Commit | Line | Data |
68dc0745 |
1 | =head1 NAME |
2 | |
ed8cf1fe |
3 | perlfaq3 - Programming Tools ($Revision: 1.21 $, $Date: 2002/04/28 15:51:10 $) |
68dc0745 |
4 | |
5 | =head1 DESCRIPTION |
6 | |
7 | This section of the FAQ answers questions related to programmer tools |
8 | and programming support. |
9 | |
10 | =head2 How do I do (anything)? |
11 | |
12 | Have you looked at CPAN (see L<perlfaq2>)? The chances are that |
13 | someone has already written a module that can solve your problem. |
3958b146 |
14 | Have you read the appropriate manpages? Here's a brief index: |
68dc0745 |
15 | |
5a964f20 |
16 | Basics perldata, perlvar, perlsyn, perlop, perlsub |
17 | Execution perlrun, perldebug |
18 | Functions perlfunc |
68dc0745 |
19 | Objects perlref, perlmod, perlobj, perltie |
20 | Data Structures perlref, perllol, perldsc |
f102b883 |
21 | Modules perlmod, perlmodlib, perlsub |
d92eb7b0 |
22 | Regexes perlre, perlfunc, perlop, perllocale |
68dc0745 |
23 | Moving to perl5 perltrap, perl |
24 | Linking w/C perlxstut, perlxs, perlcall, perlguts, perlembed |
06a5f41f |
25 | Various http://www.cpan.org/misc/olddoc/FMTEYEWTK.tgz |
26 | (not a man-page but still useful, a collection |
27 | of various essays on Perl techniques) |
68dc0745 |
28 | |
3958b146 |
29 | A crude table of contents for the Perl manpage set is found in L<perltoc>. |
68dc0745 |
30 | |
31 | =head2 How can I use Perl interactively? |
32 | |
33 | The typical approach uses the Perl debugger, described in the |
3958b146 |
34 | perldebug(1) manpage, on an ``empty'' program, like this: |
68dc0745 |
35 | |
36 | perl -de 42 |
37 | |
38 | Now just type in any legal Perl code, and it will be immediately |
39 | evaluated. You can also examine the symbol table, get stack |
40 | backtraces, check variable values, set breakpoints, and other |
92c2ed05 |
41 | operations typically found in symbolic debuggers. |
68dc0745 |
42 | |
43 | =head2 Is there a Perl shell? |
44 | |
55e174a4 |
45 | In general, not yet. There is psh available at |
46 | |
47 | http://www.focusresearch.com/gregor/psh |
48 | |
49 | Which includes the following description: |
50 | |
51 | The Perl Shell is a shell that combines the interactive nature |
52 | of a Unix shell with the power of Perl. The goal is to eventually |
53 | have a full featured shell that behaves as expected for normal |
54 | shell activity. But, the Perl Shell will use Perl syntax and |
da75cd15 |
55 | functionality for control-flow statements and other things. |
55e174a4 |
56 | |
57 | The Shell.pm module (distributed with Perl) makes Perl try commands |
58 | which aren't part of the Perl language as shell commands. perlsh |
59 | from the source distribution is simplistic and uninteresting, but |
60 | may still be what you want. |
68dc0745 |
61 | |
62 | =head2 How do I debug my Perl programs? |
63 | |
9f1b1f2d |
64 | Have you tried C<use warnings> or used C<-w>? They enable warnings |
a6dd486b |
65 | to detect dubious practices. |
68dc0745 |
66 | |
92c2ed05 |
67 | Have you tried C<use strict>? It prevents you from using symbolic |
68 | references, makes you predeclare any subroutines that you call as bare |
69 | words, and (probably most importantly) forces you to predeclare your |
a6dd486b |
70 | variables with C<my>, C<our>, or C<use vars>. |
68dc0745 |
71 | |
a6dd486b |
72 | Did you check the return values of each and every system call? The operating |
73 | system (and thus Perl) tells you whether they worked, and if not |
92c2ed05 |
74 | why. |
68dc0745 |
75 | |
92c2ed05 |
76 | open(FH, "> /etc/cantwrite") |
77 | or die "Couldn't write to /etc/cantwrite: $!\n"; |
68dc0745 |
78 | |
92c2ed05 |
79 | Did you read L<perltrap>? It's full of gotchas for old and new Perl |
a6dd486b |
80 | programmers and even has sections for those of you who are upgrading |
92c2ed05 |
81 | from languages like I<awk> and I<C>. |
82 | |
83 | Have you tried the Perl debugger, described in L<perldebug>? You can |
84 | step through your program and see what it's doing and thus work out |
85 | why what it's doing isn't what it should be doing. |
68dc0745 |
86 | |
87 | =head2 How do I profile my Perl programs? |
88 | |
e083a89c |
89 | You should get the Devel::DProf module from the standard distribution |
733271b5 |
90 | (or separately on CPAN) and also use Benchmark.pm from the standard |
91 | distribution. The Benchmark module lets you time specific portions of |
92 | your code, while Devel::DProf gives detailed breakdowns of where your |
e083a89c |
93 | code spends its time. |
68dc0745 |
94 | |
92c2ed05 |
95 | Here's a sample use of Benchmark: |
96 | |
97 | use Benchmark; |
98 | |
99 | @junk = `cat /etc/motd`; |
100 | $count = 10_000; |
101 | |
102 | timethese($count, { |
103 | 'map' => sub { my @a = @junk; |
104 | map { s/a/b/ } @a; |
105 | return @a |
106 | }, |
107 | 'for' => sub { my @a = @junk; |
108 | local $_; |
109 | for (@a) { s/a/b/ }; |
110 | return @a }, |
111 | }); |
112 | |
113 | This is what it prints (on one machine--your results will be dependent |
114 | on your hardware, operating system, and the load on your machine): |
115 | |
116 | Benchmark: timing 10000 iterations of for, map... |
117 | for: 4 secs ( 3.97 usr 0.01 sys = 3.98 cpu) |
118 | map: 6 secs ( 4.97 usr 0.00 sys = 4.97 cpu) |
119 | |
65acb1b1 |
120 | Be aware that a good benchmark is very hard to write. It only tests the |
a6dd486b |
121 | data you give it and proves little about the differing complexities |
65acb1b1 |
122 | of contrasting algorithms. |
123 | |
68dc0745 |
124 | =head2 How do I cross-reference my Perl programs? |
125 | |
83ded9ee |
126 | The B::Xref module can be used to generate cross-reference reports |
127 | for Perl programs. |
68dc0745 |
128 | |
c8db1d39 |
129 | perl -MO=Xref[,OPTIONS] scriptname.plx |
68dc0745 |
130 | |
131 | =head2 Is there a pretty-printer (formatter) for Perl? |
132 | |
55e174a4 |
133 | Perltidy is a Perl script which indents and reformats Perl scripts |
134 | to make them easier to read by trying to follow the rules of the |
135 | L<perlstyle>. If you write Perl scripts, or spend much time reading |
136 | them, you will probably find it useful. It is available at |
137 | http://perltidy.sourceforge.net |
138 | |
139 | Of course, if you simply follow the guidelines in L<perlstyle>, |
140 | you shouldn't need to reformat. The habit of formatting your code |
141 | as you write it will help prevent bugs. Your editor can and should |
142 | help you with this. The perl-mode or newer cperl-mode for emacs |
143 | can provide remarkable amounts of help with most (but not all) |
144 | code, and even less programmable editors can provide significant |
145 | assistance. Tom Christiansen and many other VI users swear by |
146 | the following settings in vi and its clones: |
65acb1b1 |
147 | |
148 | set ai sw=4 |
d92eb7b0 |
149 | map! ^O {^M}^[O^T |
65acb1b1 |
150 | |
55e174a4 |
151 | Put that in your F<.exrc> file (replacing the caret characters |
65acb1b1 |
152 | with control characters) and away you go. In insert mode, ^T is |
a6dd486b |
153 | for indenting, ^D is for undenting, and ^O is for blockdenting-- |
55e174a4 |
154 | as it were. A more complete example, with comments, can be found at |
213329dd |
155 | http://www.cpan.org/authors/id/TOMC/scripts/toms.exrc.gz |
92c2ed05 |
156 | |
f05bbc40 |
157 | The a2ps http://www-inf.enst.fr/%7Edemaille/a2ps/black+white.ps does |
06a5f41f |
158 | lots of things related to generating nicely printed output of |
f05bbc40 |
159 | documents, as does enscript at http://people.ssh.fi/mtr/genscript/ . |
65acb1b1 |
160 | |
d92eb7b0 |
161 | =head2 Is there a ctags for Perl? |
68dc0745 |
162 | |
bc06af74 |
163 | Recent versions of ctags do much more than older versions did. |
164 | EXUBERANT CTAGS is available from http://ctags.sourceforge.net/ |
165 | and does a good job of making tags files for perl code. |
166 | |
167 | There is also a simple one at |
a93751fa |
168 | http://www.cpan.org/authors/id/TOMC/scripts/ptags.gz which may do |
bc06af74 |
169 | the trick. It can be easy to hack this into what you want. |
65acb1b1 |
170 | |
171 | =head2 Is there an IDE or Windows Perl Editor? |
172 | |
6641ed39 |
173 | Perl programs are just plain text, so any editor will do. |
174 | |
6641ed39 |
175 | If you're on Unix, you already have an IDE--Unix itself. The UNIX |
176 | philosophy is the philosophy of several small tools that each do one |
177 | thing and do it well. It's like a carpenter's toolbox. |
178 | |
5ca69f12 |
179 | If you want an IDE, check the following: |
68fbfbd7 |
180 | |
181 | =over 4 |
182 | |
68fbfbd7 |
183 | =item Komodo |
184 | |
5ca69f12 |
185 | ActiveState's cross-platform (as of April 2001 Windows and Linux), |
186 | multi-language IDE has Perl support, including a regular expression |
187 | debugger and remote debugging |
f224927c |
188 | ( http://www.ActiveState.com/Products/Komodo/index.html ). (Visual |
5ca69f12 |
189 | Perl, a Visual Studio.NET plug-in is currently (early 2001) in beta |
1577cd80 |
190 | ( http://www.ActiveState.com/Products/VisualPerl/index.html )). |
68fbfbd7 |
191 | |
06e809ab |
192 | =item The Object System |
193 | |
bfeeaf1b |
194 | ( http://www.castlelink.co.uk/object_system/ ) is a Perl web |
06e809ab |
195 | applications development IDE, apparently for any platform |
196 | that runs Perl. |
197 | |
ac1094a1 |
198 | =item Open Perl IDE |
199 | |
200 | ( http://open-perl-ide.sourceforge.net/ ) |
201 | Open Perl IDE is an integrated development environment for writing |
202 | and debugging Perl scripts with ActiveState's ActivePerl distribution |
203 | under Windows 95/98/NT/2000. |
204 | |
5ca69f12 |
205 | =item PerlBuilder |
206 | |
f224927c |
207 | ( http://www.solutionsoft.com/perl.htm ) is an integrated development |
5ca69f12 |
208 | environment for Windows that supports Perl development. |
8782d048 |
209 | |
68fbfbd7 |
210 | =item visiPerl+ |
211 | |
ac1094a1 |
212 | ( http://helpconsulting.net/visiperl/ ) |
213 | From Help Consulting, for Windows. |
68fbfbd7 |
214 | |
29b1171f |
215 | =item OptiPerl |
216 | |
217 | ( http://www.optiperl.com/ ) is a Windows IDE with simulated CGI |
218 | environment, including debugger and syntax highlighting editor. |
219 | |
68fbfbd7 |
220 | =back |
221 | |
06e809ab |
222 | For Windows there's also the |
223 | |
224 | =over 4 |
225 | |
226 | =item CodeMagicCD |
227 | |
228 | ( http://www.codemagiccd.com/ ) Collection of various programming |
229 | tools for Windows: Perl (5.005_03), TclTk, Python, GNU programming |
230 | tools, REBOL, wxWindows toolkit, the MinGW GNU C/C++ compiler, DJGPP |
231 | GNU C/C++ compiler, Cint C interpreter, YaBasic. |
232 | |
233 | =back |
234 | |
5a13f98a |
235 | For editors: if you're on Unix you probably have vi or a vi clone already, |
6641ed39 |
236 | and possibly an emacs too, so you may not need to download anything. |
5a13f98a |
237 | In any emacs the cperl-mode (M-x cperl-mode) gives you perhaps the |
6641ed39 |
238 | best available Perl editing mode in any editor. |
239 | |
cc30d1a7 |
240 | If you are using Windows, you can use any editor that lets |
241 | you work with plain text, such as NotePad or WordPad. Word |
242 | processors, such as Microsoft Word or WordPerfect, typically |
243 | do not work since they insert all sorts of behind-the-scenes |
244 | information, although some allow you to save files as "Text |
245 | Only". You can also download text editors designed |
246 | specifically for programming, such as Textpad |
f224927c |
247 | ( http://www.textpad.com/ ) and UltraEdit |
bfeeaf1b |
248 | ( http://www.ultraedit.com/ ), among others. |
cc30d1a7 |
249 | |
250 | If you are using MacOS, the same concerns apply. MacPerl |
251 | (for Classic environments) comes with a simple editor. |
bfeeaf1b |
252 | Popular external editors are BBEdit ( http://www.bbedit.com/ ) |
877ae92e |
253 | or Alpha ( http://www.kelehers.org/alpha/ ). MacOS X users can |
254 | use Unix editors as well. |
68fbfbd7 |
255 | |
256 | =over 4 |
257 | |
258 | =item GNU Emacs |
259 | |
260 | http://www.gnu.org/software/emacs/windows/ntemacs.html |
261 | |
262 | =item MicroEMACS |
263 | |
264 | http://members.nbci.com/uemacs/ |
265 | |
266 | =item XEmacs |
267 | |
268 | http://www.xemacs.org/Download/index.html |
269 | |
270 | =back |
271 | |
272 | or a vi clone such as |
273 | |
274 | =over 4 |
275 | |
276 | =item Elvis |
277 | |
278 | ftp://ftp.cs.pdx.edu/pub/elvis/ http://www.fh-wedel.de/elvis/ |
279 | |
280 | =item Vile |
281 | |
282 | http://vile.cx/ |
283 | |
284 | =item Vim |
285 | |
286 | http://www.vim.org/ |
287 | |
288 | win32: http://www.cs.vu.nl/%7Etmgil/vi.html |
289 | |
290 | =back |
291 | |
5a13f98a |
292 | For vi lovers in general, Windows or elsewhere: |
f05bbc40 |
293 | |
294 | http://www.thomer.com/thomer/vi/vi.html |
6641ed39 |
295 | |
f224927c |
296 | nvi ( http://www.bostic.com/vi/ , available from CPAN in src/misc/) is |
5a13f98a |
297 | yet another vi clone, unfortunately not available for Windows, but in |
6641ed39 |
298 | UNIX platforms you might be interested in trying it out, firstly because |
299 | strictly speaking it is not a vi clone, it is the real vi, or the new |
300 | incarnation of it, and secondly because you can embed Perl inside it |
301 | to use Perl as the scripting language. nvi is not alone in this, |
7c82de66 |
302 | though: at least also vim and vile offer an embedded Perl. |
614a1598 |
303 | |
68fbfbd7 |
304 | The following are Win32 multilanguage editor/IDESs that support Perl: |
305 | |
306 | =over 4 |
307 | |
308 | =item Codewright |
309 | |
310 | http://www.starbase.com/ |
311 | |
312 | =item MultiEdit |
313 | |
314 | http://www.MultiEdit.com/ |
315 | |
316 | =item SlickEdit |
317 | |
318 | http://www.slickedit.com/ |
319 | |
320 | =back |
8782d048 |
321 | |
6641ed39 |
322 | There is also a toyedit Text widget based editor written in Perl |
323 | that is distributed with the Tk module on CPAN. The ptkdb |
f224927c |
324 | ( http://world.std.com/~aep/ptkdb/ ) is a Perl/tk based debugger that |
8782d048 |
325 | acts as a development environment of sorts. Perl Composer |
f224927c |
326 | ( http://perlcomposer.sourceforge.net/vperl.html ) is an IDE for Perl/Tk |
e083a89c |
327 | GUI creation. |
328 | |
8782d048 |
329 | In addition to an editor/IDE you might be interested in a more |
68fbfbd7 |
330 | powerful shell environment for Win32. Your options include |
331 | |
332 | =over 4 |
333 | |
334 | =item Bash |
335 | |
1577cd80 |
336 | from the Cygwin package ( http://sources.redhat.com/cygwin/ ) |
68fbfbd7 |
337 | |
338 | =item Ksh |
339 | |
f224927c |
340 | from the MKS Toolkit ( http://www.mks.com/ ), or the Bourne shell of |
1577cd80 |
341 | the U/WIN environment ( http://www.research.att.com/sw/tools/uwin/ ) |
68fbfbd7 |
342 | |
343 | =item Tcsh |
344 | |
f224927c |
345 | ftp://ftp.astron.com/pub/tcsh/ , see also |
68fbfbd7 |
346 | http://www.primate.wisc.edu/software/csh-tcsh-book/ |
347 | |
348 | =item Zsh |
349 | |
f224927c |
350 | ftp://ftp.blarg.net/users/amol/zsh/ , see also http://www.zsh.org/ |
68fbfbd7 |
351 | |
352 | =back |
353 | |
614a1598 |
354 | MKS and U/WIN are commercial (U/WIN is free for educational and |
355 | research purposes), Cygwin is covered by the GNU Public License (but |
356 | that shouldn't matter for Perl use). The Cygwin, MKS, and U/WIN all |
357 | contain (in addition to the shells) a comprehensive set of standard |
358 | UNIX toolkit utilities. |
8782d048 |
359 | |
5a13f98a |
360 | If you're transferring text files between Unix and Windows using FTP |
361 | be sure to transfer them in ASCII mode so the ends of lines are |
362 | appropriately converted. |
363 | |
e083a89c |
364 | On Mac OS the MacPerl Application comes with a simple 32k text editor |
365 | that behaves like a rudimentary IDE. In contrast to the MacPerl Application |
733271b5 |
366 | the MPW Perl tool can make use of the MPW Shell itself as an editor (with |
68fbfbd7 |
367 | no 32k limit). |
368 | |
369 | =over 4 |
370 | |
371 | =item BBEdit and BBEdit Lite |
372 | |
373 | are text editors for Mac OS that have a Perl sensitivity mode |
1577cd80 |
374 | ( http://web.barebones.com/ ). |
68fbfbd7 |
375 | |
376 | =item Alpha |
377 | |
378 | is an editor, written and extensible in Tcl, that nonetheless has |
733271b5 |
379 | built in support for several popular markup and programming languages |
1577cd80 |
380 | including Perl and HTML ( http://alpha.olm.net/ ). |
68fbfbd7 |
381 | |
382 | =back |
383 | |
384 | Pepper and Pe are programming language sensitive text editors for Mac |
1577cd80 |
385 | OS X and BeOS respectively ( http://www.hekkelman.com/ ). |
68dc0745 |
386 | |
387 | =head2 Where can I get Perl macros for vi? |
388 | |
389 | For a complete version of Tom Christiansen's vi configuration file, |
a93751fa |
390 | see http://www.cpan.org/authors/Tom_Christiansen/scripts/toms.exrc.gz , |
a6dd486b |
391 | the standard benchmark file for vi emulators. The file runs best with nvi, |
5a964f20 |
392 | the current version of vi out of Berkeley, which incidentally can be built |
bfeeaf1b |
393 | with an embedded Perl interpreter--see http://www.cpan.org/src/misc/ . |
68dc0745 |
394 | |
395 | =head2 Where can I get perl-mode for emacs? |
396 | |
397 | Since Emacs version 19 patchlevel 22 or so, there have been both a |
87275199 |
398 | perl-mode.el and support for the Perl debugger built in. These should |
68dc0745 |
399 | come with the standard Emacs 19 distribution. |
400 | |
87275199 |
401 | In the Perl source directory, you'll find a directory called "emacs", |
68dc0745 |
402 | which contains a cperl-mode that color-codes keywords, provides |
403 | context-sensitive help, and other nifty things. |
404 | |
92c2ed05 |
405 | Note that the perl-mode of emacs will have fits with C<"main'foo"> |
d92eb7b0 |
406 | (single quote), and mess up the indentation and highlighting. You |
65acb1b1 |
407 | are probably using C<"main::foo"> in new Perl code anyway, so this |
92c2ed05 |
408 | shouldn't be an issue. |
68dc0745 |
409 | |
410 | =head2 How can I use curses with Perl? |
411 | |
412 | The Curses module from CPAN provides a dynamically loadable object |
5a964f20 |
413 | module interface to a curses library. A small demo can be found at the |
1577cd80 |
414 | directory http://www.cpan.org/authors/Tom_Christiansen/scripts/rep ; |
5a964f20 |
415 | this program repeats a command and updates the screen as needed, rendering |
416 | B<rep ps axu> similar to B<top>. |
68dc0745 |
417 | |
418 | =head2 How can I use X or Tk with Perl? |
419 | |
5a964f20 |
420 | Tk is a completely Perl-based, object-oriented interface to the Tk toolkit |
421 | that doesn't force you to use Tcl just to get at Tk. Sx is an interface |
422 | to the Athena Widget set. Both are available from CPAN. See the |
a93751fa |
423 | directory http://www.cpan.org/modules/by-category/08_User_Interfaces/ |
68dc0745 |
424 | |
a6dd486b |
425 | Invaluable for Perl/Tk programming are the Perl/Tk FAQ at |
87275199 |
426 | http://w4.lns.cornell.edu/%7Epvhp/ptk/ptkTOC.html , the Perl/Tk Reference |
92c2ed05 |
427 | Guide available at |
213329dd |
428 | http://www.cpan.org/authors/Stephen_O_Lidie/ , and the |
92c2ed05 |
429 | online manpages at |
87275199 |
430 | http://www-users.cs.umn.edu/%7Eamundson/perl/perltk/toc.html . |
92c2ed05 |
431 | |
68dc0745 |
432 | =head2 How can I generate simple menus without using CGI or Tk? |
433 | |
a93751fa |
434 | The http://www.cpan.org/authors/id/SKUNZ/perlmenu.v4.0.tar.gz |
68dc0745 |
435 | module, which is curses-based, can help with this. |
436 | |
68dc0745 |
437 | =head2 How can I make my Perl program run faster? |
438 | |
92c2ed05 |
439 | The best way to do this is to come up with a better algorithm. This |
b73a15ae |
440 | can often make a dramatic difference. Jon Bentley's book |
92c2ed05 |
441 | ``Programming Pearls'' (that's not a misspelling!) has some good tips |
442 | on optimization, too. Advice on benchmarking boils down to: benchmark |
443 | and profile to make sure you're optimizing the right part, look for |
444 | better algorithms instead of microtuning your code, and when all else |
57b19278 |
445 | fails consider just buying faster hardware. You will probably want to |
446 | read the answer to the earlier question ``How do I profile my Perl programs?'' |
447 | if you haven't done so already. |
68dc0745 |
448 | |
92c2ed05 |
449 | A different approach is to autoload seldom-used Perl code. See the |
68dc0745 |
450 | AutoSplit and AutoLoader modules in the standard distribution for |
451 | that. Or you could locate the bottleneck and think about writing just |
452 | that part in C, the way we used to take bottlenecks in C code and |
a6dd486b |
453 | write them in assembler. Similar to rewriting in C, |
454 | modules that have critical sections can be written in C (for instance, the |
68dc0745 |
455 | PDL module from CPAN). |
456 | |
457 | In some cases, it may be worth it to use the backend compiler to |
458 | produce byte code (saving compilation time) or compile into C, which |
459 | will certainly save compilation time and sometimes a small amount (but |
460 | not much) execution time. See the question about compiling your Perl |
92c2ed05 |
461 | programs for more on the compiler--the wins aren't as obvious as you'd |
462 | hope. |
68dc0745 |
463 | |
92c2ed05 |
464 | If you're currently linking your perl executable to a shared I<libc.so>, |
68dc0745 |
465 | you can often gain a 10-25% performance benefit by rebuilding it to |
466 | link with a static libc.a instead. This will make a bigger perl |
467 | executable, but your Perl programs (and programmers) may thank you for |
468 | it. See the F<INSTALL> file in the source distribution for more |
469 | information. |
470 | |
471 | Unsubstantiated reports allege that Perl interpreters that use sfio |
87275199 |
472 | outperform those that don't (for I/O intensive applications). To try |
68dc0745 |
473 | this, see the F<INSTALL> file in the source distribution, especially |
87275199 |
474 | the ``Selecting File I/O mechanisms'' section. |
68dc0745 |
475 | |
476 | The undump program was an old attempt to speed up your Perl program |
477 | by storing the already-compiled form to disk. This is no longer |
478 | a viable option, as it only worked on a few architectures, and |
479 | wasn't a good solution anyway. |
480 | |
481 | =head2 How can I make my Perl program take less memory? |
482 | |
483 | When it comes to time-space tradeoffs, Perl nearly always prefers to |
484 | throw memory at a problem. Scalars in Perl use more memory than |
65acb1b1 |
485 | strings in C, arrays take more than that, and hashes use even more. While |
68dc0745 |
486 | there's still a lot to be done, recent releases have been addressing |
487 | these issues. For example, as of 5.004, duplicate hash keys are |
488 | shared amongst all hashes using them, so require no reallocation. |
489 | |
490 | In some cases, using substr() or vec() to simulate arrays can be |
491 | highly beneficial. For example, an array of a thousand booleans will |
492 | take at least 20,000 bytes of space, but it can be turned into one |
a6dd486b |
493 | 125-byte bit vector--a considerable memory savings. The standard |
68dc0745 |
494 | Tie::SubstrHash module can also help for certain types of data |
495 | structure. If you're working with specialist data structures |
496 | (matrices, for instance) modules that implement these in C may use |
497 | less memory than equivalent Perl modules. |
498 | |
499 | Another thing to try is learning whether your Perl was compiled with |
54310121 |
500 | the system malloc or with Perl's builtin malloc. Whichever one it |
68dc0745 |
501 | is, try using the other one and see whether this makes a difference. |
502 | Information about malloc is in the F<INSTALL> file in the source |
503 | distribution. You can find out whether you are using perl's malloc by |
504 | typing C<perl -V:usemymalloc>. |
505 | |
24f1ba9b |
506 | Of course, the best way to save memory is to not do anything to waste |
507 | it in the first place. Good programming practices can go a long way |
508 | toward this: |
509 | |
510 | =over 4 |
511 | |
512 | =item * Don't slurp! |
513 | |
514 | Don't read an entire file into memory if you can process it line |
515 | by line. Or more concretely, use a loop like this: |
516 | |
517 | # |
518 | # Good Idea |
519 | # |
520 | while (<FILE>) { |
521 | # ... |
522 | } |
523 | |
524 | instead of this: |
525 | |
526 | # |
527 | # Bad Idea |
528 | # |
529 | @data = <FILE>; |
530 | foreach (@data) { |
531 | # ... |
532 | } |
533 | |
534 | When the files you're processing are small, it doesn't much matter which |
535 | way you do it, but it makes a huge difference when they start getting |
536 | larger. |
537 | |
bc06af74 |
538 | =item * Use map and grep selectively |
539 | |
540 | Remember that both map and grep expect a LIST argument, so doing this: |
541 | |
542 | @wanted = grep {/pattern/} <FILE>; |
543 | |
544 | will cause the entire file to be slurped. For large files, it's better |
545 | to loop: |
546 | |
547 | while (<FILE>) { |
548 | push(@wanted, $_) if /pattern/; |
549 | } |
550 | |
551 | =item * Avoid unnecessary quotes and stringification |
552 | |
553 | Don't quote large strings unless absolutely necessary: |
554 | |
555 | my $copy = "$large_string"; |
556 | |
557 | makes 2 copies of $large_string (one for $copy and another for the |
558 | quotes), whereas |
559 | |
560 | my $copy = $large_string; |
561 | |
562 | only makes one copy. |
563 | |
564 | Ditto for stringifying large arrays: |
565 | |
566 | { |
567 | local $, = "\n"; |
568 | print @big_array; |
569 | } |
570 | |
571 | is much more memory-efficient than either |
572 | |
573 | print join "\n", @big_array; |
574 | |
575 | or |
576 | |
577 | { |
578 | local $" = "\n"; |
579 | print "@big_array"; |
580 | } |
581 | |
582 | |
24f1ba9b |
583 | =item * Pass by reference |
584 | |
585 | Pass arrays and hashes by reference, not by value. For one thing, it's |
586 | the only way to pass multiple lists or hashes (or both) in a single |
587 | call/return. It also avoids creating a copy of all the contents. This |
588 | requires some judgment, however, because any changes will be propagated |
589 | back to the original data. If you really want to mangle (er, modify) a |
590 | copy, you'll have to sacrifice the memory needed to make one. |
591 | |
592 | =item * Tie large variables to disk. |
593 | |
594 | For "big" data stores (i.e. ones that exceed available memory) consider |
595 | using one of the DB modules to store it on disk instead of in RAM. This |
ed8cf1fe |
596 | will incur a penalty in access time, but that's probably better than |
24f1ba9b |
597 | causing your hard disk to thrash due to massive swapping. |
598 | |
599 | =back |
600 | |
68dc0745 |
601 | =head2 Is it unsafe to return a pointer to local data? |
602 | |
603 | No, Perl's garbage collection system takes care of this. |
604 | |
605 | sub makeone { |
606 | my @a = ( 1 .. 10 ); |
607 | return \@a; |
608 | } |
609 | |
610 | for $i ( 1 .. 10 ) { |
611 | push @many, makeone(); |
612 | } |
613 | |
614 | print $many[4][5], "\n"; |
615 | |
616 | print "@many\n"; |
617 | |
618 | =head2 How can I free an array or hash so my program shrinks? |
619 | |
c8db1d39 |
620 | You can't. On most operating systems, memory allocated to a program |
621 | can never be returned to the system. That's why long-running programs |
65acb1b1 |
622 | sometimes re-exec themselves. Some operating systems (notably, |
623 | FreeBSD and Linux) allegedly reclaim large chunks of memory that is no |
624 | longer used, but it doesn't appear to happen with Perl (yet). The Mac |
625 | appears to be the only platform that will reliably (albeit, slowly) |
626 | return memory to the OS. |
627 | |
628 | We've had reports that on Linux (Redhat 5.1) on Intel, C<undef |
629 | $scalar> will return memory to the system, while on Solaris 2.6 it |
630 | won't. In general, try it yourself and see. |
68dc0745 |
631 | |
632 | However, judicious use of my() on your variables will help make sure |
a6dd486b |
633 | that they go out of scope so that Perl can free up that space for |
92c2ed05 |
634 | use in other parts of your program. A global variable, of course, never |
68dc0745 |
635 | goes out of scope, so you can't get its space automatically reclaimed, |
636 | although undef()ing and/or delete()ing it will achieve the same effect. |
46fc3d4c |
637 | In general, memory allocation and de-allocation isn't something you can |
68dc0745 |
638 | or should be worrying about much in Perl, but even this capability |
639 | (preallocation of data types) is in the works. |
640 | |
641 | =head2 How can I make my CGI script more efficient? |
642 | |
643 | Beyond the normal measures described to make general Perl programs |
644 | faster or smaller, a CGI program has additional issues. It may be run |
645 | several times per second. Given that each time it runs it will need |
46fc3d4c |
646 | to be re-compiled and will often allocate a megabyte or more of system |
68dc0745 |
647 | memory, this can be a killer. Compiling into C B<isn't going to help |
46fc3d4c |
648 | you> because the process start-up overhead is where the bottleneck is. |
68dc0745 |
649 | |
92c2ed05 |
650 | There are two popular ways to avoid this overhead. One solution |
651 | involves running the Apache HTTP server (available from |
f224927c |
652 | http://www.apache.org/ ) with either of the mod_perl or mod_fastcgi |
92c2ed05 |
653 | plugin modules. |
654 | |
655 | With mod_perl and the Apache::Registry module (distributed with |
656 | mod_perl), httpd will run with an embedded Perl interpreter which |
657 | pre-compiles your script and then executes it within the same address |
658 | space without forking. The Apache extension also gives Perl access to |
659 | the internal server API, so modules written in Perl can do just about |
660 | anything a module written in C can. For more on mod_perl, see |
661 | http://perl.apache.org/ |
662 | |
65acb1b1 |
663 | With the FCGI module (from CPAN) and the mod_fastcgi |
bfeeaf1b |
664 | module (available from http://www.fastcgi.com/ ) each of your Perl |
87275199 |
665 | programs becomes a permanent CGI daemon process. |
68dc0745 |
666 | |
667 | Both of these solutions can have far-reaching effects on your system |
87275199 |
668 | and on the way you write your CGI programs, so investigate them with |
68dc0745 |
669 | care. |
670 | |
a93751fa |
671 | See http://www.cpan.org/modules/by-category/15_World_Wide_Web_HTML_HTTP_CGI/ . |
5a964f20 |
672 | |
65acb1b1 |
673 | A non-free, commercial product, ``The Velocity Engine for Perl'', |
a6dd486b |
674 | (http://www.binevolve.com/ or http://www.binevolve.com/velocigen/ ) |
675 | might also be worth looking at. It will allow you to increase the |
676 | performance of your Perl programs, running programs up to 25 times |
677 | faster than normal CGI Perl when running in persistent Perl mode or 4 |
678 | to 5 times faster without any modification to your existing CGI |
679 | programs. Fully functional evaluation copies are available from the |
680 | web site. |
c8db1d39 |
681 | |
68dc0745 |
682 | =head2 How can I hide the source for my Perl program? |
683 | |
684 | Delete it. :-) Seriously, there are a number of (mostly |
92c2ed05 |
685 | unsatisfactory) solutions with varying levels of ``security''. |
68dc0745 |
686 | |
687 | First of all, however, you I<can't> take away read permission, because |
688 | the source code has to be readable in order to be compiled and |
689 | interpreted. (That doesn't mean that a CGI script's source is |
a6dd486b |
690 | readable by people on the web, though--only by people with access to |
691 | the filesystem.) So you have to leave the permissions at the socially |
92c2ed05 |
692 | friendly 0755 level. |
68dc0745 |
693 | |
694 | Some people regard this as a security problem. If your program does |
a6dd486b |
695 | insecure things and relies on people not knowing how to exploit those |
68dc0745 |
696 | insecurities, it is not secure. It is often possible for someone to |
697 | determine the insecure things and exploit them without viewing the |
698 | source. Security through obscurity, the name for hiding your bugs |
699 | instead of fixing them, is little security indeed. |
700 | |
83df6a1d |
701 | You can try using encryption via source filters (Starting from Perl |
702 | 5.8 the Filter::Simple and Filter::Util::Call modules are included in |
703 | the standard distribution), but any decent programmer will be able to |
704 | decrypt it. You can try using the byte code compiler and interpreter |
705 | described below, but the curious might still be able to de-compile it. |
706 | You can try using the native-code compiler described below, but |
707 | crackers might be able to disassemble it. These pose varying degrees |
708 | of difficulty to people wanting to get at your code, but none can |
709 | definitively conceal it (true of every language, not just Perl). |
68dc0745 |
710 | |
711 | If you're concerned about people profiting from your code, then the |
d92eb7b0 |
712 | bottom line is that nothing but a restrictive license will give you |
68dc0745 |
713 | legal security. License your software and pepper it with threatening |
92c2ed05 |
714 | statements like ``This is unpublished proprietary software of XYZ Corp. |
68dc0745 |
715 | Your access to it does not give you permission to use it blah blah |
92c2ed05 |
716 | blah.'' We are not lawyers, of course, so you should see a lawyer if |
d92eb7b0 |
717 | you want to be sure your license's wording will stand up in court. |
68dc0745 |
718 | |
54310121 |
719 | =head2 How can I compile my Perl program into byte code or C? |
68dc0745 |
720 | |
721 | Malcolm Beattie has written a multifunction backend compiler, |
5e3006a4 |
722 | available from CPAN, that can do both these things. It is included |
723 | in the perl5.005 release, but is still considered experimental. |
724 | This means it's fun to play with if you're a programmer but not |
725 | really for people looking for turn-key solutions. |
68dc0745 |
726 | |
92c2ed05 |
727 | Merely compiling into C does not in and of itself guarantee that your |
728 | code will run very much faster. That's because except for lucky cases |
729 | where a lot of native type inferencing is possible, the normal Perl |
a6dd486b |
730 | run-time system is still present and so your program will take just as |
92c2ed05 |
731 | long to run and be just as big. Most programs save little more than |
732 | compilation time, leaving execution no more than 10-30% faster. A few |
a6dd486b |
733 | rare programs actually benefit significantly (even running several times |
92c2ed05 |
734 | faster), but this takes some tweaking of your code. |
68dc0745 |
735 | |
68dc0745 |
736 | You'll probably be astonished to learn that the current version of the |
737 | compiler generates a compiled form of your script whose executable is |
738 | just as big as the original perl executable, and then some. That's |
739 | because as currently written, all programs are prepared for a full |
740 | eval() statement. You can tremendously reduce this cost by building a |
92c2ed05 |
741 | shared I<libperl.so> library and linking against that. See the |
87275199 |
742 | F<INSTALL> podfile in the Perl source distribution for details. If |
d92eb7b0 |
743 | you link your main perl binary with this, it will make it minuscule. |
92c2ed05 |
744 | For example, on one author's system, F</usr/bin/perl> is only 11k in |
68dc0745 |
745 | size! |
746 | |
5a964f20 |
747 | In general, the compiler will do nothing to make a Perl program smaller, |
a6dd486b |
748 | faster, more portable, or more secure. In fact, it can make your |
749 | situation worse. The executable will be bigger, your VM system may take |
5a964f20 |
750 | longer to load the whole thing, the binary is fragile and hard to fix, |
751 | and compilation never stopped software piracy in the form of crackers, |
752 | viruses, or bootleggers. The real advantage of the compiler is merely |
753 | packaging, and once you see the size of what it makes (well, unless |
754 | you use a shared I<libperl.so>), you'll probably want a complete |
5e3006a4 |
755 | Perl install anyway. |
5a964f20 |
756 | |
65acb1b1 |
757 | =head2 How can I compile Perl into Java? |
758 | |
a6dd486b |
759 | You can also integrate Java and Perl with the |
65acb1b1 |
760 | Perl Resource Kit from O'Reilly and Associates. See |
a6dd486b |
761 | http://www.oreilly.com/catalog/prkunix/ . |
762 | |
763 | Perl 5.6 comes with Java Perl Lingo, or JPL. JPL, still in |
764 | development, allows Perl code to be called from Java. See jpl/README |
765 | in the Perl source tree. |
65acb1b1 |
766 | |
92c2ed05 |
767 | =head2 How can I get C<#!perl> to work on [MS-DOS,NT,...]? |
68dc0745 |
768 | |
769 | For OS/2 just use |
770 | |
771 | extproc perl -S -your_switches |
772 | |
773 | as the first line in C<*.cmd> file (C<-S> due to a bug in cmd.exe's |
46fc3d4c |
774 | `extproc' handling). For DOS one should first invent a corresponding |
a6dd486b |
775 | batch file and codify it in C<ALTERNATIVE_SHEBANG> (see the |
68dc0745 |
776 | F<INSTALL> file in the source distribution for more information). |
777 | |
92c2ed05 |
778 | The Win95/NT installation, when using the ActiveState port of Perl, |
779 | will modify the Registry to associate the C<.pl> extension with the |
d92eb7b0 |
780 | perl interpreter. If you install another port, perhaps even building |
781 | your own Win95/NT Perl from the standard sources by using a Windows port |
d702ae42 |
782 | of gcc (e.g., with cygwin or mingw32), then you'll have to modify |
d92eb7b0 |
783 | the Registry yourself. In addition to associating C<.pl> with the |
784 | interpreter, NT people can use: C<SET PATHEXT=%PATHEXT%;.PL> to let them |
785 | run the program C<install-linux.pl> merely by typing C<install-linux>. |
68dc0745 |
786 | |
87275199 |
787 | Macintosh Perl programs will have the appropriate Creator and |
788 | Type, so that double-clicking them will invoke the Perl application. |
68dc0745 |
789 | |
790 | I<IMPORTANT!>: Whatever you do, PLEASE don't get frustrated, and just |
791 | throw the perl interpreter into your cgi-bin directory, in order to |
87275199 |
792 | get your programs working for a web server. This is an EXTREMELY big |
68dc0745 |
793 | security risk. Take the time to figure out how to do it correctly. |
794 | |
87275199 |
795 | =head2 Can I write useful Perl programs on the command line? |
68dc0745 |
796 | |
797 | Yes. Read L<perlrun> for more information. Some examples follow. |
798 | (These assume standard Unix shell quoting rules.) |
799 | |
800 | # sum first and last fields |
5a964f20 |
801 | perl -lane 'print $F[0] + $F[-1]' * |
68dc0745 |
802 | |
803 | # identify text files |
804 | perl -le 'for(@ARGV) {print if -f && -T _}' * |
805 | |
5a964f20 |
806 | # remove (most) comments from C program |
68dc0745 |
807 | perl -0777 -pe 's{/\*.*?\*/}{}gs' foo.c |
808 | |
809 | # make file a month younger than today, defeating reaper daemons |
810 | perl -e '$X=24*60*60; utime(time(),time() + 30 * $X,@ARGV)' * |
811 | |
812 | # find first unused uid |
813 | perl -le '$i++ while getpwuid($i); print $i' |
814 | |
815 | # display reasonable manpath |
816 | echo $PATH | perl -nl -072 -e ' |
817 | s![^/+]*$!man!&&-d&&!$s{$_}++&&push@m,$_;END{print"@m"}' |
818 | |
87275199 |
819 | OK, the last one was actually an Obfuscated Perl Contest entry. :-) |
68dc0745 |
820 | |
87275199 |
821 | =head2 Why don't Perl one-liners work on my DOS/Mac/VMS system? |
68dc0745 |
822 | |
823 | The problem is usually that the command interpreters on those systems |
824 | have rather different ideas about quoting than the Unix shells under |
825 | which the one-liners were created. On some systems, you may have to |
826 | change single-quotes to double ones, which you must I<NOT> do on Unix |
827 | or Plan9 systems. You might also have to change a single % to a %%. |
828 | |
829 | For example: |
830 | |
831 | # Unix |
832 | perl -e 'print "Hello world\n"' |
833 | |
46fc3d4c |
834 | # DOS, etc. |
68dc0745 |
835 | perl -e "print \"Hello world\n\"" |
836 | |
46fc3d4c |
837 | # Mac |
68dc0745 |
838 | print "Hello world\n" |
839 | (then Run "Myscript" or Shift-Command-R) |
840 | |
d2321c93 |
841 | # MPW |
842 | perl -e 'print "Hello world\n"' |
843 | |
68dc0745 |
844 | # VMS |
845 | perl -e "print ""Hello world\n""" |
846 | |
a6dd486b |
847 | The problem is that none of these examples are reliable: they depend on the |
92c2ed05 |
848 | command interpreter. Under Unix, the first two often work. Under DOS, |
a6dd486b |
849 | it's entirely possible that neither works. If 4DOS was the command shell, |
92c2ed05 |
850 | you'd probably have better luck like this: |
68dc0745 |
851 | |
852 | perl -e "print <Ctrl-x>"Hello world\n<Ctrl-x>"" |
853 | |
46fc3d4c |
854 | Under the Mac, it depends which environment you are using. The MacPerl |
68dc0745 |
855 | shell, or MPW, is much like Unix shells in its support for several |
46fc3d4c |
856 | quoting variants, except that it makes free use of the Mac's non-ASCII |
68dc0745 |
857 | characters as control characters. |
858 | |
65acb1b1 |
859 | Using qq(), q(), and qx(), instead of "double quotes", 'single |
860 | quotes', and `backticks`, may make one-liners easier to write. |
861 | |
d2321c93 |
862 | There is no general solution to all of this. It is a mess. |
68dc0745 |
863 | |
864 | [Some of this answer was contributed by Kenneth Albanowski.] |
865 | |
866 | =head2 Where can I learn about CGI or Web programming in Perl? |
867 | |
868 | For modules, get the CGI or LWP modules from CPAN. For textbooks, |
869 | see the two especially dedicated to web stuff in the question on |
92c2ed05 |
870 | books. For problems and questions related to the web, like ``Why |
871 | do I get 500 Errors'' or ``Why doesn't it run from the browser right |
8305e449 |
872 | when it runs fine on the command line'', see the troubleshooting |
873 | guides and references in L<perlfaq9> or in the CGI MetaFAQ: |
68dc0745 |
874 | |
8305e449 |
875 | http://www.perl.org/CGI_MetaFAQ.html |
0f542199 |
876 | |
68dc0745 |
877 | =head2 Where can I learn about object-oriented Perl programming? |
878 | |
a6dd486b |
879 | A good place to start is L<perltoot>, and you can use L<perlobj>, |
06a5f41f |
880 | L<perlboot>, L<perltoot>, L<perltooc>, and L<perlbot> for reference. |
881 | (If you are using really old Perl, you may not have all of these, |
882 | try http://www.perldoc.com/ , but consider upgrading your perl.) |
883 | |
884 | A good book on OO on Perl is the "Object-Oriented Perl" |
885 | by Damian Conway from Manning Publications, |
886 | http://www.manning.com/Conway/index.html |
68dc0745 |
887 | |
888 | =head2 Where can I learn about linking C with Perl? [h2xs, xsubpp] |
889 | |
890 | If you want to call C from Perl, start with L<perlxstut>, |
891 | moving on to L<perlxs>, L<xsubpp>, and L<perlguts>. If you want to |
892 | call Perl from C, then read L<perlembed>, L<perlcall>, and |
893 | L<perlguts>. Don't forget that you can learn a lot from looking at |
894 | how the authors of existing extension modules wrote their code and |
895 | solved their problems. |
896 | |
897 | =head2 I've read perlembed, perlguts, etc., but I can't embed perl in |
a6dd486b |
898 | my C program; what am I doing wrong? |
68dc0745 |
899 | |
900 | Download the ExtUtils::Embed kit from CPAN and run `make test'. If |
901 | the tests pass, read the pods again and again and again. If they |
87275199 |
902 | fail, see L<perlbug> and send a bug report with the output of |
68dc0745 |
903 | C<make test TEST_VERBOSE=1> along with C<perl -V>. |
904 | |
83ded9ee |
905 | =head2 When I tried to run my script, I got this message. What does it mean? |
68dc0745 |
906 | |
87275199 |
907 | A complete list of Perl's error messages and warnings with explanatory |
908 | text can be found in L<perldiag>. You can also use the splain program |
909 | (distributed with Perl) to explain the error messages: |
68dc0745 |
910 | |
911 | perl program 2>diag.out |
912 | splain [-v] [-p] diag.out |
913 | |
914 | or change your program to explain the messages for you: |
915 | |
916 | use diagnostics; |
917 | |
918 | or |
919 | |
920 | use diagnostics -verbose; |
921 | |
922 | =head2 What's MakeMaker? |
923 | |
87275199 |
924 | This module (part of the standard Perl distribution) is designed to |
68dc0745 |
925 | write a Makefile for an extension module from a Makefile.PL. For more |
926 | information, see L<ExtUtils::MakeMaker>. |
927 | |
928 | =head1 AUTHOR AND COPYRIGHT |
929 | |
0bc0ad85 |
930 | Copyright (c) 1997-2002 Tom Christiansen and Nathan Torkington. |
5a964f20 |
931 | All rights reserved. |
932 | |
5a7beb56 |
933 | This documentation is free; you can redistribute it and/or modify it |
934 | under the same terms as Perl itself. |
c8db1d39 |
935 | |
87275199 |
936 | Irrespective of its distribution, all code examples here are in the public |
c8db1d39 |
937 | domain. You are permitted and encouraged to use this code and any |
938 | derivatives thereof in your own programs for fun or for profit as you |
939 | see fit. A simple comment in the code giving credit to the FAQ would |
940 | be courteous but is not required. |