3 perltodo - Perl TO-DO List
7 This is a list of wishes for Perl. The tasks we think are smaller or
8 easier are listed first. Anyone is welcome to work on any of these,
9 but it's a good idea to first contact I<perl5-porters@perl.org> to
10 avoid duplication of effort, and to learn from any previous attempts.
11 By all means contact a pumpking privately first if you prefer.
13 Whilst patches to make the list shorter are most welcome, ideas to add to
14 the list are also encouraged. Check the perl5-porters archives for past
15 ideas, and any discussion about them. One set of archives may be found at:
17 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
19 What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
20 not, but if your patch is incorporated, then we'll add your name to the
21 F<AUTHORS> file, which ships in the official distribution. How many other
22 programming languages offer you 1 line of immortality?
24 =head1 Tasks that only need Perl knowledge
26 =head2 Remove duplication of test setup.
28 Schwern notes, that there's duplication of code - lots and lots of tests have
29 some variation on the big block of C<$Is_Foo> checks. We can safely put this
30 into a file, change it to build an C<%Is> hash and require it. Maybe just put
31 it into F<test.pl>. Throw in the handy tainting subroutines.
33 =head2 merge common code in installperl and installman
35 There are some common subroutines and a common C<BEGIN> block in F<installperl>
36 and F<installman>. These should probably be merged. It would also be good to
37 check for duplication in all the utility scripts supplied in the source
38 tarball. It might be good to move them all to a subdirectory, but this would
39 require careful checking to find all places that call them, and change those
42 =head2 common test code for timed bail out
44 Write portable self destruct code for tests to stop them burning CPU in
45 infinite loops. This needs to avoid using alarm, as some of the tests are
46 testing alarm/sleep or timers.
48 =head2 POD -E<gt> HTML conversion in the core still sucks
50 Which is crazy given just how simple POD purports to be, and how simple HTML
51 can be. It's not actually I<as> simple as it sounds, particularly with the
52 flexibility POD allows for C<=item>, but it would be good to improve the
53 visual appeal of the HTML generated, and to avoid it having any validation
54 errors. See also L</make HTML install work>, as the layout of installation tree
55 is needed to improve the cross-linking.
57 The addition of C<Pod::Simple> and its related modules may make this task
60 =head2 merge checkpods and podchecker
62 F<pod/checkpods.PL> (and C<make check> in the F<pod/> subdirectory)
63 implements a very basic check for pod files, but the errors it discovers
64 aren't found by podchecker. Add this check to podchecker, get rid of
65 checkpods and have C<make check> use podchecker.
67 =head2 Parallel testing
69 (This probably impacts much more than the core: also the Test::Harness
70 and TAP::* modules on CPAN.)
72 The core regression test suite is getting ever more comprehensive, which has
73 the side effect that it takes longer to run. This isn't so good. Investigate
74 whether it would be feasible to give the harness script the B<option> of
75 running sets of tests in parallel. This would be useful for tests in
76 F<t/op/*.t> and F<t/uni/*.t> and maybe some sets of tests in F<lib/>.
84 How does screen layout work when you're running more than one test?
88 How does the caller of test specify how many tests to run in parallel?
92 How do setup/teardown tests identify themselves?
96 Pugs already does parallel testing - can their approach be re-used?
98 =head2 Make Schwern poorer
100 We should have tests for everything. When all the core's modules are tested,
101 Schwern has promised to donate to $500 to TPF. We may need volunteers to
102 hold him upside down and shake vigorously in order to actually extract the
105 =head2 Improve the coverage of the core tests
107 Use Devel::Cover to ascertain the core modules's test coverage, then add
108 tests that are currently missing.
112 A full test suite for the B module would be nice.
114 =head2 Deparse inlined constants
118 use constant PI => 4;
121 will currently deparse as
123 use constant ('PI', 4);
126 because the tokenizer inlines the value of the constant subroutine C<PI>.
127 This allows various compile time optimisations, such as constant folding
128 and dead code elimination. Where these haven't happened (such as the example
129 above) it ought be possible to make B::Deparse work out the name of the
130 original constant, because just enough information survives in the symbol
131 table to do this. Specifically, the same scalar is used for the constant in
132 the optree as is used for the constant subroutine, so by iterating over all
133 symbol tables and generating a mapping of SV address to constant name, it
134 would be possible to provide B::Deparse with this functionality.
136 =head2 A decent benchmark
138 C<perlbench> seems impervious to any recent changes made to the perl core. It
139 would be useful to have a reasonable general benchmarking suite that roughly
140 represented what current perl programs do, and measurably reported whether
141 tweaks to the core improve, degrade or don't really affect performance, to
142 guide people attempting to optimise the guts of perl. Gisle would welcome
143 new tests for perlbench.
145 =head2 fix tainting bugs
147 Fix the bugs revealed by running the test suite with the C<-t> switch (via
148 C<make test.taintwarn>).
150 =head2 Dual life everything
152 As part of the "dists" plan, anything that doesn't belong in the smallest perl
153 distribution needs to be dual lifed. Anything else can be too. Figure out what
154 changes would be needed to package that module and its tests up for CPAN, and
155 do so. Test it with older perl releases, and fix the problems you find.
157 To make a minimal perl distribution, it's useful to look at
158 F<t/lib/commonsense.t>.
160 =head2 Improving C<threads::shared>
162 Investigate whether C<threads::shared> could share aggregates properly with
163 only Perl level changes to shared.pm
165 =head2 POSIX memory footprint
167 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
168 various times worked to cut it down. There is probably still fat to cut out -
169 for example POSIX passes Exporter some very memory hungry data structures.
171 =head2 embed.pl/makedef.pl
173 There is a script F<embed.pl> that generates several header files to prefix
174 all of Perl's symbols in a consistent way, to provide some semblance of
175 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
176 in F<interpvar.h>. Quite a few of the functions and variables
177 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
178 doesn't understand the C macros, so the rules about which symbols are present
179 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
180 It would be good to teach C<embed.pl> to understand the conditional
181 compilation, and hence remove the duplication, and the mistakes it has caused.
183 =head2 use strict; and AutoLoad
185 Currently if you write
188 use AutoLoader 'AUTOLOAD';
193 print join (' ', No, strict, here), "!\n";
196 then C<use strict;> isn't in force within the autoloaded subroutines. It would
197 be more consistent (and less surprising) to arrange for all lexical pragmas
198 in force at the __END__ block to be in force within each autoloaded subroutine.
200 There's a similar problem with SelfLoader.
202 =head1 Tasks that need a little sysadmin-type knowledge
204 Or if you prefer, tasks that you would learn from, and broaden your skills
207 =head2 make HTML install work
209 There is an C<installhtml> target in the Makefile. It's marked as
210 "experimental". It would be good to get this tested, make it work reliably, and
211 remove the "experimental" tag. This would include
217 Checking that cross linking between various parts of the documentation works.
218 In particular that links work between the modules (files with POD in F<lib/>)
219 and the core documentation (files in F<pod/>)
223 Work out how to split C<perlfunc> into chunks, preferably one per function
224 group, preferably with general case code that could be used elsewhere.
225 Challenges here are correctly identifying the groups of functions that go
226 together, and making the right named external cross-links point to the right
227 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
228 C<endservent>, two or more C<=items> giving the different parameter lists, such
231 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
232 =item substr EXPR,OFFSET,LENGTH
233 =item substr EXPR,OFFSET
235 and different parameter lists having different meanings. (eg C<select>)
239 =head2 compressed man pages
241 Be able to install them. This would probably need a configure test to see how
242 the system does compressed man pages (same directory/different directory?
243 same filename/different filename), as well as tweaking the F<installman> script
244 to compress as necessary.
246 =head2 Add a code coverage target to the Makefile
248 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
249 to do this manually are roughly
255 do a normal C<Configure>, but include Devel::Cover as a module to install
256 (see F<INSTALL> for how to do this)
264 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
268 Process the resulting Devel::Cover database
272 This just give you the coverage of the F<.pm>s. To also get the C level
279 Additionally tell C<Configure> to use the appropriate C compiler flags for
286 (instead of C<make perl>)
290 After running the tests run C<gcov> to generate all the F<.gcov> files.
291 (Including down in the subdirectories of F<ext/>
295 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
296 to get their stats into the cover_db directory.
300 Then process the Devel::Cover database
304 It would be good to add a single switch to C<Configure> to specify that you
305 wanted to perform perl level coverage, and another to specify C level
306 coverage, and have C<Configure> and the F<Makefile> do all the right things
309 =head2 Make Config.pm cope with differences between built and installed perl
311 Quite often vendors ship a perl binary compiled with their (pay-for)
312 compilers. People install a free compiler, such as gcc. To work out how to
313 build extensions, Perl interrogates C<%Config>, so in this situation
314 C<%Config> describes compilers that aren't there, and extension building
315 fails. This forces people into choosing between re-compiling perl themselves
316 using the compiler they have, or only using modules that the vendor ships.
318 It would be good to find a way teach C<Config.pm> about the installation setup,
319 possibly involving probing at install time or later, so that the C<%Config> in
320 a binary distribution better describes the installed machine, when the
321 installed machine differs from the build machine in some significant way.
323 =head2 linker specification files
325 Some platforms mandate that you provide a list of a shared library's external
326 symbols to the linker, so the core already has the infrastructure in place to
327 do this for generating shared perl libraries. My understanding is that the
328 GNU toolchain can accept an optional linker specification file, and restrict
329 visibility just to symbols declared in that file. It would be good to extend
330 F<makedef.pl> to support this format, and to provide a means within
331 C<Configure> to enable it. This would allow Unix users to test that the
332 export list is correct, and to build a perl that does not pollute the global
333 namespace with private symbols.
335 =head2 Cross-compile support
337 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
338 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
339 assumed then to be copied to TARGET machine and used as a replacement of full
342 This could be done little differently. Namely C<miniperl> should be built for
343 HOST and then full C<perl> with extensions should be compiled for TARGET.
344 This, however, might require extra trickery for %Config: we have one config
345 first for HOST and then another for TARGET. Tools like MakeMaker will be
346 mightily confused. Having around two different types of executables and
347 libraries (HOST and TARGET) makes life interesting for Makefiles and
348 shell (and Perl) scripts. There is $Config{run}, normally empty, which
349 can be used as an execution wrapper. Also note that in some
350 cross-compilation/execution environments the HOST and the TARGET do
351 not see the same filesystem(s), the $Config{run} may need to do some
352 file/directory copying back and forth.
356 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
358 =head1 Tasks that need a little C knowledge
360 These tasks would need a little C knowledge, but don't need any specific
361 background or experience with XS, or how the Perl interpreter works
363 =head2 Weed out needless PERL_UNUSED_ARG
365 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
366 unused arguments. Often the arguments can't be removed, as there is an
367 external constraint that determines the prototype of the function, so this
368 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
369 could be removed. Specifically
375 The prototypes of (nearly all) static functions can be changed
379 Unused arguments generated by short cut macros are wasteful - the short cut
380 macro used can be changed.
384 =head2 Modernize the order of directories in @INC
386 The way @INC is laid out by default, one cannot upgrade core (dual-life)
387 modules without overwriting files. This causes problems for binary
388 package builders. One possible proposal is laid out in this
390 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
394 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
395 On these systems, it might be the default compilation mode, and there
396 is currently no guarantee that passing no use64bitall option to the
397 Configure process will build a 32bit perl. Implementing -Duse32bit*
398 options would be nice for perl 5.12.
400 =head2 Make it clear from -v if this is the exact official release
402 Currently perl from C<p4>/C<rsync> ships with a F<patchlevel.h> file that
403 usually defines one local patch, of the form "MAINT12345" or "RC1". The output
404 of perl -v doesn't report that a perl isn't an official release, and this
405 information can get lost in bugs reports. Because of this, the minor version
406 isn't bumped up until RC time, to minimise the possibility of versions of perl
407 escaping that believe themselves to be newer than they actually are.
409 It would be useful to find an elegant way to have the "this is an interim
410 maintenance release" or "this is a release candidate" in the terse -v output,
411 and have it so that it's easy for the pumpking to remove this just as the
412 release tarball is rolled up. This way the version pulled out of rsync would
413 always say "I'm a development release" and it would be safe to bump the
414 reported minor version as soon as a release ships, which would aid perl
417 This task is really about thinking of an elegant way to arrange the C source
418 such that it's trivial for the Pumpking to flag "this is an official release"
419 when making a tarball, yet leave the default source saying "I'm not the
422 =head2 Profile Perl - am I hot or not?
424 The Perl source code is stable enough that it makes sense to profile it,
425 identify and optimise the hotspots. It would be good to measure the
426 performance of the Perl interpreter using free tools such as cachegrind,
427 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
429 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
430 the ops that are most commonly used. The idea is that by grouping them, their
431 object code will be adjacent in the executable, so they have a greater chance
432 of already being in the CPU cache (or swapped in) due to being near another op
435 Except that it's not clear if these really are the most commonly used ops. So
436 as part of exercising your skills with coverage and profiling tools you might
437 want to determine what ops I<really> are the most commonly used. And in turn
438 suggest evictions and promotions to achieve a better F<pp_hot.c>.
440 =head2 Allocate OPs from arenas
442 Currently all new OP structures are individually malloc()ed and free()d.
443 All C<malloc> implementations have space overheads, and are now as fast as
444 custom allocates so it would both use less memory and less CPU to allocate
445 the various OP structures from arenas. The SV arena code can probably be
448 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
449 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
450 probably superior to the use of OP arenas, esp. from a cache locality
451 standpoint. See L<Profile Perl - am I hot or not?>.
453 =head2 Improve win32/wince.c
455 Currently, numerous functions look virtually, if not completely,
456 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
459 =head2 Use secure CRT functions when building with VC8 on Win32
461 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
462 that they were "unsafe" and introduced differently named secure versions of
463 them as replacements, e.g. instead of writing
465 FILE* f = fopen(__FILE__, "r");
470 errno_t err = fopen_s(&f, __FILE__, "r");
472 Currently, the warnings about these deprecations have been disabled by adding
473 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
474 warning suppressant and actually make use of the new secure CRT functions.
476 There is also a similar issue with POSIX CRT function names like fileno having
477 been deprecated in favour of ISO C++ conformant names like _fileno. These
478 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
479 might be nice to do as Microsoft suggest here too, although, unlike the secure
480 functions issue, there is presumably little or no benefit in this case.
482 =head2 Fix POSIX::access() and chdir() on Win32
484 These functions currently take no account of DACLs and therefore do not behave
485 correctly in situations where access is restricted by DACLs (as opposed to the
486 read-only attribute).
488 Furthermore, POSIX::access() behaves differently for directories having the
489 read-only attribute set depending on what CRT library is being used. For
490 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
491 such directories are not writable, whereas in fact all directories are writable
492 unless access is denied by DACLs. (In the case of directories, the read-only
493 attribute actually only means that the directory cannot be deleted.) This CRT
494 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
495 not actually be writable if access is indeed denied by DACLs).
497 For the chdir() issue, see ActiveState bug #74552:
498 http://bugs.activestate.com/show_bug.cgi?id=74552
500 Therefore, DACLs should be checked both for consistency across CRTs and for
503 (Note that perl's -w operator should not be modified to check DACLs. It has
504 been written so that it reflects the state of the read-only attribute, even
505 for directories (whatever CRT is being used), for symmetry with chmod().)
507 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
509 Maybe create a utility that checks after each libperl.a creation that
510 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
511 ever creep back to libperl.a.
513 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
515 Note, of course, that this will only tell whether B<your> platform
516 is using those naughty interfaces.
518 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
520 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
521 (4.1 onwards?) supports C<-fstack-protector>, both of which give
522 protection against various kinds of buffer overflow problems.
523 These should probably be used for compiling Perl whenever available,
524 Configure and/or hints files should be adjusted to probe for the
525 availability of these features and enable them as appropriate.
527 =head2 Arenas for GPs? For MAGIC?
529 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
530 It might be a speed or memory saving to change to using arenas. Or it might
531 not. It would need some suitable benchmarking first. In particular, C<GP>s
532 can probably be changed with minimal compatibility impact (probably nothing
533 outside of the core, or even outside of F<gv.c> allocates them), but they
534 probably aren't allocated/deallocated often enough for a speed saving. Whereas
535 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
536 more externally visible, so changing the rules here may bite external code.
539 =head1 Tasks that need a knowledge of XS
541 These tasks would need C knowledge, and roughly the level of knowledge of
542 the perl API that comes from writing modules that use XS to interface to
545 =head2 safely supporting POSIX SA_SIGINFO
547 Some years ago Jarkko supplied patches to provide support for the POSIX
548 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
550 Unfortunately, it only works with "unsafe" signals, because under safe
551 signals, by the time Perl gets to run the signal handler, the extra
552 information has been lost. Moreover, it's not easy to store it somewhere,
553 as you can't call mutexs, or do anything else fancy, from inside a signal
556 So it strikes me that we could provide safe SA_SIGINFO support
562 Provide global variables for two file descriptors
566 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
567 pipe, store the reader in one, the writer in the other
571 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
572 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
578 serialise signal number, C<struct siginfo_t> (or at least the parts we care
579 about) into a small auto char buff
583 C<write()> that (non-blocking) to the writer fd
589 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
590 to the current per-signal-number counts
594 if it writes 0%, assume the pipe is full. Flag the data as lost?
598 if it writes partially, croak a panic, as your OS is broken.
606 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
607 the pipe", read the data out, deserialise, build the Perl structures on
608 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
613 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
614 of running Perl code inside the signal handler context. (With all the dangers
615 of things like C<malloc> corruption that that currently offers us)
617 For more information see the thread starting with this message:
618 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
620 =head2 autovivification
622 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
624 This task is incremental - even a little bit of work on it will help.
626 =head2 Unicode in Filenames
628 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
629 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
630 system, truncate, unlink, utime, -X. All these could potentially accept
631 Unicode filenames either as input or output (and in the case of system
632 and qx Unicode in general, as input or output to/from the shell).
633 Whether a filesystem - an operating system pair understands Unicode in
636 Known combinations that have some level of understanding include
637 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
638 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
639 create Unicode filenames, what forms of Unicode are accepted and used
640 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
641 and so on, varies. Finding the right level of interfacing to Perl
642 requires some thought. Remember that an OS does not implicate a
645 (The Windows -C command flag "wide API support" has been at least
646 temporarily retired in 5.8.1, and the -C has been repurposed, see
649 Most probably the right way to do this would be this:
650 L</"Virtualize operating system access">.
652 =head2 Unicode in %ENV
654 Currently the %ENV entries are always byte strings.
655 See L</"Virtualize operating system access">.
657 =head2 Unicode and glob()
659 Currently glob patterns and filenames returned from File::Glob::glob()
660 are always byte strings. See L</"Virtualize operating system access">.
662 =head2 Unicode and lc/uc operators
664 Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
665 what the internal encoding of their argument is. That should not be the
666 case. Maybe add a pragma to switch behaviour.
668 =head2 use less 'memory'
670 Investigate trade offs to switch out perl's choices on memory usage.
671 Particularly perl should be able to give memory back.
673 This task is incremental - even a little bit of work on it will help.
675 =head2 Re-implement C<:unique> in a way that is actually thread-safe
677 The old implementation made bad assumptions on several levels. A good 90%
678 solution might be just to make C<:unique> work to share the string buffer
679 of SvPVs. That way large constant strings can be shared between ithreads,
680 such as the configuration information in F<Config>.
682 =head2 Make tainting consistent
684 Tainting would be easier to use if it didn't take documented shortcuts and
685 allow taint to "leak" everywhere within an expression.
687 =head2 readpipe(LIST)
689 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
690 running a shell. readpipe() (the function behind qx//) could be similarly
693 =head2 Audit the code for destruction ordering assumptions
697 /* Need to check SvMAGICAL, as during global destruction it may be that
698 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
699 is now part of the linked list of SV heads, rather than pointing to
700 the original body. */
701 /* FIXME - audit the code for other bugs like this one. */
703 adding the C<SvMAGICAL> check to
705 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
706 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
708 Go through the core and look for similar assumptions that SVs have particular
709 types, as all bets are off during global destruction.
711 =head2 Extend PerlIO and PerlIO::Scalar
713 PerlIO::Scalar doesn't know how to truncate(). Implementing this
714 would require extending the PerlIO vtable.
716 Similarly the PerlIO vtable doesn't know about formats (write()), or
717 about stat(), or chmod()/chown(), utime(), or flock().
719 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
722 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
723 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
726 See also L</"Virtualize operating system access">.
728 =head2 -C on the #! line
730 It should be possible to make -C work correctly if found on the #! line,
731 given that all perl command line options are strict ASCII, and -C changes
732 only the interpretation of non-ASCII characters, and not for the script file
733 handle. To make it work needs some investigation of the ordering of function
734 calls during startup, and (by implication) a bit of tweaking of that order.
736 =head2 Organize error messages
738 Perl's diagnostics (error messages, see L<perldiag>) could use
739 reorganizing and formalizing so that each error message has its
740 stable-for-all-eternity unique id, categorized by severity, type, and
741 subsystem. (The error messages would be listed in a datafile outside
742 of the Perl source code, and the source code would only refer to the
743 messages by the id.) This clean-up and regularizing should apply
744 for all croak() messages.
746 This would enable all sorts of things: easier translation/localization
747 of the messages (though please do keep in mind the caveats of
748 L<Locale::Maketext> about too straightforward approaches to
749 translation), filtering by severity, and instead of grepping for a
750 particular error message one could look for a stable error id. (Of
751 course, changing the error messages by default would break all the
752 existing software depending on some particular error message...)
754 This kind of functionality is known as I<message catalogs>. Look for
755 inspiration for example in the catgets() system, possibly even use it
756 if available-- but B<only> if available, all platforms will B<not>
759 For the really pure at heart, consider extending this item to cover
760 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
762 =head1 Tasks that need a knowledge of the interpreter
764 These tasks would need C knowledge, and knowledge of how the interpreter works,
765 or a willingness to learn.
767 =head2 lexicals used only once
771 $ perl -we '$pie = 42'
772 Name "main::pie" used only once: possible typo at -e line 1.
776 $ perl -we 'my $pie = 42'
778 Logically all lexicals used only once should warn, if the user asks for
779 warnings. An unworked RT ticket (#5087) has been open for almost seven
780 years for this discrepancy.
784 The handling of Unicode is unclean in many places. For example, the regexp
785 engine matches in Unicode semantics whenever the string or the pattern is
786 flagged as UTF-8, but that should not be dependent on an internal storage
787 detail of the string. Likewise, case folding behaviour is dependent on the
788 UTF8 internal flag being on or off.
790 =head2 Properly Unicode safe tokeniser and pads.
792 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
793 variable names are stored in stashes as raw bytes, without the utf-8 flag
794 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
795 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
796 source filters. All this could be fixed.
798 =head2 state variable initialization in list context
800 Currently this is illegal:
802 state ($a, $b) = foo();
804 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
805 semantics, which is tricky to implement in Perl 5 as currently they produce
806 the same opcode trees. The Perl 6 design is firm, so it would be good to
807 implement the necessary code in Perl 5. There are comments in
808 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
809 constructions involving state variables.
811 =head2 Implement $value ~~ 0 .. $range
813 It would be nice to extend the syntax of the C<~~> operator to also
814 understand numeric (and maybe alphanumeric) ranges.
816 =head2 A does() built-in
818 Like ref(), only useful. It would call the C<DOES> method on objects; it
819 would also tell whether something can be dereferenced as an
820 array/hash/etc., or used as a regexp, etc.
821 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
823 =head2 Tied filehandles and write() don't mix
825 There is no method on tied filehandles to allow them to be called back by
828 =head2 Attach/detach debugger from running program
830 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
831 program if you pass the process ID. It would be good to do this with the Perl
832 debugger on a running Perl program, although I'm not sure how it would be
833 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
835 =head2 Optimize away empty destructors
837 Defining an empty DESTROY method might be useful (notably in
838 AUTOLOAD-enabled classes), but it's still a bit expensive to call. That
839 could probably be optimized.
841 =head2 LVALUE functions for lists
843 The old perltodo notes that lvalue functions don't work for list or hash
844 slices. This would be good to fix.
846 =head2 LVALUE functions in the debugger
848 The old perltodo notes that lvalue functions don't work in the debugger. This
849 would be good to fix.
851 =head2 regexp optimiser optional
853 The regexp optimiser is not optional. It should configurable to be, to allow
854 its performance to be measured, and its bugs to be easily demonstrated.
856 =head2 delete &function
858 Allow to delete functions. One can already undef them, but they're still
861 =head2 C</w> regex modifier
863 That flag would enable to match whole words, and also to interpolate
864 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
866 do { local $"='|'; /\b(?:P)\b/ }
868 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
871 =head2 optional optimizer
873 Make the peephole optimizer optional. Currently it performs two tasks as
874 it walks the optree - genuine peephole optimisations, and necessary fixups of
875 ops. It would be good to find an efficient way to switch out the
876 optimisations whilst keeping the fixups.
878 =head2 You WANT *how* many
880 Currently contexts are void, scalar and list. split has a special mechanism in
881 place to pass in the number of return values wanted. It would be useful to
882 have a general mechanism for this, backwards compatible and little speed hit.
883 This would allow proposals such as short circuiting sort to be implemented
886 =head2 lexical aliases
888 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
890 =head2 entersub XS vs Perl
892 At the moment pp_entersub is huge, and has code to deal with entering both
893 perl and XS subroutines. Subroutine implementations rarely change between
894 perl and XS at run time, so investigate using 2 ops to enter subs (one for
895 XS, one for perl) and swap between if a sub is redefined.
899 Self-ties are currently illegal because they caused too many segfaults. Maybe
900 the causes of these could be tracked down and self-ties on all types
903 =head2 Optimize away @_
905 The old perltodo notes "Look at the "reification" code in C<av.c>".
907 =head2 The yada yada yada operators
909 Perl 6's Synopsis 3 says:
911 I<The ... operator is the "yada, yada, yada" list operator, which is used as
912 the body in function prototypes. It complains bitterly (by calling fail)
913 if it is ever executed. Variant ??? calls warn, and !!! calls die.>
915 Those would be nice to add to Perl 5. That could be done without new ops.
917 =head2 Virtualize operating system access
919 Implement a set of "vtables" that virtualizes operating system access
920 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
921 least these interfaces should take SVs as "name" arguments instead of
922 bare char pointers; probably the most flexible and extensible way
923 would be for the Perl-facing interfaces to accept HVs. The system
924 needs to be per-operating-system and per-file-system
925 hookable/filterable, preferably both from XS and Perl level
926 (L<perlport/"Files and Filesystems"> is good reading at this point,
927 in fact, all of L<perlport> is.)
929 This has actually already been implemented (but only for Win32),
930 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
931 variants go through a set of "vtables" for operating system access,
932 non-Win32 systems currently go straight for the POSIX/UNIX-style
933 system/library call. Similar system as for Win32 should be
934 implemented for all platforms. The existing Win32 implementation
935 probably does not need to survive alongside this proposed new
936 implementation, the approaches could be merged.
938 What would this give us? One often-asked-for feature this would
939 enable is using Unicode for filenames, and other "names" like %ENV,
940 usernames, hostnames, and so forth.
941 (See L<perlunicode/"When Unicode Does Not Happen">.)
943 But this kind of virtualization would also allow for things like
944 virtual filesystems, virtual networks, and "sandboxes" (though as long
945 as dynamic loading of random object code is allowed, not very safe
946 sandboxes since external code of course know not of Perl's vtables).
947 An example of a smaller "sandbox" is that this feature can be used to
948 implement per-thread working directories: Win32 already does this.
950 See also L</"Extend PerlIO and PerlIO::Scalar">.
952 =head2 Investigate PADTMP hash pessimisation
954 The peephole optimier converts constants used for hash key lookups to shared
955 hash key scalars. Under ithreads, something is undoing this work.
956 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
958 =head2 Store the current pad in the OP slab allocator
961 I hope that I got that "current pad" part correct
963 Currently we leak ops in various cases of parse failure. I suggested that we
964 could solve this by always using the op slab allocator, and walking it to
965 free ops. Dave comments that as some ops are already freed during optree
966 creation one would have to mark which ops are freed, and not double free them
967 when walking the slab. He notes that one problem with this is that for some ops
968 you have to know which pad was current at the time of allocation, which does
969 change. I suggested storing a pointer to the current pad in the memory allocated
970 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
971 that this would work.
973 =head2 repack the optree
975 Repacking the optree after execution order is determined could allow
976 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
977 filling. The slab allocator could be reused for this purpose. I think that
978 the best way to do this is to make it an optional step just before the
979 completed optree is attached to anything else, and to use the slab allocator
980 unchanged, so that freeing ops is identical whether or not this step runs.
981 Note that the slab allocator allocates ops downwards in memory, so one would
982 have to actually "allocate" the ops in reverse-execution order to get them
983 contiguous in memory in execution order.
985 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
987 Note that running this copy, and then freeing all the old location ops would
988 cause their slabs to be freed, which would eliminate possible memory wastage if
989 the previous suggestion is implemented, and we swap slabs more frequently.
991 =head2 eliminate incorrect line numbers in warnings
999 } elsif ($undef == 0) {
1002 used to produce this output:
1004 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1005 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1007 where the line of the second warning was misreported - it should be line 5.
1008 Rafael fixed this - the problem arose because there was no nextstate OP
1009 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1010 reports that the currently executing line is line 4. The solution was to inject
1011 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1012 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1013 numbers became misreported. (Jenga!)
1015 The problem is more general than C<elsif> (although the C<elsif> case is the
1016 most common and the most confusing). Ideally this code
1026 would produce this output
1028 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1029 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1031 (rather than lines 4 and 5), but this would seem to require every OP to carry
1032 (at least) line number information.
1034 What might work is to have an optional line number in memory just before the
1035 BASEOP structure, with a flag bit in the op to say whether it's present.
1036 Initially during compile every OP would carry its line number. Then add a late
1037 pass to the optimiser (potentially combined with L</repack the optree>) which
1038 looks at the two ops on every edge of the graph of the execution path. If
1039 the line number changes, flags the destination OP with this information.
1040 Once all paths are traced, replace every op with the flag with a
1041 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1042 control on to the true op. All ops would then be replaced by variants that
1043 do not store the line number. (Which, logically, why it would work best in
1044 conjunction with L</repack the optree>, as that is already copying/reallocating
1047 (Although I should note that we're not certain that doing this for the general
1050 =head2 optimize tail-calls
1052 Tail-calls present an opportunity for broadly applicable optimization;
1053 anywhere that C<< return foo(...) >> is called, the outer return can
1054 be replaced by a goto, and foo will return directly to the outer
1055 caller, saving (conservatively) 25% of perl's call&return cost, which
1056 is relatively higher than in C. The scheme language is known to do
1057 this heavily. B::Concise provides good insight into where this
1058 optimization is possible, ie anywhere entersub,leavesub op-sequence
1061 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1063 Bottom line on this is probably a new pp_tailcall function which
1064 combines the code in pp_entersub, pp_leavesub. This should probably
1065 be done 1st in XS, and using B::Generate to patch the new OP into the
1070 Tasks that will get your name mentioned in the description of the "Highlights
1073 =head2 make ithreads more robust
1075 Generally make ithreads more robust. See also L</iCOW>
1077 This task is incremental - even a little bit of work on it will help, and
1078 will be greatly appreciated.
1080 One bit would be to write the missing code in sv.c:Perl_dirp_dup.
1082 Fix Perl_sv_dup, et al so that threads can return objects.
1086 Sarathy and Arthur have a proposal for an improved Copy On Write which
1087 specifically will be able to COW new ithreads. If this can be implemented
1088 it would be a good thing.
1090 =head2 (?{...}) closures in regexps
1092 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1094 =head2 A re-entrant regexp engine
1096 This will allow the use of a regex from inside (?{ }), (??{ }) and
1097 (?(?{ })|) constructs.
1099 =head2 Add class set operations to regexp engine
1101 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1103 demerphq has this on his todo list, but right at the bottom.