n/10)2113 1116 w
10 R f
(and)2379 1116 w
10 CW f
(n%10)2549 1116 w
10 R f
(to represent the leading digits and the trailing digit of)9 2147 1 2816 1116 t
10 I f
(n)4990 1116 w
10 R f
( that integer division behaves in a somewhat implementation-dependent)8 2869( Recall)1 305(\(with suitable sign changes\).)3 1146 3 720 1236 t
( that reason, it might actually be that)7 1502( For)1 195( operands is negative.)3 883(way when one of the)4 851 4 720 1356 t
10 CW f
(n%10)4182 1356 w
10 R f
( In)1 139(is positive!)1 448 2 4453 1356 t
(that case,)1 371 1 720 1476 t
10 CW f
(-\(n%10\))1116 1476 w
10 R f
(would be negative, and we would run off the end of our digit array.)13 2686 1 1561 1476 t
( temporary variables to hold the quotient and remainder.)8 2353(We cater to this problem by creating two)7 1717 2 970 1632 t
( the division, we check that the remainder is in range and adjust both variables if not.)16 3411(After we do)2 480 2 720 1752 t
10 I f
(Printnum)4662 1752 w
10 R f
(has not changed, so we show only)6 1362 1 720 1872 t
10 I f
(printneg)2107 1872 w
10 R f
(:)2446 1872 w
10 CW f
(void)1080 2052 w
(printneg \(n, p\))2 900 1 1080 2172 t
(long n;)1 420 1 1330 2292 t
(void \(*p\)\(\);)1 720 1 1330 2412 t
({)1080 2532 w
(long q;)1 420 1 1330 2652 t
(int r;)1 360 1 1330 2772 t
(q = n / 10;)4 660 1 1330 3012 t
(r = n % 10;)4 660 1 1330 3132 t
(if \(r > 0\) {)4 720 1 1330 3252 t
(r -= 10;)2 480 1 1580 3372 t
(q++;)1580 3492 w
(})1330 3612 w
(if \(n <= -10\))3 780 1 1330 3732 t
(printneg \(q, p\);)2 960 1 1580 3852 t
(\(*p\) \("0123456789"[-r]\);)1 1440 1 1330 3972 t
(})1080 4092 w
10 B f
( Space Available)2 706(8. This)1 315 2 720 4392 t
10 R f
( If)1 121(There are many ways for C programmers to go astray that have not been mentioned in this paper.)17 3949 2 970 4548 t
( may well be included, with an acknowledging footnote, in a)10 2549( It)1 124(you find one, please contact the author.)6 1647 3 720 4668 t
(future revision.)1 610 1 720 4788 t
10 B f
(References)720 5028 w
10 I f
(The C Programming Language)3 1256 1 970 5184 t
10 R f
(\(Kernighan and Ritchie, Prentice-Hall 1978\) is the definitive work on)9 2788 1 2252 5184 t
( already familiar with other high-level lan-)6 1702( contains both an excellent tutorial, aimed at people who are)10 2414(C. It)1 204 3 720 5304 t
( the language has)3 750( While)1 314( entire language succinctly.)3 1152(guages, and a reference manual that describes the)7 2104 4 720 5424 t
( book also contains the)4 928( This)1 231(expanded slightly since 1978, this book is still the last word on most subjects.)13 3161 3 720 5544 t
(``C Reference Manual'' we have mentioned several times in this paper.)10 2851 1 720 5664 t
10 I f
( Book)1 231(The C Puzzle)2 528 2 970 5820 t
10 R f
( The)1 206(\(Feuer, Prentice-Hall, 1982\) is an unusual way to hone one's syntactic skills.)11 3079 2 1755 5820 t
( a collection of puzzles \(and answers\) whose solutions test the reader's knowledge of C's fine)15 4009(book is)1 311 2 720 5940 t
(points.)720 6060 w
10 I f
(C: A Reference Manual)3 963 1 970 6216 t
10 R f
(\(Harbison and Steele, Prentice Hall 1984\) is mostly intended as a reference)11 3076 1 1964 6216 t
( users may also find it useful, particularly because of its meticulous cross)12 3017( Other)1 285( implementors.)1 608(source for)1 410 4 720 6336 t
(references.)720 6456 w
cleartomark
showpage
saveobj restore
%%EndPage: 27 29
%%Trailer
done
%%Pages: 29
%%DocumentFonts: Courier Times-Bold Times-Italic Times-Roman Times-Roman Symbol