3 perltodo - Perl TO-DO List
7 This is a list of wishes for Perl. The tasks we think are smaller or easier
8 are listed first. Anyone is welcome to work on any of these, but it's a good
9 idea to first contact I<perl5-porters@perl.org> to avoid duplication of
10 effort. By all means contact a pumpking privately first if you prefer.
12 Whilst patches to make the list shorter are most welcome, ideas to add to
13 the list are also encouraged. Check the perl5-porters archives for past
14 ideas, and any discussion about them. One set of archives may be found at:
16 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
18 What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
19 not, but if your patch is incorporated, then we'll add your name to the
20 F<AUTHORS> file, which ships in the official distribution. How many other
21 programming languages offer you 1 line of immortality?
23 =head1 Tasks that only need Perl knowledge
25 =head2 common test code for timed bail out
27 Write portable self destruct code for tests to stop them burning CPU in
28 infinite loops. This needs to avoid using alarm, as some of the tests are
29 testing alarm/sleep or timers.
31 =head2 POD -E<gt> HTML conversion in the core still sucks
33 Which is crazy given just how simple POD purports to be, and how simple HTML
34 can be. It's not actually I<as> simple as it sounds, particularly with the
35 flexibility POD allows for C<=item>, but it would be good to improve the
36 visual appeal of the HTML generated, and to avoid it having any validation
37 errors. See also L</make HTML install work>, as the layout of installation tree
38 is needed to improve the cross-linking.
40 The addition of C<Pod::Simple> and its related modules may make this task
43 =head2 Parallel testing
45 (This probably impacts much more than the core: also the Test::Harness
46 and TAP::* modules on CPAN.)
48 The core regression test suite is getting ever more comprehensive, which has
49 the side effect that it takes longer to run. This isn't so good. Investigate
50 whether it would be feasible to give the harness script the B<option> of
51 running sets of tests in parallel. This would be useful for tests in
52 F<t/op/*.t> and F<t/uni/*.t> and maybe some sets of tests in F<lib/>.
60 How does screen layout work when you're running more than one test?
64 How does the caller of test specify how many tests to run in parallel?
68 How do setup/teardown tests identify themselves?
72 Pugs already does parallel testing - can their approach be re-used?
74 =head2 Make Schwern poorer
76 We should have tests for everything. When all the core's modules are tested,
77 Schwern has promised to donate to $500 to TPF. We may need volunteers to
78 hold him upside down and shake vigorously in order to actually extract the
81 =head2 Improve the coverage of the core tests
83 Use Devel::Cover to ascertain the core modules's test coverage, then add
84 tests that are currently missing.
88 A full test suite for the B module would be nice.
90 =head2 A decent benchmark
92 C<perlbench> seems impervious to any recent changes made to the perl core. It
93 would be useful to have a reasonable general benchmarking suite that roughly
94 represented what current perl programs do, and measurably reported whether
95 tweaks to the core improve, degrade or don't really affect performance, to
96 guide people attempting to optimise the guts of perl. Gisle would welcome
97 new tests for perlbench.
99 =head2 fix tainting bugs
101 Fix the bugs revealed by running the test suite with the C<-t> switch (via
102 C<make test.taintwarn>).
104 =head2 Dual life everything
106 As part of the "dists" plan, anything that doesn't belong in the smallest perl
107 distribution needs to be dual lifed. Anything else can be too. Figure out what
108 changes would be needed to package that module and its tests up for CPAN, and
109 do so. Test it with older perl releases, and fix the problems you find.
111 =head2 Improving C<threads::shared>
113 Investigate whether C<threads::shared> could share aggregates properly with
114 only Perl level changes to shared.pm
116 =head2 POSIX memory footprint
118 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
119 various times worked to cut it down. There is probably still fat to cut out -
120 for example POSIX passes Exporter some very memory hungry data structures.
122 =head2 embed.pl/makedef.pl
124 There is a script F<embed.pl> that generates several header files to prefix
125 all of Perl's symbols in a consistent way, to provide some semblance of
126 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
127 in F<interpvar.h>. Quite a few of the functions and variables
128 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
129 doesn't understand the C macros, so the rules about which symbols are present
130 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
131 It would be good to teach C<embed.pl> to understand the conditional
132 compilation, and hence remove the duplication, and the mistakes it has caused.
134 =head2 use strict; and AutoLoad
136 Currently if you write
139 use AutoLoader 'AUTOLOAD';
144 print join (' ', No, strict, here), "!\n";
147 then C<use strict;> isn't in force within the autoloaded subroutines. It would
148 be more consistent (and less surprising) to arrange for all lexical pragmas
149 in force at the __END__ block to be in force within each autoloaded subroutine.
151 There's a similar problem with SelfLoader.
153 =head1 Tasks that need a little sysadmin-type knowledge
155 Or if you prefer, tasks that you would learn from, and broaden your skills
158 =head2 make HTML install work
160 There is an C<installhtml> target in the Makefile. It's marked as
161 "experimental". It would be good to get this tested, make it work reliably, and
162 remove the "experimental" tag. This would include
168 Checking that cross linking between various parts of the documentation works.
169 In particular that links work between the modules (files with POD in F<lib/>)
170 and the core documentation (files in F<pod/>)
174 Work out how to split C<perlfunc> into chunks, preferably one per function
175 group, preferably with general case code that could be used elsewhere.
176 Challenges here are correctly identifying the groups of functions that go
177 together, and making the right named external cross-links point to the right
178 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
179 C<endservent>, two or more C<=items> giving the different parameter lists, such
182 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
183 =item substr EXPR,OFFSET,LENGTH
184 =item substr EXPR,OFFSET
186 and different parameter lists having different meanings. (eg C<select>)
190 =head2 compressed man pages
192 Be able to install them. This would probably need a configure test to see how
193 the system does compressed man pages (same directory/different directory?
194 same filename/different filename), as well as tweaking the F<installman> script
195 to compress as necessary.
197 =head2 Add a code coverage target to the Makefile
199 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
200 to do this manually are roughly
206 do a normal C<Configure>, but include Devel::Cover as a module to install
207 (see F<INSTALL> for how to do this)
215 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
219 Process the resulting Devel::Cover database
223 This just give you the coverage of the F<.pm>s. To also get the C level
230 Additionally tell C<Configure> to use the appropriate C compiler flags for
237 (instead of C<make perl>)
241 After running the tests run C<gcov> to generate all the F<.gcov> files.
242 (Including down in the subdirectories of F<ext/>
246 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
247 to get their stats into the cover_db directory.
251 Then process the Devel::Cover database
255 It would be good to add a single switch to C<Configure> to specify that you
256 wanted to perform perl level coverage, and another to specify C level
257 coverage, and have C<Configure> and the F<Makefile> do all the right things
260 =head2 Make Config.pm cope with differences between built and installed perl
262 Quite often vendors ship a perl binary compiled with their (pay-for)
263 compilers. People install a free compiler, such as gcc. To work out how to
264 build extensions, Perl interrogates C<%Config>, so in this situation
265 C<%Config> describes compilers that aren't there, and extension building
266 fails. This forces people into choosing between re-compiling perl themselves
267 using the compiler they have, or only using modules that the vendor ships.
269 It would be good to find a way teach C<Config.pm> about the installation setup,
270 possibly involving probing at install time or later, so that the C<%Config> in
271 a binary distribution better describes the installed machine, when the
272 installed machine differs from the build machine in some significant way.
274 =head2 linker specification files
276 Some platforms mandate that you provide a list of a shared library's external
277 symbols to the linker, so the core already has the infrastructure in place to
278 do this for generating shared perl libraries. My understanding is that the
279 GNU toolchain can accept an optional linker specification file, and restrict
280 visibility just to symbols declared in that file. It would be good to extend
281 F<makedef.pl> to support this format, and to provide a means within
282 C<Configure> to enable it. This would allow Unix users to test that the
283 export list is correct, and to build a perl that does not pollute the global
284 namespace with private symbols.
286 =head2 Cross-compile support
288 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
289 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
290 assumed then to be copied to TARGET machine and used as a replacement of full
293 This could be done little differently. Namely C<miniperl> should be built for
294 HOST and then full C<perl> with extensions should be compiled for TARGET.
295 This, however, might require extra trickery for %Config: we have one config
296 first for HOST and then another for TARGET. Tools like MakeMaker will be
297 mightily confused. Having around two different types of executables and
298 libraries (HOST and TARGET) makes life interesting for Makefiles and
299 shell (and Perl) scripts. There is $Config{run}, normally empty, which
300 can be used as an execution wrapper. Also note that in some
301 cross-compilation/execution environments the HOST and the TARGET do
302 not see the same filesystem(s), the $Config{run} may need to do some
303 file/directory copying back and forth.
305 =head1 Tasks that need a little C knowledge
307 These tasks would need a little C knowledge, but don't need any specific
308 background or experience with XS, or how the Perl interpreter works
310 =head2 Make it clear from -v if this is the exact official release
312 Currently perl from C<p4>/C<rsync> ships with a F<patchlevel.h> file that
313 usually defines one local patch, of the form "MAINT12345" or "RC1". The output
314 of perl -v doesn't report that a perl isn't an official release, and this
315 information can get lost in bugs reports. Because of this, the minor version
316 isn't bumped up until RC time, to minimise the possibility of versions of perl
317 escaping that believe themselves to be newer than they actually are.
319 It would be useful to find an elegant way to have the "this is an interim
320 maintenance release" or "this is a release candidate" in the terse -v output,
321 and have it so that it's easy for the pumpking to remove this just as the
322 release tarball is rolled up. This way the version pulled out of rsync would
323 always say "I'm a development release" and it would be safe to bump the
324 reported minor version as soon as a release ships, which would aid perl
327 This task is really about thinking of an elegant way to arrange the C source
328 such that it's trivial for the Pumpking to flag "this is an official release"
329 when making a tarball, yet leave the default source saying "I'm not the
332 =head2 Profile Perl - am I hot or not?
334 The Perl source code is stable enough that it makes sense to profile it,
335 identify and optimise the hotspots. It would be good to measure the
336 performance of the Perl interpreter using free tools such as cachegrind,
337 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
339 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
340 the ops that are most commonly used. The idea is that by grouping them, their
341 object code will be adjacent in the executable, so they have a greater chance
342 of already being in the CPU cache (or swapped in) due to being near another op
345 Except that it's not clear if these really are the most commonly used ops. So
346 as part of exercising your skills with coverage and profiling tools you might
347 want to determine what ops I<really> are the most commonly used. And in turn
348 suggest evictions and promotions to achieve a better F<pp_hot.c>.
350 =head2 Allocate OPs from arenas
352 Currently all new OP structures are individually malloc()ed and free()d.
353 All C<malloc> implementations have space overheads, and are now as fast as
354 custom allocates so it would both use less memory and less CPU to allocate
355 the various OP structures from arenas. The SV arena code can probably be
358 =head2 Improve win32/wince.c
360 Currently, numerous functions look virtually, if not completely,
361 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
364 =head2 Use secure CRT functions when building with VC8 on Win32
366 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
367 that they were "unsafe" and introduced differently named secure versions of
368 them as replacements, e.g. instead of writing
370 FILE* f = fopen(__FILE__, "r");
375 errno_t err = fopen_s(&f, __FILE__, "r");
377 Currently, the warnings about these deprecations have been disabled by adding
378 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
379 warning suppressant and actually make use of the new secure CRT functions.
381 There is also a similar issue with POSIX CRT function names like fileno having
382 been deprecated in favour of ISO C++ conformant names like _fileno. These
383 warnings are also currently suppressed with the compiler option /wd4996. It
384 might be nice to do as Microsoft suggest here too, although, unlike the secure
385 functions issue, there is presumably little or no benefit in this case.
387 =head1 Tasks that need a knowledge of XS
389 These tasks would need C knowledge, and roughly the level of knowledge of
390 the perl API that comes from writing modules that use XS to interface to
393 =head2 autovivification
395 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
397 This task is incremental - even a little bit of work on it will help.
399 =head2 Unicode in Filenames
401 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
402 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
403 system, truncate, unlink, utime, -X. All these could potentially accept
404 Unicode filenames either as input or output (and in the case of system
405 and qx Unicode in general, as input or output to/from the shell).
406 Whether a filesystem - an operating system pair understands Unicode in
409 Known combinations that have some level of understanding include
410 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
411 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
412 create Unicode filenames, what forms of Unicode are accepted and used
413 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
414 and so on, varies. Finding the right level of interfacing to Perl
415 requires some thought. Remember that an OS does not implicate a
418 (The Windows -C command flag "wide API support" has been at least
419 temporarily retired in 5.8.1, and the -C has been repurposed, see
422 Most probably the right way to do this would be this:
423 L</"Virtualize operating system access">.
425 =head2 Unicode in %ENV
427 Currently the %ENV entries are always byte strings.
428 See L</"Virtualize operating system access">.
430 =head2 Unicode and glob()
432 Currently glob patterns and filenames returned from File::Glob::glob()
433 are always byte strings. See L</"Virtualize operating system access">.
435 =head2 use less 'memory'
437 Investigate trade offs to switch out perl's choices on memory usage.
438 Particularly perl should be able to give memory back.
440 This task is incremental - even a little bit of work on it will help.
442 =head2 Re-implement C<:unique> in a way that is actually thread-safe
444 The old implementation made bad assumptions on several levels. A good 90%
445 solution might be just to make C<:unique> work to share the string buffer
446 of SvPVs. That way large constant strings can be shared between ithreads,
447 such as the configuration information in F<Config>.
449 =head2 Make tainting consistent
451 Tainting would be easier to use if it didn't take documented shortcuts and
452 allow taint to "leak" everywhere within an expression.
454 =head2 readpipe(LIST)
456 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
457 running a shell. readpipe() (the function behind qx//) could be similarly
460 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
462 Maybe create a utility that checks after each libperl.a creation that
463 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
464 ever creep back to libperl.a.
466 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
468 Note, of course, that this will only tell whether B<your> platform
469 is using those naughty interfaces.
471 =head2 Audit the code for destruction ordering assumptions
475 /* Need to check SvMAGICAL, as during global destruction it may be that
476 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
477 is now part of the linked list of SV heads, rather than pointing to
478 the original body. */
479 /* FIXME - audit the code for other bugs like this one. */
481 adding the C<SvMAGICAL> check to
483 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
484 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
486 Go through the core and look for similar assumptions that SVs have particular
487 types, as all bets are off during global destruction.
489 =head2 Extend PerlIO and PerlIO::Scalar
491 PerlIO::Scalar doesn't know how to truncate(). Implementing this
492 would require extending the PerlIO vtable.
494 Similarly the PerlIO vtable doesn't know about formats (write()), or
495 about stat(), or chmod()/chown(), utime(), or flock().
497 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
500 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
501 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
504 See also L</"Virtualize operating system access">.
506 =head2 -C on the #! line
508 It should be possible to make -C work correctly if found on the #! line,
509 given that all perl command line options are strict ASCII, and -C changes
510 only the interpretation of non-ASCII characters, and not for the script file
511 handle. To make it work needs some investigation of the ordering of function
512 calls during startup, and (by implication) a bit of tweaking of that order.
515 =head1 Tasks that need a knowledge of the interpreter
517 These tasks would need C knowledge, and knowledge of how the interpreter works,
518 or a willingness to learn.
520 =head2 Implement $value ~~ 0 .. $range
522 It would be nice to extend the syntax of the C<~~> operator to also
523 understand numeric (and maybe alphanumeric) ranges.
525 =head2 Attach/detach debugger from running program
527 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
528 program if you pass the process ID. It would be good to do this with the Perl
529 debugger on a running Perl program, although I'm not sure how it would be
530 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
532 =head2 LVALUE functions for lists
534 The old perltodo notes that lvalue functions don't work for list or hash
535 slices. This would be good to fix.
537 =head2 LVALUE functions in the debugger
539 The old perltodo notes that lvalue functions don't work in the debugger. This
540 would be good to fix.
542 =head2 regexp optimiser optional
544 The regexp optimiser is not optional. It should configurable to be, to allow
545 its performance to be measured, and its bugs to be easily demonstrated.
547 =head2 delete &function
549 Allow to delete functions. One can already undef them, but they're still
552 =head2 C</w> regex modifier
554 That flag would enable to match whole words, and also to interpolate
555 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
557 do { local $"='|'; /\b(?:P)\b/ }
559 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
562 =head2 optional optimizer
564 Make the peephole optimizer optional. Currently it performs two tasks as
565 it walks the optree - genuine peephole optimisations, and necessary fixups of
566 ops. It would be good to find an efficient way to switch out the
567 optimisations whilst keeping the fixups.
569 =head2 You WANT *how* many
571 Currently contexts are void, scalar and list. split has a special mechanism in
572 place to pass in the number of return values wanted. It would be useful to
573 have a general mechanism for this, backwards compatible and little speed hit.
574 This would allow proposals such as short circuiting sort to be implemented
577 =head2 lexical aliases
579 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
581 =head2 entersub XS vs Perl
583 At the moment pp_entersub is huge, and has code to deal with entering both
584 perl and XS subroutines. Subroutine implementations rarely change between
585 perl and XS at run time, so investigate using 2 ops to enter subs (one for
586 XS, one for perl) and swap between if a sub is redefined.
590 self ties are currently illegal because they caused too many segfaults. Maybe
591 the causes of these could be tracked down and self-ties on all types re-
594 =head2 Optimize away @_
596 The old perltodo notes "Look at the "reification" code in C<av.c>".
598 =head2 Properly Unicode safe tokeniser and pads.
600 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
601 variable names are stored in stashes as raw bytes, without the utf-8 flag
602 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
603 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
604 source filters. All this could be fixed.
606 =head2 The yada yada yada operators
608 Perl 6's Synopsis 3 says:
610 I<The ... operator is the "yada, yada, yada" list operator, which is used as
611 the body in function prototypes. It complains bitterly (by calling fail)
612 if it is ever executed. Variant ??? calls warn, and !!! calls die.>
614 Those would be nice to add to Perl 5. That could be done without new ops.
616 =head2 Virtualize operating system access
618 Implement a set of "vtables" that virtualizes operating system access
619 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
620 least these interfaces should take SVs as "name" arguments instead of
621 bare char pointers; probably the most flexible and extensible way
622 would be for the Perl-facing interfaces to accept HVs. The system
623 needs to be per-operating-system and per-file-system
624 hookable/filterable, preferably both from XS and Perl level
625 (L<perlport/"Files and Filesystems"> is good reading at this point,
626 in fact, all of L<perlport> is.)
628 This has actually already been implemented (but only for Win32),
629 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
630 variants go through a set of "vtables" for operating system access,
631 non-Win32 systems currently go straight for the POSIX/UNIX-style
632 system/library call. Similar system as for Win32 should be
633 implemented for all platforms. The existing Win32 implementation
634 probably does not need to survive alongside this proposed new
635 implementation, the approaches could be merged.
637 What would this give us? One often-asked-for feature this would
638 enable is using Unicode for filenames, and other "names" like %ENV,
639 usernames, hostnames, and so forth.
640 (See L<perlunicode/"When Unicode Does Not Happen">.)
642 But this kind of virtualization would also allow for things like
643 virtual filesystems, virtual networks, and "sandboxes" (though as long
644 as dynamic loading of random object code is allowed, not very safe
645 sandboxes since external code of course know not of Perl's vtables).
646 An example of a smaller "sandbox" is that this feature can be used to
647 implement per-thread working directories: Win32 already does this.
649 See also L</"Extend PerlIO and PerlIO::Scalar">.
653 Tasks that will get your name mentioned in the description of the "Highlights
656 =head2 make ithreads more robust
658 Generally make ithreads more robust. See also L</iCOW>
660 This task is incremental - even a little bit of work on it will help, and
661 will be greatly appreciated.
663 One bit would be to write the missing code in sv.c:Perl_dirp_dup.
665 Fix Perl_sv_dup, et al so that threads can return objects.
669 Sarathy and Arthur have a proposal for an improved Copy On Write which
670 specifically will be able to COW new ithreads. If this can be implemented
671 it would be a good thing.
673 =head2 (?{...}) closures in regexps
675 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
677 =head2 A re-entrant regexp engine
679 This will allow the use of a regex from inside (?{ }), (??{ }) and
680 (?(?{ })|) constructs.
682 =head2 Add class set operations to regexp engine
684 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
686 demerphq has this on his todo list, but right at the bottom.