1PERLTRAP(1) Perl Programmers Reference Guide PERLTRAP(1)
2
6 perltrap - Perl traps for the unwary
7
9 The biggest trap of all is forgetting to "use warnings" or use the -w
10 switch; see perllexwarn and perlrun. The second biggest trap is not
11 making your entire program runnable under "use strict". The third big‐
12 gest trap is not reading the list of changes in this version of Perl;
13 see perldelta.
14 Awk Traps
16 Accustomed awk users should take special note of the following:
18 · A Perl program executes only once, not once for each input line.
20 You can do an implicit loop with "-n" or "-p".
21 · The English module, loaded via
23 use English;
25 allows you to refer to special variables (like $/) with names (like
27 $RS), as though they were in awk; see perlvar for details.
28 · Semicolons are required after all simple statements in Perl (except
30 at the end of a block). Newline is not a statement delimiter.
31 · Curly brackets are required on "if"s and "while"s.
33 · Variables begin with "$", "@" or "%" in Perl.
35 · Arrays index from 0. Likewise string positions in substr() and
37 index().
38 · You have to decide whether your array has numeric or string
40 indices.
41 · Hash values do not spring into existence upon mere reference.
43 · You have to decide whether you want to use string or numeric com‐
45 parisons.
46 · Reading an input line does not split it for you. You get to split
48 it to an array yourself. And the split() operator has different
49 arguments than awk's.
50 · The current input line is normally in $_, not $0. It generally
52 does not have the newline stripped. ($0 is the name of the program
53 executed.) See perlvar.
54 · $<digit> does not refer to fields--it refers to substrings matched
56 by the last match pattern.
57 · The print() statement does not add field and record separators
59 unless you set $, and "$\". You can set $OFS and $ORS if you're
60 using the English module.
61 · You must open your files before you print to them.
63 · The range operator is "..", not comma. The comma operator works as
65 in C.
66 · The match operator is "=~", not "~". ("~" is the one's complement
68 operator, as in C.)
69 · The exponentiation operator is "**", not "^". "^" is the XOR oper‐
71 ator, as in C. (You know, one could get the feeling that awk is
72 basically incompatible with C.)
73 · The concatenation operator is ".", not the null string. (Using the
75 null string would render "/pat/ /pat/" unparsable, because the
76 third slash would be interpreted as a division operator--the tok‐
77 enizer is in fact slightly context sensitive for operators like
78 "/", "?", and ">". And in fact, "." itself can be the beginning of
79 a number.)
80 · The "next", "exit", and "continue" keywords work differently.
82 · The following variables work differently:
84 Awk Perl
86 ARGC scalar @ARGV (compare with $#ARGV)
87 ARGV[0] $0
88 FILENAME $ARGV
89 FNR $. - something
90 FS (whatever you like)
91 NF $#Fld, or some such
92 NR $.
93 OFMT $#
94 OFS $,
95 ORS $\
96 RLENGTH length($&)
97 RS $/
98 RSTART length($`)
99 SUBSEP $;
100 · You cannot set $RS to a pattern, only a string.
102 · When in doubt, run the awk construct through a2p and see what it
104 gives you.
105 C/C++ Traps
107 Cerebral C and C++ programmers should take note of the following:
109 · Curly brackets are required on "if"'s and "while"'s.
111 · You must use "elsif" rather than "else if".
113 · The "break" and "continue" keywords from C become in Perl "last"
115 and "next", respectively. Unlike in C, these do not work within a
116 "do { } while" construct. See "Loop Control" in perlsyn.
117 · There's no switch statement. (But it's easy to build one on the
119 fly, see "Basic BLOCKs and Switch Statements" in perlsyn)
120 · Variables begin with "$", "@" or "%" in Perl.
122 · Comments begin with "#", not "/*" or "//". Perl may interpret
124 C/C++ comments as division operators, unterminated regular expres‐
125 sions or the defined-or operator.
126 · You can't take the address of anything, although a similar operator
128 in Perl is the backslash, which creates a reference.
129 · "ARGV" must be capitalized. $ARGV[0] is C's "argv[1]", and
131 "argv[0]" ends up in $0.
132 · System calls such as link(), unlink(), rename(), etc. return
134 nonzero for success, not 0. (system(), however, returns zero for
135 success.)
136 · Signal handlers deal with signal names, not numbers. Use "kill -l"
138 to find their names on your system.
139 Sed Traps
141 Seasoned sed programmers should take note of the following:
143 · A Perl program executes only once, not once for each input line.
145 You can do an implicit loop with "-n" or "-p".
146 · Backreferences in substitutions use "$" rather than "\".
148 · The pattern matching metacharacters "(", ")", and "⎪" do not have
150 backslashes in front.
151 · The range operator is "...", rather than comma.
153 Shell Traps
155 Sharp shell programmers should take note of the following:
157 · The backtick operator does variable interpolation without regard to
159 the presence of single quotes in the command.
160 · The backtick operator does no translation of the return value,
162 unlike csh.
163 · Shells (especially csh) do several levels of substitution on each
165 command line. Perl does substitution in only certain constructs
166 such as double quotes, backticks, angle brackets, and search pat‐
167 terns.
168 · Shells interpret scripts a little bit at a time. Perl compiles the
170 entire program before executing it (except for "BEGIN" blocks,
171 which execute at compile time).
172 · The arguments are available via @ARGV, not $1, $2, etc.
174 · The environment is not automatically made available as separate
176 scalar variables.
177 · The shell's "test" uses "=", "!=", "<" etc for string comparisons
179 and "-eq", "-ne", "-lt" etc for numeric comparisons. This is the
180 reverse of Perl, which uses "eq", "ne", "lt" for string compar‐
181 isons, and "==", "!=" "<" etc for numeric comparisons.
182 Perl Traps
184 Practicing Perl Programmers should take note of the following:
186 · Remember that many operations behave differently in a list context
188 than they do in a scalar one. See perldata for details.
189 · Avoid barewords if you can, especially all lowercase ones. You
191 can't tell by just looking at it whether a bareword is a function
192 or a string. By using quotes on strings and parentheses on func‐
193 tion calls, you won't ever get them confused.
194 · You cannot discern from mere inspection which builtins are unary
196 operators (like chop() and chdir()) and which are list operators
197 (like print() and unlink()). (Unless prototyped, user-defined sub‐
198 routines can only be list operators, never unary ones.) See perlop
199 and perlsub.
200 · People have a hard time remembering that some functions default to
202 $_, or @ARGV, or whatever, but that others which you might expect
203 to do not.
204 · The <FH> construct is not the name of the filehandle, it is a read‐
206 line operation on that handle. The data read is assigned to $_
207 only if the file read is the sole condition in a while loop:
208 while (<FH>) { }
210 while (defined($_ = <FH>)) { }..
211 <FH>; # data discarded!
212 · Remember not to use "=" when you need "=~"; these two constructs
214 are quite different:
215 $x = /foo/;
217 $x =~ /foo/;
218 · The "do {}" construct isn't a real loop that you can use loop con‐
220 trol on.
221 · Use "my()" for local variables whenever you can get away with it
223 (but see perlform for where you can't). Using "local()" actually
224 gives a local value to a global variable, which leaves you open to
225 unforeseen side-effects of dynamic scoping.
226 · If you localize an exported variable in a module, its exported
228 value will not change. The local name becomes an alias to a new
229 value but the external name is still an alias for the original.
230 Perl4 to Perl5 Traps
232 Practicing Perl4 Programmers should take note of the following
234 Perl4-to-Perl5 specific traps.
235 They're crudely ordered according to the following list:
237 Discontinuance, Deprecation, and BugFix traps
239 Anything that's been fixed as a perl4 bug, removed as a perl4 fea‐
240 ture or deprecated as a perl4 feature with the intent to encourage
241 usage of some other perl5 feature.
242 Parsing Traps
244 Traps that appear to stem from the new parser.
245 Numerical Traps
247 Traps having to do with numerical or mathematical operators.
248 General data type traps
250 Traps involving perl standard data types.
251 Context Traps - scalar, list contexts
253 Traps related to context within lists, scalar statements/declara‐
254 tions.
255 Precedence Traps
257 Traps related to the precedence of parsing, evaluation, and execu‐
258 tion of code.
259 General Regular Expression Traps using s///, etc.
261 Traps related to the use of pattern matching.
262 Subroutine, Signal, Sorting Traps
264 Traps related to the use of signals and signal handlers, general
265 subroutines, and sorting, along with sorting subroutines.
266 OS Traps
268 OS-specific traps.
269 DBM Traps
271 Traps specific to the use of "dbmopen()", and specific dbm imple‐
272 mentations.
273 Unclassified Traps
275 Everything else.
276 If you find an example of a conversion trap that is not listed here,
278 please submit it to <perlbug@perl.org> for inclusion. Also note that
279 at least some of these can be caught with the "use warnings" pragma or
280 the -w switch.
281 Discontinuance, Deprecation, and BugFix traps
283 Anything that has been discontinued, deprecated, or fixed as a bug from
285 perl4.
286 * Symbols starting with "_" no longer forced into main
288 Symbols starting with "_" are no longer forced into package main,
289 except for $_ itself (and @_, etc.).
290 package test;
292 $_legacy = 1;
293 package main;
295 print "\$_legacy is ",$_legacy,"\n";
296 # perl4 prints: $_legacy is 1
298 # perl5 prints: $_legacy is
299 * Double-colon valid package separator in variable name
301 Double-colon is now a valid package separator in a variable name.
302 Thus these behave differently in perl4 vs. perl5, because the pack‐
303 ages don't exist.
304 $a=1;$b=2;$c=3;$var=4;
306 print "$a::$b::$c ";
307 print "$var::abc::xyz\n";
308 # perl4 prints: 1::2::3 4::abc::xyz
310 # perl5 prints: 3
311 Given that "::" is now the preferred package delimiter, it is
313 debatable whether this should be classed as a bug or not. (The
314 older package delimiter, ' ,is used here)
315 $x = 10;
317 print "x=${'x}\n";
318 # perl4 prints: x=10
320 # perl5 prints: Can't find string terminator "'" anywhere before EOF
321 You can avoid this problem, and remain compatible with perl4, if
323 you always explicitly include the package name:
324 $x = 10;
326 print "x=${main'x}\n";
327 Also see precedence traps, for parsing $:.
329 * 2nd and 3rd args to "splice()" are now in scalar context
331 The second and third arguments of "splice()" are now evaluated in
332 scalar context (as the Camel says) rather than list context.
333 sub sub1{return(0,2) } # return a 2-element list
335 sub sub2{ return(1,2,3)} # return a 3-element list
336 @a1 = ("a","b","c","d","e");
337 @a2 = splice(@a1,&sub1,&sub2);
338 print join(' ',@a2),"\n";
339 # perl4 prints: a b
341 # perl5 prints: c d e
342 * Can't do "goto" into a block that is optimized away
344 You can't do a "goto" into a block that is optimized away. Darn.
345 goto marker1;
347 for(1){
349 marker1:
350 print "Here I is!\n";
351 }
352 # perl4 prints: Here I is!
354 # perl5 errors: Can't "goto" into the middle of a foreach loop
355 * Can't use whitespace as variable name or quote delimiter
357 It is no longer syntactically legal to use whitespace as the name
358 of a variable, or as a delimiter for any kind of quote construct.
359 Double darn.
360 $a = ("foo bar");
362 $b = q baz;
363 print "a is $a, b is $b\n";
364 # perl4 prints: a is foo bar, b is baz
366 # perl5 errors: Bareword found where operator expected
367 * "while/if BLOCK BLOCK" gone
369 The archaic while/if BLOCK BLOCK syntax is no longer supported.
370 if { 1 } {
372 print "True!";
373 }
374 else {
375 print "False!";
376 }
377 # perl4 prints: True!
379 # perl5 errors: syntax error at test.pl line 1, near "if {"
380 * "**" binds tighter than unary minus
382 The "**" operator now binds more tightly than unary minus. It was
383 documented to work this way before, but didn't.
384 print -4**2,"\n";
386 # perl4 prints: 16
388 # perl5 prints: -16
389 * "foreach" changed when iterating over a list
391 The meaning of "foreach{}" has changed slightly when it is iterat‐
392 ing over a list which is not an array. This used to assign the
393 list to a temporary array, but no longer does so (for efficiency).
394 This means that you'll now be iterating over the actual values, not
395 over copies of the values. Modifications to the loop variable can
396 change the original values.
397 @list = ('ab','abc','bcd','def');
399 foreach $var (grep(/ab/,@list)){
400 $var = 1;
401 }
402 print (join(':',@list));
403 # perl4 prints: ab:abc:bcd:def
405 # perl5 prints: 1:1:bcd:def
406 To retain Perl4 semantics you need to assign your list explicitly
408 to a temporary array and then iterate over that. For example, you
409 might need to change
410 foreach $var (grep(/ab/,@list)){
412 to
414 foreach $var (@tmp = grep(/ab/,@list)){
416 Otherwise changing $var will clobber the values of @list. (This
418 most often happens when you use $_ for the loop variable, and call
419 subroutines in the loop that don't properly localize $_.)
420 * "split" with no args behavior changed
422 "split" with no arguments now behaves like "split ' '" (which
423 doesn't return an initial null field if $_ starts with whitespace),
424 it used to behave like "split /\s+/" (which does).
425 $_ = ' hi mom';
427 print join(':', split);
428 # perl4 prints: :hi:mom
430 # perl5 prints: hi:mom
431 * -e behavior fixed
433 Perl 4 would ignore any text which was attached to an -e switch,
434 always taking the code snippet from the following arg. Addition‐
435 ally, it would silently accept an -e switch without a following
436 arg. Both of these behaviors have been fixed.
437 perl -e'print "attached to -e"' 'print "separate arg"'
439 # perl4 prints: separate arg
441 # perl5 prints: attached to -e
442 perl -e
444 # perl4 prints:
446 # perl5 dies: No code specified for -e.
447 * "push" returns number of elements in resulting list
449 In Perl 4 the return value of "push" was undocumented, but it was
450 actually the last value being pushed onto the target list. In Perl
451 5 the return value of "push" is documented, but has changed, it is
452 the number of elements in the resulting list.
453 @x = ('existing');
455 print push(@x, 'first new', 'second new');
456 # perl4 prints: second new
458 # perl5 prints: 3
459 * Some error messages differ
461 Some error messages will be different.
462 * "split()" honors subroutine args
464 In Perl 4, if in list context the delimiters to the first argument
465 of "split()" were "??", the result would be placed in @_ as well as
466 being returned. Perl 5 has more respect for your subroutine argu‐
467 ments.
468 * Bugs removed
470 Some bugs may have been inadvertently removed. :-)
471 Parsing Traps
473 Perl4-to-Perl5 traps from having to do with parsing.
475 * Space between . and = triggers syntax error
477 Note the space between . and =
478 $string . = "more string";
480 print $string;
481 # perl4 prints: more string
483 # perl5 prints: syntax error at - line 1, near ". ="
484 * Better parsing in perl 5
486 Better parsing in perl 5
487 sub foo {}
489 &foo
490 print("hello, world\n");
491 # perl4 prints: hello, world
493 # perl5 prints: syntax error
494 * Function parsing
496 "if it looks like a function, it is a function" rule.
497 print
499 ($foo == 1) ? "is one\n" : "is zero\n";
500 # perl4 prints: is zero
502 # perl5 warns: "Useless use of a constant in void context" if using -w
503 * String interpolation of $#array differs
505 String interpolation of the $#array construct differs when braces
506 are to used around the name.
507 @a = (1..3);
509 print "${#a}";
510 # perl4 prints: 2
512 # perl5 fails with syntax error
513 @ = (1..3);
515 print "$#{a}";
516 # perl4 prints: {a}
518 # perl5 prints: 2
519 * Perl guesses on "map", "grep" followed by "{" if it starts BLOCK or
521 hash ref
522 When perl sees "map {" (or "grep {"), it has to guess whether the
523 "{" starts a BLOCK or a hash reference. If it guesses wrong, it
524 will report a syntax error near the "}" and the missing (or unex‐
525 pected) comma.
526 Use unary "+" before "{" on a hash reference, and unary "+" applied
528 to the first thing in a BLOCK (after "{"), for perl to guess right
529 all the time. (See "map" in perlfunc.)
530 Numerical Traps
532 Perl4-to-Perl5 traps having to do with numerical operators, operands,
534 or output from same.
535 * Formatted output and significant digits
537 Formatted output and significant digits. In general, Perl 5 tries
538 to be more precise. For example, on a Solaris Sparc:
539 print 7.373504 - 0, "\n";
541 printf "%20.18f\n", 7.373504 - 0;
542 # Perl4 prints:
544 7.3750399999999996141
545 7.375039999999999614
546 # Perl5 prints:
548 7.373504
549 7.375039999999999614
550 Notice how the first result looks better in Perl 5.
552 Your results may vary, since your floating point formatting rou‐
554 tines and even floating point format may be slightly different.
555 * Auto-increment operator over signed int limit deleted
557 This specific item has been deleted. It demonstrated how the
558 auto-increment operator would not catch when a number went over
559 the signed int limit. Fixed in version 5.003_04. But always be
560 wary when using large integers. If in doubt:
561 use Math::BigInt;
563 * Assignment of return values from numeric equality tests doesn't work
565 Assignment of return values from numeric equality tests does not
566 work in perl5 when the test evaluates to false (0). Logical tests
567 now return a null, instead of 0
568 $p = ($test == 1);
570 print $p,"\n";
571 # perl4 prints: 0
573 # perl5 prints:
574 Also see "General Regular Expression Traps using s///, etc." for
576 another example of this new feature...
577 * Bitwise string ops
579 When bitwise operators which can operate upon either numbers or
580 strings ("& ⎪ ^ ~") are given only strings as arguments, perl4
581 would treat the operands as bitstrings so long as the program con‐
582 tained a call to the "vec()" function. perl5 treats the string op‐
583 erands as bitstrings. (See "Bitwise String Operators" in perlop
584 for more details.)
585 $fred = "10";
587 $barney = "12";
588 $betty = $fred & $barney;
589 print "$betty\n";
590 # Uncomment the next line to change perl4's behavior
591 # ($dummy) = vec("dummy", 0, 0);
592 # Perl4 prints:
594 8
595 # Perl5 prints:
597 10
598 # If vec() is used anywhere in the program, both print:
600 10
601 General data type traps
603 Perl4-to-Perl5 traps involving most data-types, and their usage within
605 certain expressions and/or context.
606 * Negative array subscripts now count from the end of array
608 Negative array subscripts now count from the end of the array.
609 @a = (1, 2, 3, 4, 5);
611 print "The third element of the array is $a[3] also expressed as $a[-2] \n";
612 # perl4 prints: The third element of the array is 4 also expressed as
614 # perl5 prints: The third element of the array is 4 also expressed as 4
615 * Setting $#array lower now discards array elements
617 Setting $#array lower now discards array elements, and makes them
618 impossible to recover.
619 @a = (a,b,c,d,e);
621 print "Before: ",join('',@a);
622 $#a =1;
623 print ", After: ",join('',@a);
624 $#a =3;
625 print ", Recovered: ",join('',@a),"\n";
626 # perl4 prints: Before: abcde, After: ab, Recovered: abcd
628 # perl5 prints: Before: abcde, After: ab, Recovered: ab
629 * Hashes get defined before use
631 Hashes get defined before use
632 local($s,@a,%h);
634 die "scalar \$s defined" if defined($s);
635 die "array \@a defined" if defined(@a);
636 die "hash \%h defined" if defined(%h);
637 # perl4 prints:
639 # perl5 dies: hash %h defined
640 Perl will now generate a warning when it sees defined(@a) and
642 defined(%h).
643 * Glob assignment from localized variable to variable
645 glob assignment from variable to variable will fail if the
646 assigned variable is localized subsequent to the assignment
647 @a = ("This is Perl 4");
649 *b = *a;
650 local(@a);
651 print @b,"\n";
652 # perl4 prints: This is Perl 4
654 # perl5 prints:
655 * Assigning "undef" to glob
657 Assigning "undef" to a glob has no effect in Perl 5. In Perl 4
658 it undefines the associated scalar (but may have other side
659 effects including SEGVs). Perl 5 will also warn if "undef" is
660 assigned to a typeglob. (Note that assigning "undef" to a typeglob
661 is different than calling the "undef" function on a typeglob
662 ("undef *foo"), which has quite a few effects.
663 $foo = "bar";
665 *foo = undef;
666 print $foo;
667 # perl4 prints:
669 # perl4 warns: "Use of uninitialized variable" if using -w
670 # perl5 prints: bar
671 # perl5 warns: "Undefined value assigned to typeglob" if using -w
672 * Changes in unary negation (of strings)
674 Changes in unary negation (of strings) This change effects both
675 the return value and what it does to auto(magic)increment.
676 $x = "aaa";
678 print ++$x," : ";
679 print -$x," : ";
680 print ++$x,"\n";
681 # perl4 prints: aab : -0 : 1
683 # perl5 prints: aab : -aab : aac
684 * Modifying of constants prohibited
686 perl 4 lets you modify constants:
687 $foo = "x";
689 &mod($foo);
690 for ($x = 0; $x < 3; $x++) {
691 &mod("a");
692 }
693 sub mod {
694 print "before: $_[0]";
695 $_[0] = "m";
696 print " after: $_[0]\n";
697 }
698 # perl4:
700 # before: x after: m
701 # before: a after: m
702 # before: m after: m
703 # before: m after: m
704 # Perl5:
706 # before: x after: m
707 # Modification of a read-only value attempted at foo.pl line 12.
708 # before: a
709 * "defined $var" behavior changed
711 The behavior is slightly different for:
712 print "$x", defined $x
714 # perl 4: 1
716 # perl 5: <no output, $x is not called into existence>
717 * Variable Suicide
719 Variable suicide behavior is more consistent under Perl 5. Perl5
720 exhibits the same behavior for hashes and scalars, that perl4
721 exhibits for only scalars.
722 $aGlobal{ "aKey" } = "global value";
724 print "MAIN:", $aGlobal{"aKey"}, "\n";
725 $GlobalLevel = 0;
726 &test( *aGlobal );
727 sub test {
729 local( *theArgument ) = @_;
730 local( %aNewLocal ); # perl 4 != 5.001l,m
731 $aNewLocal{"aKey"} = "this should never appear";
732 print "SUB: ", $theArgument{"aKey"}, "\n";
733 $aNewLocal{"aKey"} = "level $GlobalLevel"; # what should print
734 $GlobalLevel++;
735 if( $GlobalLevel<4 ) {
736 &test( *aNewLocal );
737 }
738 }
739 # Perl4:
741 # MAIN:global value
742 # SUB: global value
743 # SUB: level 0
744 # SUB: level 1
745 # SUB: level 2
746 # Perl5:
748 # MAIN:global value
749 # SUB: global value
750 # SUB: this should never appear
751 # SUB: this should never appear
752 # SUB: this should never appear
753 Context Traps - scalar, list contexts
755 * Elements of argument lists for formats evaluated in list context
757 The elements of argument lists for formats are now evaluated in
758 list context. This means you can interpolate list values now.
759 @fmt = ("foo","bar","baz");
761 format STDOUT=
762 @<<<<< @⎪⎪⎪⎪⎪ @>>>>>
763 @fmt;
764 .
765 write;
766 # perl4 errors: Please use commas to separate fields in file
768 # perl5 prints: foo bar baz
769 * "caller()" returns false value in scalar context if no caller present
771 The "caller()" function now returns a false value in a scalar con‐
772 text if there is no caller. This lets library files determine if
773 they're being required.
774 caller() ? (print "You rang?\n") : (print "Got a 0\n");
776 # perl4 errors: There is no caller
778 # perl5 prints: Got a 0
779 * Comma operator in scalar context gives scalar context to args
781 The comma operator in a scalar context is now guaranteed to give a
782 scalar context to its arguments.
783 @y= ('a','b','c');
785 $x = (1, 2, @y);
786 print "x = $x\n";
787 # Perl4 prints: x = c # Thinks list context interpolates list
789 # Perl5 prints: x = 3 # Knows scalar uses length of list
790 * "sprintf()" prototyped as "($;@)"
792 "sprintf()" is prototyped as ($;@), so its first argument is given
793 scalar context. Thus, if passed an array, it will probably not do
794 what you want, unlike Perl 4:
795 @z = ('%s%s', 'foo', 'bar');
797 $x = sprintf(@z);
798 print $x;
799 # perl4 prints: foobar
801 # perl5 prints: 3
802 "printf()" works the same as it did in Perl 4, though:
804 @z = ('%s%s', 'foo', 'bar');
806 printf STDOUT (@z);
807 # perl4 prints: foobar
809 # perl5 prints: foobar
810 Precedence Traps
812 Perl4-to-Perl5 traps involving precedence order.
814 Perl 4 has almost the same precedence rules as Perl 5 for the operators
816 that they both have. Perl 4 however, seems to have had some inconsis‐
817 tencies that made the behavior differ from what was documented.
818 * LHS vs. RHS of any assignment operator
820 LHS vs. RHS of any assignment operator. LHS is evaluated first in
821 perl4, second in perl5; this can affect the relationship between
822 side-effects in sub-expressions.
823 @arr = ( 'left', 'right' );
825 $a{shift @arr} = shift @arr;
826 print join( ' ', keys %a );
827 # perl4 prints: left
829 # perl5 prints: right
830 * Semantic errors introduced due to precedence
832 These are now semantic errors because of precedence:
833 @list = (1,2,3,4,5);
835 %map = ("a",1,"b",2,"c",3,"d",4);
836 $n = shift @list + 2; # first item in list plus 2
837 print "n is $n, ";
838 $m = keys %map + 2; # number of items in hash plus 2
839 print "m is $m\n";
840 # perl4 prints: n is 3, m is 6
842 # perl5 errors and fails to compile
843 * Precedence of assignment operators same as the precedence of assign‐
845 ment
846 The precedence of assignment operators is now the same as the
847 precedence of assignment. Perl 4 mistakenly gave them the prece‐
848 dence of the associated operator. So you now must parenthesize
849 them in expressions like
850 /foo/ ? ($a += 2) : ($a -= 2);
852 Otherwise
854 /foo/ ? $a += 2 : $a -= 2
856 would be erroneously parsed as
858 (/foo/ ? $a += 2 : $a) -= 2;
860 On the other hand,
862 $a += /foo/ ? 1 : 2;
864 now works as a C programmer would expect.
866 * "open" requires parentheses around filehandle
868 open FOO ⎪⎪ die;
869 is now incorrect. You need parentheses around the filehandle.
871 Otherwise, perl5 leaves the statement as its default precedence:
872 open(FOO ⎪⎪ die);
874 # perl4 opens or dies
876 # perl5 opens FOO, dying only if 'FOO' is false, i.e. never
877 * $: precedence over $:: gone
879 perl4 gives the special variable, $: precedence, where perl5
880 treats $:: as main "package"
881 $a = "x"; print "$::a";
883 # perl 4 prints: -:a
885 # perl 5 prints: x
886 * Precedence of file test operators documented
888 perl4 had buggy precedence for the file test operators vis-a-vis
889 the assignment operators. Thus, although the precedence table for
890 perl4 leads one to believe "-e $foo .= "q"" should parse as "((-e
891 $foo) .= "q")", it actually parses as "(-e ($foo .= "q"))". In
892 perl5, the precedence is as documented.
893 -e $foo .= "q"
895 # perl4 prints: no output
897 # perl5 prints: Can't modify -e in concatenation
898 * "keys", "each", "values" are regular named unary operators
900 In perl4, keys(), each() and values() were special high-precedence
901 operators that operated on a single hash, but in perl5, they are
902 regular named unary operators. As documented, named unary opera‐
903 tors have lower precedence than the arithmetic and concatenation
904 operators "+ - .", but the perl4 variants of these operators actu‐
905 ally bind tighter than "+ - .". Thus, for:
906 %foo = 1..10;
908 print keys %foo - 1
909 # perl4 prints: 4
911 # perl5 prints: Type of arg 1 to keys must be hash (not subtraction)
912 The perl4 behavior was probably more useful, if less consistent.
914 General Regular Expression Traps using s///, etc.
916 All types of RE traps.
918 * "s'$lhs'$rhs'" interpolates on either side
920 "s'$lhs'$rhs'" now does no interpolation on either side. It used
921 to interpolate $lhs but not $rhs. (And still does not match a
922 literal '$' in string)
923 $a=1;$b=2;
925 $string = '1 2 $a $b';
926 $string =~ s'$a'$b';
927 print $string,"\n";
928 # perl4 prints: $b 2 $a $b
930 # perl5 prints: 1 2 $a $b
931 * "m//g" attaches its state to the searched string
933 "m//g" now attaches its state to the searched string rather than
934 the regular expression. (Once the scope of a block is left for
935 the sub, the state of the searched string is lost)
936 $_ = "ababab";
938 while(m/ab/g){
939 &doit("blah");
940 }
941 sub doit{local($_) = shift; print "Got $_ "}
942 # perl4 prints: Got blah Got blah Got blah Got blah
944 # perl5 prints: infinite loop blah...
945 * "m//o" used within an anonymous sub
947 Currently, if you use the "m//o" qualifier on a regular expression
948 within an anonymous sub, all closures generated from that anony‐
949 mous sub will use the regular expression as it was compiled when
950 it was used the very first time in any such closure. For
951 instance, if you say
952 sub build_match {
954 my($left,$right) = @_;
955 return sub { $_[0] =~ /$left stuff $right/o; };
956 }
957 $good = build_match('foo','bar');
958 $bad = build_match('baz','blarch');
959 print $good->('foo stuff bar') ? "ok\n" : "not ok\n";
960 print $bad->('baz stuff blarch') ? "ok\n" : "not ok\n";
961 print $bad->('foo stuff bar') ? "not ok\n" : "ok\n";
962 For most builds of Perl5, this will print: ok not ok not ok
964 build_match() will always return a sub which matches the contents
966 of $left and $right as they were the first time that build_match()
967 was called, not as they are in the current call.
968 * $+ isn't set to whole match
970 If no parentheses are used in a match, Perl4 sets $+ to the whole
971 match, just like $&. Perl5 does not.
972 "abcdef" =~ /b.*e/;
974 print "\$+ = $+\n";
975 # perl4 prints: bcde
977 # perl5 prints:
978 * Substitution now returns null string if it fails
980 substitution now returns the null string if it fails
981 $string = "test";
983 $value = ($string =~ s/foo//);
984 print $value, "\n";
985 # perl4 prints: 0
987 # perl5 prints:
988 Also see "Numerical Traps" for another example of this new fea‐
990 ture.
991 * "s`lhs`rhs`" is now a normal substitution
993 "s`lhs`rhs`" (using backticks) is now a normal substitution, with
994 no backtick expansion
995 $string = "";
997 $string =~ s`^`hostname`;
998 print $string, "\n";
999 # perl4 prints: <the local hostname>
1001 # perl5 prints: hostname
1002 * Stricter parsing of variables in regular expressions
1004 Stricter parsing of variables used in regular expressions
1005 s/^([^$grpc]*$grpc[$opt$plus$rep]?)//o;
1007 # perl4: compiles w/o error
1009 # perl5: with Scalar found where operator expected ..., near "$opt$plus"
1010 an added component of this example, apparently from the same
1012 script, is the actual value of the s'd string after the substitu‐
1013 tion. "[$opt]" is a character class in perl4 and an array sub‐
1014 script in perl5
1015 $grpc = 'a';
1017 $opt = 'r';
1018 $_ = 'bar';
1019 s/^([^$grpc]*$grpc[$opt]?)/foo/;
1020 print;
1021 # perl4 prints: foo
1023 # perl5 prints: foobar
1024 * "m?x?" matches only once
1026 Under perl5, "m?x?" matches only once, like "?x?". Under perl4, it
1027 matched repeatedly, like "/x/" or "m!x!".
1028 $test = "once";
1030 sub match { $test =~ m?once?; }
1031 &match();
1032 if( &match() ) {
1033 # m?x? matches more then once
1034 print "perl4\n";
1035 } else {
1036 # m?x? matches only once
1037 print "perl5\n";
1038 }
1039 # perl4 prints: perl4
1041 # perl5 prints: perl5
1042 * Failed matches don't reset the match variables
1044 Unlike in Ruby, failed matches in Perl do not reset the match
1045 variables ($1, $2, ..., $`, ...).
1046 Subroutine, Signal, Sorting Traps
1048 The general group of Perl4-to-Perl5 traps having to do with Signals,
1050 Sorting, and their related subroutines, as well as general subroutine
1051 traps. Includes some OS-Specific traps.
1052 * Barewords that used to look like strings look like subroutine calls
1054 Barewords that used to look like strings to Perl will now look
1055 like subroutine calls if a subroutine by that name is defined
1056 before the compiler sees them.
1057 sub SeeYa { warn"Hasta la vista, baby!" }
1059 $SIG{'TERM'} = SeeYa;
1060 print "SIGTERM is now $SIG{'TERM'}\n";
1061 # perl4 prints: SIGTERM is now main'SeeYa
1063 # perl5 prints: SIGTERM is now main::1 (and warns "Hasta la vista, baby!")
1064 Use -w to catch this one
1066 * Reverse is no longer allowed as the name of a sort subroutine
1068 reverse is no longer allowed as the name of a sort subroutine.
1069 sub reverse{ print "yup "; $a <=> $b }
1071 print sort reverse (2,1,3);
1072 # perl4 prints: yup yup 123
1074 # perl5 prints: 123
1075 # perl5 warns (if using -w): Ambiguous call resolved as CORE::reverse()
1076 * "warn()" won't let you specify a filehandle.
1078 Although it _always_ printed to STDERR, warn() would let you spec‐
1079 ify a filehandle in perl4. With perl5 it does not.
1080 warn STDERR "Foo!";
1082 # perl4 prints: Foo!
1084 # perl5 prints: String found where operator expected
1085 OS Traps
1087 * SysV resets signal handler correctly
1089 Under HPUX, and some other SysV OSes, one had to reset any signal
1090 handler, within the signal handler function, each time a signal
1091 was handled with perl4. With perl5, the reset is now done cor‐
1092 rectly. Any code relying on the handler _not_ being reset will
1093 have to be reworked.
1094 Since version 5.002, Perl uses sigaction() under SysV.
1096 sub gotit {
1098 print "Got @_... ";
1099 }
1100 $SIG{'INT'} = 'gotit';
1101 $⎪ = 1;
1103 $pid = fork;
1104 if ($pid) {
1105 kill('INT', $pid);
1106 sleep(1);
1107 kill('INT', $pid);
1108 } else {
1109 while (1) {sleep(10);}
1110 }
1111 # perl4 (HPUX) prints: Got INT...
1113 # perl5 (HPUX) prints: Got INT... Got INT...
1114 * SysV "seek()" appends correctly
1116 Under SysV OSes, "seek()" on a file opened to append ">>" now does
1117 the right thing w.r.t. the fopen() manpage. e.g., - When a file is
1118 opened for append, it is impossible to overwrite information
1119 already in the file.
1120 open(TEST,">>seek.test");
1122 $start = tell TEST;
1123 foreach(1 .. 9){
1124 print TEST "$_ ";
1125 }
1126 $end = tell TEST;
1127 seek(TEST,$start,0);
1128 print TEST "18 characters here";
1129 # perl4 (solaris) seek.test has: 18 characters here
1131 # perl5 (solaris) seek.test has: 1 2 3 4 5 6 7 8 9 18 characters here
1132 Interpolation Traps
1134 Perl4-to-Perl5 traps having to do with how things get interpolated
1136 within certain expressions, statements, contexts, or whatever.
1137 * "@" always interpolates an array in double-quotish strings
1139 @ now always interpolates an array in double-quotish strings.
1140 print "To: someone@somewhere.com\n";
1142 # perl4 prints: To:someone@somewhere.com
1144 # perl < 5.6.1, error : In string, @somewhere now must be written as \@somewhere
1145 # perl >= 5.6.1, warning : Possible unintended interpolation of @somewhere in string
1146 * Double-quoted strings may no longer end with an unescaped $
1148 Double-quoted strings may no longer end with an unescaped $.
1149 $foo = "foo$";
1151 print "foo is $foo\n";
1152 # perl4 prints: foo is foo$
1154 # perl5 errors: Final $ should be \$ or $name
1155 Note: perl5 DOES NOT error on the terminating @ in $bar
1157 * Arbitrary expressions are evaluated inside braces within double
1159 quotes
1160 Perl now sometimes evaluates arbitrary expressions inside braces
1161 that occur within double quotes (usually when the opening brace is
1162 preceded by "$" or "@").
1163 @www = "buz";
1165 $foo = "foo";
1166 $bar = "bar";
1167 sub foo { return "bar" };
1168 print "⎪@{w.w.w}⎪${main'foo}⎪";
1169 # perl4 prints: ⎪@{w.w.w}⎪foo⎪
1171 # perl5 prints: ⎪buz⎪bar⎪
1172 Note that you can "use strict;" to ward off such trappiness under
1174 perl5.
1175 * $$x now tries to dereference $x
1177 The construct "this is $$x" used to interpolate the pid at that
1178 point, but now tries to dereference $x. $$ by itself still works
1179 fine, however.
1180 $s = "a reference";
1182 $x = *s;
1183 print "this is $$x\n";
1184 # perl4 prints: this is XXXx (XXX is the current pid)
1186 # perl5 prints: this is a reference
1187 * Creation of hashes on the fly with "eval "EXPR"" requires protection
1189 Creation of hashes on the fly with "eval "EXPR"" now requires
1190 either both "$"'s to be protected in the specification of the hash
1191 name, or both curlies to be protected. If both curlies are pro‐
1192 tected, the result will be compatible with perl4 and perl5. This
1193 is a very common practice, and should be changed to use the block
1194 form of "eval{}" if possible.
1195 $hashname = "foobar";
1197 $key = "baz";
1198 $value = 1234;
1199 eval "\$$hashname{'$key'} = q⎪$value⎪";
1200 (defined($foobar{'baz'})) ? (print "Yup") : (print "Nope");
1201 # perl4 prints: Yup
1203 # perl5 prints: Nope
1204 Changing
1206 eval "\$$hashname{'$key'} = q⎪$value⎪";
1208 to
1210 eval "\$\$hashname{'$key'} = q⎪$value⎪";
1212 causes the following result:
1214 # perl4 prints: Nope
1216 # perl5 prints: Yup
1217 or, changing to
1219 eval "\$$hashname\{'$key'\} = q⎪$value⎪";
1221 causes the following result:
1223 # perl4 prints: Yup
1225 # perl5 prints: Yup
1226 # and is compatible for both versions
1227 * Bugs in earlier perl versions
1229 perl4 programs which unconsciously rely on the bugs in earlier
1230 perl versions.
1231 perl -e '$bar=q/not/; print "This is $foo{$bar} perl5"'
1233 # perl4 prints: This is not perl5
1235 # perl5 prints: This is perl5
1236 * Array and hash brackets during interpolation
1238 You also have to be careful about array and hash brackets during
1239 interpolation.
1240 print "$foo["
1242 perl 4 prints: [
1244 perl 5 prints: syntax error
1245 print "$foo{"
1247 perl 4 prints: {
1249 perl 5 prints: syntax error
1250 Perl 5 is expecting to find an index or key name following the
1252 respective brackets, as well as an ending bracket of the appropri‐
1253 ate type. In order to mimic the behavior of Perl 4, you must
1254 escape the bracket like so.
1255 print "$foo\[";
1257 print "$foo\{";
1258 * Interpolation of "\$$foo{bar}"
1260 Similarly, watch out for: "\$$foo{bar}"
1261 $foo = "baz";
1263 print "\$$foo{bar}\n";
1264 # perl4 prints: $baz{bar}
1266 # perl5 prints: $
1267 Perl 5 is looking for $foo{bar} which doesn't exist, but perl 4 is
1269 happy just to expand $foo to "baz" by itself. Watch out for this
1270 especially in "eval"'s.
1271 * "qq()" string passed to "eval" will not find string terminator
1273 "qq()" string passed to "eval"
1274 eval qq(
1276 foreach \$y (keys %\$x\) {
1277 \$count++;
1278 }
1279 );
1280 # perl4 runs this ok
1282 # perl5 prints: Can't find string terminator ")"
1283 DBM Traps
1285 General DBM traps.
1287 * Perl5 must have been linked with same dbm/ndbm as the default for
1289 "dbmopen()"
1290 Existing dbm databases created under perl4 (or any other dbm/ndbm
1291 tool) may cause the same script, run under perl5, to fail. The
1292 build of perl5 must have been linked with the same dbm/ndbm as the
1293 default for "dbmopen()" to function properly without "tie"'ing to
1294 an extension dbm implementation.
1295 dbmopen (%dbm, "file", undef);
1297 print "ok\n";
1298 # perl4 prints: ok
1300 # perl5 prints: ok (IFF linked with -ldbm or -lndbm)
1301 * DBM exceeding limit on the key/value size will cause perl5 to exit
1303 immediately
1304 Existing dbm databases created under perl4 (or any other dbm/ndbm
1305 tool) may cause the same script, run under perl5, to fail. The
1306 error generated when exceeding the limit on the key/value size
1307 will cause perl5 to exit immediately.
1308 dbmopen(DB, "testdb",0600) ⎪⎪ die "couldn't open db! $!";
1310 $DB{'trap'} = "x" x 1024; # value too large for most dbm/ndbm
1311 print "YUP\n";
1312 # perl4 prints:
1314 dbm store returned -1, errno 28, key "trap" at - line 3.
1315 YUP
1316 # perl5 prints:
1318 dbm store returned -1, errno 28, key "trap" at - line 3.
1319 Unclassified Traps
1321 Everything else.
1323 * "require"/"do" trap using returned value
1325 If the file doit.pl has:
1326 sub foo {
1328 $rc = do "./do.pl";
1329 return 8;
1330 }
1331 print &foo, "\n";
1332 And the do.pl file has the following single line:
1334 return 3;
1336 Running doit.pl gives the following:
1338 # perl 4 prints: 3 (aborts the subroutine early)
1340 # perl 5 prints: 8
1341 Same behavior if you replace "do" with "require".
1343 * "split" on empty string with LIMIT specified
1345 $string = '';
1346 @list = split(/foo/, $string, 2)
1347 Perl4 returns a one element list containing the empty string but
1349 Perl5 returns an empty list.
1350 As always, if any of these are ever officially declared as bugs,
1352 they'll be fixed and removed.
1353perl v5.8.8 2006-01-07 PERLTRAP(1)