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 POD -E<gt> HTML conversion in the core still sucks
35 Which is crazy given just how simple POD purports to be, and how simple HTML
36 can be. It's not actually I<as> simple as it sounds, particularly with the
37 flexibility POD allows for C<=item>, but it would be good to improve the
38 visual appeal of the HTML generated, and to avoid it having any validation
39 errors. See also L</make HTML install work>, as the layout of installation tree
40 is needed to improve the cross-linking.
42 The addition of C<Pod::Simple> and its related modules may make this task
45 =head2 merge checkpods and podchecker
47 F<pod/checkpods.PL> (and C<make check> in the F<pod/> subdirectory)
48 implements a very basic check for pod files, but the errors it discovers
49 aren't found by podchecker. Add this check to podchecker, get rid of
50 checkpods and have C<make check> use podchecker.
52 =head2 Parallel testing
54 (This probably impacts much more than the core: also the Test::Harness
55 and TAP::* modules on CPAN.)
57 The core regression test suite is getting ever more comprehensive, which has
58 the side effect that it takes longer to run. This isn't so good. Investigate
59 whether it would be feasible to give the harness script the B<option> of
60 running sets of tests in parallel. This would be useful for tests in
61 F<t/op/*.t> and F<t/uni/*.t> and maybe some sets of tests in F<lib/>.
69 How does screen layout work when you're running more than one test?
73 How does the caller of test specify how many tests to run in parallel?
77 How do setup/teardown tests identify themselves?
81 Pugs already does parallel testing - can their approach be re-used?
83 =head2 Make Schwern poorer
85 We should have tests for everything. When all the core's modules are tested,
86 Schwern has promised to donate to $500 to TPF. We may need volunteers to
87 hold him upside down and shake vigorously in order to actually extract the
90 =head2 Improve the coverage of the core tests
92 Use Devel::Cover to ascertain the core modules's test coverage, then add
93 tests that are currently missing.
97 A full test suite for the B module would be nice.
99 =head2 Deparse inlined constants
103 use constant PI => 4;
106 will currently deparse as
108 use constant ('PI', 4);
111 because the tokenizer inlines the value of the constant subroutine C<PI>.
112 This allows various compile time optimisations, such as constant folding
113 and dead code elimination. Where these haven't happened (such as the example
114 above) it ought be possible to make B::Deparse work out the name of the
115 original constant, because just enough information survives in the symbol
116 table to do this. Specifically, the same scalar is used for the constant in
117 the optree as is used for the constant subroutine, so by iterating over all
118 symbol tables and generating a mapping of SV address to constant name, it
119 would be possible to provide B::Deparse with this functionality.
121 =head2 A decent benchmark
123 C<perlbench> seems impervious to any recent changes made to the perl core. It
124 would be useful to have a reasonable general benchmarking suite that roughly
125 represented what current perl programs do, and measurably reported whether
126 tweaks to the core improve, degrade or don't really affect performance, to
127 guide people attempting to optimise the guts of perl. Gisle would welcome
128 new tests for perlbench.
130 =head2 fix tainting bugs
132 Fix the bugs revealed by running the test suite with the C<-t> switch (via
133 C<make test.taintwarn>).
135 =head2 Dual life everything
137 As part of the "dists" plan, anything that doesn't belong in the smallest perl
138 distribution needs to be dual lifed. Anything else can be too. Figure out what
139 changes would be needed to package that module and its tests up for CPAN, and
140 do so. Test it with older perl releases, and fix the problems you find.
142 To make a minimal perl distribution, it's useful to look at
143 F<t/lib/commonsense.t>.
145 =head2 Bundle dual life modules in ext/
147 For maintenance (and branch merging) reasons, it would be useful to move
148 some architecture-independent dual-life modules from lib/ to ext/, if this
149 has no negative impact on the build of perl itself.
151 However, we need to make sure that they are still installed in
152 architecture-independent directories by C<make install>.
154 =head2 Improving C<threads::shared>
156 Investigate whether C<threads::shared> could share aggregates properly with
157 only Perl level changes to shared.pm
159 =head2 POSIX memory footprint
161 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
162 various times worked to cut it down. There is probably still fat to cut out -
163 for example POSIX passes Exporter some very memory hungry data structures.
165 =head2 embed.pl/makedef.pl
167 There is a script F<embed.pl> that generates several header files to prefix
168 all of Perl's symbols in a consistent way, to provide some semblance of
169 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
170 in F<interpvar.h>. Quite a few of the functions and variables
171 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
172 doesn't understand the C macros, so the rules about which symbols are present
173 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
174 It would be good to teach C<embed.pl> to understand the conditional
175 compilation, and hence remove the duplication, and the mistakes it has caused.
177 =head2 use strict; and AutoLoad
179 Currently if you write
182 use AutoLoader 'AUTOLOAD';
187 print join (' ', No, strict, here), "!\n";
190 then C<use strict;> isn't in force within the autoloaded subroutines. It would
191 be more consistent (and less surprising) to arrange for all lexical pragmas
192 in force at the __END__ block to be in force within each autoloaded subroutine.
194 There's a similar problem with SelfLoader.
196 =head2 profile installman
198 The F<installman> script is slow. All it is doing text processing, which we're
199 told is something Perl is good at. So it would be nice to know what it is doing
200 that is taking so much CPU, and where possible address it.
203 =head1 Tasks that need a little sysadmin-type knowledge
205 Or if you prefer, tasks that you would learn from, and broaden your skills
208 =head2 make HTML install work
210 There is an C<installhtml> target in the Makefile. It's marked as
211 "experimental". It would be good to get this tested, make it work reliably, and
212 remove the "experimental" tag. This would include
218 Checking that cross linking between various parts of the documentation works.
219 In particular that links work between the modules (files with POD in F<lib/>)
220 and the core documentation (files in F<pod/>)
224 Work out how to split C<perlfunc> into chunks, preferably one per function
225 group, preferably with general case code that could be used elsewhere.
226 Challenges here are correctly identifying the groups of functions that go
227 together, and making the right named external cross-links point to the right
228 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
229 C<endservent>, two or more C<=items> giving the different parameter lists, such
232 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
233 =item substr EXPR,OFFSET,LENGTH
234 =item substr EXPR,OFFSET
236 and different parameter lists having different meanings. (eg C<select>)
240 =head2 compressed man pages
242 Be able to install them. This would probably need a configure test to see how
243 the system does compressed man pages (same directory/different directory?
244 same filename/different filename), as well as tweaking the F<installman> script
245 to compress as necessary.
247 =head2 Add a code coverage target to the Makefile
249 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
250 to do this manually are roughly
256 do a normal C<Configure>, but include Devel::Cover as a module to install
257 (see F<INSTALL> for how to do this)
265 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
269 Process the resulting Devel::Cover database
273 This just give you the coverage of the F<.pm>s. To also get the C level
280 Additionally tell C<Configure> to use the appropriate C compiler flags for
287 (instead of C<make perl>)
291 After running the tests run C<gcov> to generate all the F<.gcov> files.
292 (Including down in the subdirectories of F<ext/>
296 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
297 to get their stats into the cover_db directory.
301 Then process the Devel::Cover database
305 It would be good to add a single switch to C<Configure> to specify that you
306 wanted to perform perl level coverage, and another to specify C level
307 coverage, and have C<Configure> and the F<Makefile> do all the right things
310 =head2 Make Config.pm cope with differences between built and installed perl
312 Quite often vendors ship a perl binary compiled with their (pay-for)
313 compilers. People install a free compiler, such as gcc. To work out how to
314 build extensions, Perl interrogates C<%Config>, so in this situation
315 C<%Config> describes compilers that aren't there, and extension building
316 fails. This forces people into choosing between re-compiling perl themselves
317 using the compiler they have, or only using modules that the vendor ships.
319 It would be good to find a way teach C<Config.pm> about the installation setup,
320 possibly involving probing at install time or later, so that the C<%Config> in
321 a binary distribution better describes the installed machine, when the
322 installed machine differs from the build machine in some significant way.
324 =head2 linker specification files
326 Some platforms mandate that you provide a list of a shared library's external
327 symbols to the linker, so the core already has the infrastructure in place to
328 do this for generating shared perl libraries. My understanding is that the
329 GNU toolchain can accept an optional linker specification file, and restrict
330 visibility just to symbols declared in that file. It would be good to extend
331 F<makedef.pl> to support this format, and to provide a means within
332 C<Configure> to enable it. This would allow Unix users to test that the
333 export list is correct, and to build a perl that does not pollute the global
334 namespace with private symbols.
336 =head2 Cross-compile support
338 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
339 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
340 assumed then to be copied to TARGET machine and used as a replacement of full
343 This could be done little differently. Namely C<miniperl> should be built for
344 HOST and then full C<perl> with extensions should be compiled for TARGET.
345 This, however, might require extra trickery for %Config: we have one config
346 first for HOST and then another for TARGET. Tools like MakeMaker will be
347 mightily confused. Having around two different types of executables and
348 libraries (HOST and TARGET) makes life interesting for Makefiles and
349 shell (and Perl) scripts. There is $Config{run}, normally empty, which
350 can be used as an execution wrapper. Also note that in some
351 cross-compilation/execution environments the HOST and the TARGET do
352 not see the same filesystem(s), the $Config{run} may need to do some
353 file/directory copying back and forth.
357 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
359 =head1 Tasks that need a little C knowledge
361 These tasks would need a little C knowledge, but don't need any specific
362 background or experience with XS, or how the Perl interpreter works
364 =head2 Weed out needless PERL_UNUSED_ARG
366 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
367 unused arguments. Often the arguments can't be removed, as there is an
368 external constraint that determines the prototype of the function, so this
369 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
370 could be removed. Specifically
376 The prototypes of (nearly all) static functions can be changed
380 Unused arguments generated by short cut macros are wasteful - the short cut
381 macro used can be changed.
385 =head2 Modernize the order of directories in @INC
387 The way @INC is laid out by default, one cannot upgrade core (dual-life)
388 modules without overwriting files. This causes problems for binary
389 package builders. One possible proposal is laid out in this
391 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
395 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
396 On these systems, it might be the default compilation mode, and there
397 is currently no guarantee that passing no use64bitall option to the
398 Configure process will build a 32bit perl. Implementing -Duse32bit*
399 options would be nice for perl 5.12.
401 =head2 Make it clear from -v if this is the exact official release
403 Currently perl from C<p4>/C<rsync> ships with a F<patchlevel.h> file that
404 usually defines one local patch, of the form "MAINT12345" or "RC1". The output
405 of perl -v doesn't report that a perl isn't an official release, and this
406 information can get lost in bugs reports. Because of this, the minor version
407 isn't bumped up until RC time, to minimise the possibility of versions of perl
408 escaping that believe themselves to be newer than they actually are.
410 It would be useful to find an elegant way to have the "this is an interim
411 maintenance release" or "this is a release candidate" in the terse -v output,
412 and have it so that it's easy for the pumpking to remove this just as the
413 release tarball is rolled up. This way the version pulled out of rsync would
414 always say "I'm a development release" and it would be safe to bump the
415 reported minor version as soon as a release ships, which would aid perl
418 This task is really about thinking of an elegant way to arrange the C source
419 such that it's trivial for the Pumpking to flag "this is an official release"
420 when making a tarball, yet leave the default source saying "I'm not the
423 =head2 Profile Perl - am I hot or not?
425 The Perl source code is stable enough that it makes sense to profile it,
426 identify and optimise the hotspots. It would be good to measure the
427 performance of the Perl interpreter using free tools such as cachegrind,
428 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
430 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
431 the ops that are most commonly used. The idea is that by grouping them, their
432 object code will be adjacent in the executable, so they have a greater chance
433 of already being in the CPU cache (or swapped in) due to being near another op
436 Except that it's not clear if these really are the most commonly used ops. So
437 as part of exercising your skills with coverage and profiling tools you might
438 want to determine what ops I<really> are the most commonly used. And in turn
439 suggest evictions and promotions to achieve a better F<pp_hot.c>.
441 One piece of Perl code that might make a good testbed is F<installman>.
443 =head2 Allocate OPs from arenas
445 Currently all new OP structures are individually malloc()ed and free()d.
446 All C<malloc> implementations have space overheads, and are now as fast as
447 custom allocates so it would both use less memory and less CPU to allocate
448 the various OP structures from arenas. The SV arena code can probably be
451 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
452 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
453 probably superior to the use of OP arenas, esp. from a cache locality
454 standpoint. See L<Profile Perl - am I hot or not?>.
456 =head2 Improve win32/wince.c
458 Currently, numerous functions look virtually, if not completely,
459 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
462 =head2 Use secure CRT functions when building with VC8 on Win32
464 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
465 that they were "unsafe" and introduced differently named secure versions of
466 them as replacements, e.g. instead of writing
468 FILE* f = fopen(__FILE__, "r");
473 errno_t err = fopen_s(&f, __FILE__, "r");
475 Currently, the warnings about these deprecations have been disabled by adding
476 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
477 warning suppressant and actually make use of the new secure CRT functions.
479 There is also a similar issue with POSIX CRT function names like fileno having
480 been deprecated in favour of ISO C++ conformant names like _fileno. These
481 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
482 might be nice to do as Microsoft suggest here too, although, unlike the secure
483 functions issue, there is presumably little or no benefit in this case.
485 =head2 Fix POSIX::access() and chdir() on Win32
487 These functions currently take no account of DACLs and therefore do not behave
488 correctly in situations where access is restricted by DACLs (as opposed to the
489 read-only attribute).
491 Furthermore, POSIX::access() behaves differently for directories having the
492 read-only attribute set depending on what CRT library is being used. For
493 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
494 such directories are not writable, whereas in fact all directories are writable
495 unless access is denied by DACLs. (In the case of directories, the read-only
496 attribute actually only means that the directory cannot be deleted.) This CRT
497 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
498 not actually be writable if access is indeed denied by DACLs).
500 For the chdir() issue, see ActiveState bug #74552:
501 http://bugs.activestate.com/show_bug.cgi?id=74552
503 Therefore, DACLs should be checked both for consistency across CRTs and for
506 (Note that perl's -w operator should not be modified to check DACLs. It has
507 been written so that it reflects the state of the read-only attribute, even
508 for directories (whatever CRT is being used), for symmetry with chmod().)
510 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
512 Maybe create a utility that checks after each libperl.a creation that
513 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
514 ever creep back to libperl.a.
516 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
518 Note, of course, that this will only tell whether B<your> platform
519 is using those naughty interfaces.
521 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
523 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
524 (4.1 onwards?) supports C<-fstack-protector>, both of which give
525 protection against various kinds of buffer overflow problems.
526 These should probably be used for compiling Perl whenever available,
527 Configure and/or hints files should be adjusted to probe for the
528 availability of these features and enable them as appropriate.
530 =head2 Arenas for GPs? For MAGIC?
532 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
533 It might be a speed or memory saving to change to using arenas. Or it might
534 not. It would need some suitable benchmarking first. In particular, C<GP>s
535 can probably be changed with minimal compatibility impact (probably nothing
536 outside of the core, or even outside of F<gv.c> allocates them), but they
537 probably aren't allocated/deallocated often enough for a speed saving. Whereas
538 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
539 more externally visible, so changing the rules here may bite external code.
542 =head1 Tasks that need a knowledge of XS
544 These tasks would need C knowledge, and roughly the level of knowledge of
545 the perl API that comes from writing modules that use XS to interface to
548 =head2 safely supporting POSIX SA_SIGINFO
550 Some years ago Jarkko supplied patches to provide support for the POSIX
551 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
553 Unfortunately, it only works with "unsafe" signals, because under safe
554 signals, by the time Perl gets to run the signal handler, the extra
555 information has been lost. Moreover, it's not easy to store it somewhere,
556 as you can't call mutexs, or do anything else fancy, from inside a signal
559 So it strikes me that we could provide safe SA_SIGINFO support
565 Provide global variables for two file descriptors
569 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
570 pipe, store the reader in one, the writer in the other
574 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
575 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
581 serialise signal number, C<struct siginfo_t> (or at least the parts we care
582 about) into a small auto char buff
586 C<write()> that (non-blocking) to the writer fd
592 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
593 to the current per-signal-number counts
597 if it writes 0%, assume the pipe is full. Flag the data as lost?
601 if it writes partially, croak a panic, as your OS is broken.
609 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
610 the pipe", read the data out, deserialise, build the Perl structures on
611 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
616 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
617 of running Perl code inside the signal handler context. (With all the dangers
618 of things like C<malloc> corruption that that currently offers us)
620 For more information see the thread starting with this message:
621 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
623 =head2 autovivification
625 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
627 This task is incremental - even a little bit of work on it will help.
629 =head2 Unicode in Filenames
631 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
632 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
633 system, truncate, unlink, utime, -X. All these could potentially accept
634 Unicode filenames either as input or output (and in the case of system
635 and qx Unicode in general, as input or output to/from the shell).
636 Whether a filesystem - an operating system pair understands Unicode in
639 Known combinations that have some level of understanding include
640 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
641 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
642 create Unicode filenames, what forms of Unicode are accepted and used
643 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
644 and so on, varies. Finding the right level of interfacing to Perl
645 requires some thought. Remember that an OS does not implicate a
648 (The Windows -C command flag "wide API support" has been at least
649 temporarily retired in 5.8.1, and the -C has been repurposed, see
652 Most probably the right way to do this would be this:
653 L</"Virtualize operating system access">.
655 =head2 Unicode in %ENV
657 Currently the %ENV entries are always byte strings.
658 See L</"Virtualize operating system access">.
660 =head2 Unicode and glob()
662 Currently glob patterns and filenames returned from File::Glob::glob()
663 are always byte strings. See L</"Virtualize operating system access">.
665 =head2 Unicode and lc/uc operators
667 Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
668 what the internal encoding of their argument is. That should not be the
669 case. Maybe add a pragma to switch behaviour.
671 =head2 use less 'memory'
673 Investigate trade offs to switch out perl's choices on memory usage.
674 Particularly perl should be able to give memory back.
676 This task is incremental - even a little bit of work on it will help.
678 =head2 Re-implement C<:unique> in a way that is actually thread-safe
680 The old implementation made bad assumptions on several levels. A good 90%
681 solution might be just to make C<:unique> work to share the string buffer
682 of SvPVs. That way large constant strings can be shared between ithreads,
683 such as the configuration information in F<Config>.
685 =head2 Make tainting consistent
687 Tainting would be easier to use if it didn't take documented shortcuts and
688 allow taint to "leak" everywhere within an expression.
690 =head2 readpipe(LIST)
692 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
693 running a shell. readpipe() (the function behind qx//) could be similarly
696 =head2 Audit the code for destruction ordering assumptions
700 /* Need to check SvMAGICAL, as during global destruction it may be that
701 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
702 is now part of the linked list of SV heads, rather than pointing to
703 the original body. */
704 /* FIXME - audit the code for other bugs like this one. */
706 adding the C<SvMAGICAL> check to
708 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
709 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
711 Go through the core and look for similar assumptions that SVs have particular
712 types, as all bets are off during global destruction.
714 =head2 Extend PerlIO and PerlIO::Scalar
716 PerlIO::Scalar doesn't know how to truncate(). Implementing this
717 would require extending the PerlIO vtable.
719 Similarly the PerlIO vtable doesn't know about formats (write()), or
720 about stat(), or chmod()/chown(), utime(), or flock().
722 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
725 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
726 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
729 See also L</"Virtualize operating system access">.
731 =head2 -C on the #! line
733 It should be possible to make -C work correctly if found on the #! line,
734 given that all perl command line options are strict ASCII, and -C changes
735 only the interpretation of non-ASCII characters, and not for the script file
736 handle. To make it work needs some investigation of the ordering of function
737 calls during startup, and (by implication) a bit of tweaking of that order.
739 =head2 Organize error messages
741 Perl's diagnostics (error messages, see L<perldiag>) could use
742 reorganizing and formalizing so that each error message has its
743 stable-for-all-eternity unique id, categorized by severity, type, and
744 subsystem. (The error messages would be listed in a datafile outside
745 of the Perl source code, and the source code would only refer to the
746 messages by the id.) This clean-up and regularizing should apply
747 for all croak() messages.
749 This would enable all sorts of things: easier translation/localization
750 of the messages (though please do keep in mind the caveats of
751 L<Locale::Maketext> about too straightforward approaches to
752 translation), filtering by severity, and instead of grepping for a
753 particular error message one could look for a stable error id. (Of
754 course, changing the error messages by default would break all the
755 existing software depending on some particular error message...)
757 This kind of functionality is known as I<message catalogs>. Look for
758 inspiration for example in the catgets() system, possibly even use it
759 if available-- but B<only> if available, all platforms will B<not>
762 For the really pure at heart, consider extending this item to cover
763 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
765 =head1 Tasks that need a knowledge of the interpreter
767 These tasks would need C knowledge, and knowledge of how the interpreter works,
768 or a willingness to learn.
770 =head2 lexicals used only once
774 $ perl -we '$pie = 42'
775 Name "main::pie" used only once: possible typo at -e line 1.
779 $ perl -we 'my $pie = 42'
781 Logically all lexicals used only once should warn, if the user asks for
782 warnings. An unworked RT ticket (#5087) has been open for almost seven
783 years for this discrepancy.
787 The handling of Unicode is unclean in many places. For example, the regexp
788 engine matches in Unicode semantics whenever the string or the pattern is
789 flagged as UTF-8, but that should not be dependent on an internal storage
790 detail of the string. Likewise, case folding behaviour is dependent on the
791 UTF8 internal flag being on or off.
793 =head2 Properly Unicode safe tokeniser and pads.
795 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
796 variable names are stored in stashes as raw bytes, without the utf-8 flag
797 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
798 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
799 source filters. All this could be fixed.
801 =head2 state variable initialization in list context
803 Currently this is illegal:
805 state ($a, $b) = foo();
807 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
808 semantics, which is tricky to implement in Perl 5 as currently they produce
809 the same opcode trees. The Perl 6 design is firm, so it would be good to
810 implement the necessary code in Perl 5. There are comments in
811 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
812 constructions involving state variables.
814 =head2 Implement $value ~~ 0 .. $range
816 It would be nice to extend the syntax of the C<~~> operator to also
817 understand numeric (and maybe alphanumeric) ranges.
819 =head2 A does() built-in
821 Like ref(), only useful. It would call the C<DOES> method on objects; it
822 would also tell whether something can be dereferenced as an
823 array/hash/etc., or used as a regexp, etc.
824 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
826 =head2 Tied filehandles and write() don't mix
828 There is no method on tied filehandles to allow them to be called back by
831 =head2 Attach/detach debugger from running program
833 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
834 program if you pass the process ID. It would be good to do this with the Perl
835 debugger on a running Perl program, although I'm not sure how it would be
836 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
838 =head2 Optimize away empty destructors
840 Defining an empty DESTROY method might be useful (notably in
841 AUTOLOAD-enabled classes), but it's still a bit expensive to call. That
842 could probably be optimized.
844 =head2 LVALUE functions for lists
846 The old perltodo notes that lvalue functions don't work for list or hash
847 slices. This would be good to fix.
849 =head2 LVALUE functions in the debugger
851 The old perltodo notes that lvalue functions don't work in the debugger. This
852 would be good to fix.
854 =head2 regexp optimiser optional
856 The regexp optimiser is not optional. It should configurable to be, to allow
857 its performance to be measured, and its bugs to be easily demonstrated.
859 =head2 delete &function
861 Allow to delete functions. One can already undef them, but they're still
864 =head2 C</w> regex modifier
866 That flag would enable to match whole words, and also to interpolate
867 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
869 do { local $"='|'; /\b(?:P)\b/ }
871 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
874 =head2 optional optimizer
876 Make the peephole optimizer optional. Currently it performs two tasks as
877 it walks the optree - genuine peephole optimisations, and necessary fixups of
878 ops. It would be good to find an efficient way to switch out the
879 optimisations whilst keeping the fixups.
881 =head2 You WANT *how* many
883 Currently contexts are void, scalar and list. split has a special mechanism in
884 place to pass in the number of return values wanted. It would be useful to
885 have a general mechanism for this, backwards compatible and little speed hit.
886 This would allow proposals such as short circuiting sort to be implemented
889 =head2 lexical aliases
891 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
893 =head2 entersub XS vs Perl
895 At the moment pp_entersub is huge, and has code to deal with entering both
896 perl and XS subroutines. Subroutine implementations rarely change between
897 perl and XS at run time, so investigate using 2 ops to enter subs (one for
898 XS, one for perl) and swap between if a sub is redefined.
902 Self-ties are currently illegal because they caused too many segfaults. Maybe
903 the causes of these could be tracked down and self-ties on all types
906 =head2 Optimize away @_
908 The old perltodo notes "Look at the "reification" code in C<av.c>".
910 =head2 Virtualize operating system access
912 Implement a set of "vtables" that virtualizes operating system access
913 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
914 least these interfaces should take SVs as "name" arguments instead of
915 bare char pointers; probably the most flexible and extensible way
916 would be for the Perl-facing interfaces to accept HVs. The system
917 needs to be per-operating-system and per-file-system
918 hookable/filterable, preferably both from XS and Perl level
919 (L<perlport/"Files and Filesystems"> is good reading at this point,
920 in fact, all of L<perlport> is.)
922 This has actually already been implemented (but only for Win32),
923 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
924 variants go through a set of "vtables" for operating system access,
925 non-Win32 systems currently go straight for the POSIX/UNIX-style
926 system/library call. Similar system as for Win32 should be
927 implemented for all platforms. The existing Win32 implementation
928 probably does not need to survive alongside this proposed new
929 implementation, the approaches could be merged.
931 What would this give us? One often-asked-for feature this would
932 enable is using Unicode for filenames, and other "names" like %ENV,
933 usernames, hostnames, and so forth.
934 (See L<perlunicode/"When Unicode Does Not Happen">.)
936 But this kind of virtualization would also allow for things like
937 virtual filesystems, virtual networks, and "sandboxes" (though as long
938 as dynamic loading of random object code is allowed, not very safe
939 sandboxes since external code of course know not of Perl's vtables).
940 An example of a smaller "sandbox" is that this feature can be used to
941 implement per-thread working directories: Win32 already does this.
943 See also L</"Extend PerlIO and PerlIO::Scalar">.
945 =head2 Investigate PADTMP hash pessimisation
947 The peephole optimier converts constants used for hash key lookups to shared
948 hash key scalars. Under ithreads, something is undoing this work.
949 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
951 =head2 Store the current pad in the OP slab allocator
954 I hope that I got that "current pad" part correct
956 Currently we leak ops in various cases of parse failure. I suggested that we
957 could solve this by always using the op slab allocator, and walking it to
958 free ops. Dave comments that as some ops are already freed during optree
959 creation one would have to mark which ops are freed, and not double free them
960 when walking the slab. He notes that one problem with this is that for some ops
961 you have to know which pad was current at the time of allocation, which does
962 change. I suggested storing a pointer to the current pad in the memory allocated
963 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
964 that this would work.
966 =head2 repack the optree
968 Repacking the optree after execution order is determined could allow
969 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
970 filling. The slab allocator could be reused for this purpose. I think that
971 the best way to do this is to make it an optional step just before the
972 completed optree is attached to anything else, and to use the slab allocator
973 unchanged, so that freeing ops is identical whether or not this step runs.
974 Note that the slab allocator allocates ops downwards in memory, so one would
975 have to actually "allocate" the ops in reverse-execution order to get them
976 contiguous in memory in execution order.
978 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
980 Note that running this copy, and then freeing all the old location ops would
981 cause their slabs to be freed, which would eliminate possible memory wastage if
982 the previous suggestion is implemented, and we swap slabs more frequently.
984 =head2 eliminate incorrect line numbers in warnings
992 } elsif ($undef == 0) {
995 used to produce this output:
997 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
998 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1000 where the line of the second warning was misreported - it should be line 5.
1001 Rafael fixed this - the problem arose because there was no nextstate OP
1002 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1003 reports that the currently executing line is line 4. The solution was to inject
1004 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1005 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1006 numbers became misreported. (Jenga!)
1008 The problem is more general than C<elsif> (although the C<elsif> case is the
1009 most common and the most confusing). Ideally this code
1019 would produce this output
1021 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1022 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1024 (rather than lines 4 and 5), but this would seem to require every OP to carry
1025 (at least) line number information.
1027 What might work is to have an optional line number in memory just before the
1028 BASEOP structure, with a flag bit in the op to say whether it's present.
1029 Initially during compile every OP would carry its line number. Then add a late
1030 pass to the optimiser (potentially combined with L</repack the optree>) which
1031 looks at the two ops on every edge of the graph of the execution path. If
1032 the line number changes, flags the destination OP with this information.
1033 Once all paths are traced, replace every op with the flag with a
1034 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1035 control on to the true op. All ops would then be replaced by variants that
1036 do not store the line number. (Which, logically, why it would work best in
1037 conjunction with L</repack the optree>, as that is already copying/reallocating
1040 (Although I should note that we're not certain that doing this for the general
1043 =head2 optimize tail-calls
1045 Tail-calls present an opportunity for broadly applicable optimization;
1046 anywhere that C<< return foo(...) >> is called, the outer return can
1047 be replaced by a goto, and foo will return directly to the outer
1048 caller, saving (conservatively) 25% of perl's call&return cost, which
1049 is relatively higher than in C. The scheme language is known to do
1050 this heavily. B::Concise provides good insight into where this
1051 optimization is possible, ie anywhere entersub,leavesub op-sequence
1054 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1056 Bottom line on this is probably a new pp_tailcall function which
1057 combines the code in pp_entersub, pp_leavesub. This should probably
1058 be done 1st in XS, and using B::Generate to patch the new OP into the
1063 Tasks that will get your name mentioned in the description of the "Highlights
1066 =head2 make ithreads more robust
1068 Generally make ithreads more robust. See also L</iCOW>
1070 This task is incremental - even a little bit of work on it will help, and
1071 will be greatly appreciated.
1073 One bit would be to write the missing code in sv.c:Perl_dirp_dup.
1075 Fix Perl_sv_dup, et al so that threads can return objects.
1079 Sarathy and Arthur have a proposal for an improved Copy On Write which
1080 specifically will be able to COW new ithreads. If this can be implemented
1081 it would be a good thing.
1083 =head2 (?{...}) closures in regexps
1085 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1087 =head2 A re-entrant regexp engine
1089 This will allow the use of a regex from inside (?{ }), (??{ }) and
1090 (?(?{ })|) constructs.
1092 =head2 Add class set operations to regexp engine
1094 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1096 demerphq has this on his todo list, but right at the bottom.