3 perltodo - Perl TO-DO List
7 This is a list of wishes for Perl. The most up to date version of this file
8 is at http://perl5.git.perl.org/perl.git/blob_plain/HEAD:/pod/perltodo.pod
10 The tasks we think are smaller or easier are listed first. Anyone is welcome
11 to work on any of these, but it's a good idea to first contact
12 I<perl5-porters@perl.org> to avoid duplication of effort, and to learn from
13 any previous attempts. By all means contact a pumpking privately first if you
16 Whilst patches to make the list shorter are most welcome, ideas to add to
17 the list are also encouraged. Check the perl5-porters archives for past
18 ideas, and any discussion about them. One set of archives may be found at:
20 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
22 What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
23 not, but if your patch is incorporated, then we'll add your name to the
24 F<AUTHORS> file, which ships in the official distribution. How many other
25 programming languages offer you 1 line of immortality?
27 =head1 Tasks that only need Perl knowledge
29 =head2 Automate perldelta generation
31 The perldelta file accompanying each release summaries the major changes.
32 It's mostly manually generated currently, but some of that could be
33 automated with a bit of perl, specifically the generation of
37 =item Modules and Pragmata
39 =item New Documentation
45 See F<Porting/how_to_write_a_perldelta.pod> for details.
47 =head2 Remove duplication of test setup.
49 Schwern notes, that there's duplication of code - lots and lots of tests have
50 some variation on the big block of C<$Is_Foo> checks. We can safely put this
51 into a file, change it to build an C<%Is> hash and require it. Maybe just put
52 it into F<test.pl>. Throw in the handy tainting subroutines.
54 =head2 POD -E<gt> HTML conversion in the core still sucks
56 Which is crazy given just how simple POD purports to be, and how simple HTML
57 can be. It's not actually I<as> simple as it sounds, particularly with the
58 flexibility POD allows for C<=item>, but it would be good to improve the
59 visual appeal of the HTML generated, and to avoid it having any validation
60 errors. See also L</make HTML install work>, as the layout of installation tree
61 is needed to improve the cross-linking.
63 The addition of C<Pod::Simple> and its related modules may make this task
66 =head2 Make ExtUtils::ParseXS use strict;
68 F<lib/ExtUtils/ParseXS.pm> contains this line
70 # use strict; # One of these days...
72 Simply uncomment it, and fix all the resulting issues :-)
74 The more practical approach, to break the task down into manageable chunks, is
75 to work your way though the code from bottom to top, or if necessary adding
76 extra C<{ ... }> blocks, and turning on strict within them.
78 =head2 Make Schwern poorer
80 We should have tests for everything. When all the core's modules are tested,
81 Schwern has promised to donate to $500 to TPF. We may need volunteers to
82 hold him upside down and shake vigorously in order to actually extract the
85 =head2 Improve the coverage of the core tests
87 Use Devel::Cover to ascertain the core modules's test coverage, then add
88 tests that are currently missing.
92 A full test suite for the B module would be nice.
94 =head2 A decent benchmark
96 C<perlbench> seems impervious to any recent changes made to the perl core. It
97 would be useful to have a reasonable general benchmarking suite that roughly
98 represented what current perl programs do, and measurably reported whether
99 tweaks to the core improve, degrade or don't really affect performance, to
100 guide people attempting to optimise the guts of perl. Gisle would welcome
101 new tests for perlbench.
103 =head2 fix tainting bugs
105 Fix the bugs revealed by running the test suite with the C<-t> switch (via
106 C<make test.taintwarn>).
108 =head2 Dual life everything
110 As part of the "dists" plan, anything that doesn't belong in the smallest perl
111 distribution needs to be dual lifed. Anything else can be too. Figure out what
112 changes would be needed to package that module and its tests up for CPAN, and
113 do so. Test it with older perl releases, and fix the problems you find.
115 To make a minimal perl distribution, it's useful to look at
116 F<t/lib/commonsense.t>.
118 =head2 Move dual-life pod/*.PL into ext
120 Nearly all the dual-life modules have been moved to F<ext>. However, we
121 still need to move F<pod/*.PL> into their respective directories
122 in F<ext/>. They're referenced by (at least) C<plextract> in F<Makefile.SH>
123 and C<utils> in F<win32/Makefile> and F<win32/makefile.ml>, and listed
124 explicitly in F<win32/pod.mak>, F<vms/descrip_mms.template> and F<utils.lst>
126 =head2 POSIX memory footprint
128 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
129 various times worked to cut it down. There is probably still fat to cut out -
130 for example POSIX passes Exporter some very memory hungry data structures.
132 =head2 embed.pl/makedef.pl
134 There is a script F<embed.pl> that generates several header files to prefix
135 all of Perl's symbols in a consistent way, to provide some semblance of
136 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
137 in F<interpvar.h>. Quite a few of the functions and variables
138 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
139 doesn't understand the C macros, so the rules about which symbols are present
140 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
141 It would be good to teach C<embed.pl> to understand the conditional
142 compilation, and hence remove the duplication, and the mistakes it has caused.
144 =head2 use strict; and AutoLoad
146 Currently if you write
149 use AutoLoader 'AUTOLOAD';
154 print join (' ', No, strict, here), "!\n";
157 then C<use strict;> isn't in force within the autoloaded subroutines. It would
158 be more consistent (and less surprising) to arrange for all lexical pragmas
159 in force at the __END__ block to be in force within each autoloaded subroutine.
161 There's a similar problem with SelfLoader.
163 =head2 profile installman
165 The F<installman> script is slow. All it is doing text processing, which we're
166 told is something Perl is good at. So it would be nice to know what it is doing
167 that is taking so much CPU, and where possible address it.
170 =head1 Tasks that need a little sysadmin-type knowledge
172 Or if you prefer, tasks that you would learn from, and broaden your skills
175 =head2 make HTML install work
177 There is an C<installhtml> target in the Makefile. It's marked as
178 "experimental". It would be good to get this tested, make it work reliably, and
179 remove the "experimental" tag. This would include
185 Checking that cross linking between various parts of the documentation works.
186 In particular that links work between the modules (files with POD in F<lib/>)
187 and the core documentation (files in F<pod/>)
191 Work out how to split C<perlfunc> into chunks, preferably one per function
192 group, preferably with general case code that could be used elsewhere.
193 Challenges here are correctly identifying the groups of functions that go
194 together, and making the right named external cross-links point to the right
195 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
196 C<endservent>, two or more C<=items> giving the different parameter lists, such
199 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
200 =item substr EXPR,OFFSET,LENGTH
201 =item substr EXPR,OFFSET
203 and different parameter lists having different meanings. (eg C<select>)
207 =head2 compressed man pages
209 Be able to install them. This would probably need a configure test to see how
210 the system does compressed man pages (same directory/different directory?
211 same filename/different filename), as well as tweaking the F<installman> script
212 to compress as necessary.
214 =head2 Add a code coverage target to the Makefile
216 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
217 to do this manually are roughly
223 do a normal C<Configure>, but include Devel::Cover as a module to install
224 (see F<INSTALL> for how to do this)
232 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
236 Process the resulting Devel::Cover database
240 This just give you the coverage of the F<.pm>s. To also get the C level
247 Additionally tell C<Configure> to use the appropriate C compiler flags for
254 (instead of C<make perl>)
258 After running the tests run C<gcov> to generate all the F<.gcov> files.
259 (Including down in the subdirectories of F<ext/>
263 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
264 to get their stats into the cover_db directory.
268 Then process the Devel::Cover database
272 It would be good to add a single switch to C<Configure> to specify that you
273 wanted to perform perl level coverage, and another to specify C level
274 coverage, and have C<Configure> and the F<Makefile> do all the right things
277 =head2 Make Config.pm cope with differences between built and installed perl
279 Quite often vendors ship a perl binary compiled with their (pay-for)
280 compilers. People install a free compiler, such as gcc. To work out how to
281 build extensions, Perl interrogates C<%Config>, so in this situation
282 C<%Config> describes compilers that aren't there, and extension building
283 fails. This forces people into choosing between re-compiling perl themselves
284 using the compiler they have, or only using modules that the vendor ships.
286 It would be good to find a way teach C<Config.pm> about the installation setup,
287 possibly involving probing at install time or later, so that the C<%Config> in
288 a binary distribution better describes the installed machine, when the
289 installed machine differs from the build machine in some significant way.
291 =head2 linker specification files
293 Some platforms mandate that you provide a list of a shared library's external
294 symbols to the linker, so the core already has the infrastructure in place to
295 do this for generating shared perl libraries. My understanding is that the
296 GNU toolchain can accept an optional linker specification file, and restrict
297 visibility just to symbols declared in that file. It would be good to extend
298 F<makedef.pl> to support this format, and to provide a means within
299 C<Configure> to enable it. This would allow Unix users to test that the
300 export list is correct, and to build a perl that does not pollute the global
301 namespace with private symbols.
303 =head2 Cross-compile support
305 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
306 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
307 assumed then to be copied to TARGET machine and used as a replacement of full
310 This could be done little differently. Namely C<miniperl> should be built for
311 HOST and then full C<perl> with extensions should be compiled for TARGET.
312 This, however, might require extra trickery for %Config: we have one config
313 first for HOST and then another for TARGET. Tools like MakeMaker will be
314 mightily confused. Having around two different types of executables and
315 libraries (HOST and TARGET) makes life interesting for Makefiles and
316 shell (and Perl) scripts. There is $Config{run}, normally empty, which
317 can be used as an execution wrapper. Also note that in some
318 cross-compilation/execution environments the HOST and the TARGET do
319 not see the same filesystem(s), the $Config{run} may need to do some
320 file/directory copying back and forth.
324 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
326 =head2 Split "linker" from "compiler"
328 Right now, Configure probes for two commands, and sets two variables:
332 =item * C<cc> (in F<cc.U>)
334 This variable holds the name of a command to execute a C compiler which
335 can resolve multiple global references that happen to have the same
336 name. Usual values are F<cc> and F<gcc>.
337 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
339 =item * C<ld> (in F<dlsrc.U>)
341 This variable indicates the program to be used to link
342 libraries for dynamic loading. On some systems, it is F<ld>.
343 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
344 the hint file setting.
348 There is an implicit historical assumption from around Perl5.000alpha
349 something, that C<$cc> is also the correct command for linking object files
350 together to make an executable. This may be true on Unix, but it's not true
351 on other platforms, and there are a maze of work arounds in other places (such
352 as F<Makefile.SH>) to cope with this.
354 Ideally, we should create a new variable to hold the name of the executable
355 linker program, probe for it in F<Configure>, and centralise all the special
356 case logic there or in hints files.
358 A small bikeshed issue remains - what to call it, given that C<$ld> is already
359 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
360 for creating dynamically-loadable modules) and C<$link> could be confused with
361 the Unix command line executable of the same name, which does something
362 completely different. Andy Dougherty makes the counter argument "In parrot, I
363 tried to call the command used to link object files and libraries into an
364 executable F<link>, since that's what my vaguely-remembered DOS and VMS
365 experience suggested. I don't think any real confusion has ensued, so it's
366 probably a reasonable name for perl5 to use."
368 "Alas, I've always worried that introducing it would make things worse,
369 since now the module building utilities would have to look for
370 C<$Config{link}> and institute a fall-back plan if it weren't found."
371 Although I can see that as confusing, given that C<$Config{d_link}> is true
372 when (hard) links are available.
374 =head2 Configure Windows using PowerShell
376 Currently, Windows uses hard-coded config files based to build the
377 config.h for compiling Perl. Makefiles are also hard-coded and need to be
378 hand edited prior to building Perl. While this makes it easy to create a perl.exe
379 that works across multiple Windows versions, being able to accurately
380 configure a perl.exe for a specific Windows versions and VS C++ would be
381 a nice enhancement. With PowerShell available on Windows XP and up, this
382 may now be possible. Step 1 might be to investigate whether this is possible
383 and use this to clean up our current makefile situation. Step 2 would be to
384 see if there would be a way to use our existing metaconfig units to configure a
385 Windows Perl or whether we go in a separate direction and make it so. Of
386 course, we all know what step 3 is.
388 =head2 decouple -g and -DDEBUGGING
390 Currently F<Configure> automatically adds C<-DDEBUGGING> to the C compiler
391 flags if it spots C<-g> in the optimiser flags. The pre-processor directive
392 C<DEBUGGING> enables F<perl>'s command line C<-D> options, but in the process
393 makes F<perl> slower. It would be good to disentangle this logic, so that
394 C-level debugging with C<-g> and Perl level debugging with C<-D> can easily
395 be enabled independently.
397 =head1 Tasks that need a little C knowledge
399 These tasks would need a little C knowledge, but don't need any specific
400 background or experience with XS, or how the Perl interpreter works
402 =head2 Weed out needless PERL_UNUSED_ARG
404 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
405 unused arguments. Often the arguments can't be removed, as there is an
406 external constraint that determines the prototype of the function, so this
407 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
408 could be removed. Specifically
414 The prototypes of (nearly all) static functions can be changed
418 Unused arguments generated by short cut macros are wasteful - the short cut
419 macro used can be changed.
423 =head2 Modernize the order of directories in @INC
425 The way @INC is laid out by default, one cannot upgrade core (dual-life)
426 modules without overwriting files. This causes problems for binary
427 package builders. One possible proposal is laid out in this
429 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
433 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
434 On these systems, it might be the default compilation mode, and there
435 is currently no guarantee that passing no use64bitall option to the
436 Configure process will build a 32bit perl. Implementing -Duse32bit*
437 options would be nice for perl 5.12.
439 =head2 Profile Perl - am I hot or not?
441 The Perl source code is stable enough that it makes sense to profile it,
442 identify and optimise the hotspots. It would be good to measure the
443 performance of the Perl interpreter using free tools such as cachegrind,
444 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
446 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
447 the ops that are most commonly used. The idea is that by grouping them, their
448 object code will be adjacent in the executable, so they have a greater chance
449 of already being in the CPU cache (or swapped in) due to being near another op
452 Except that it's not clear if these really are the most commonly used ops. So
453 as part of exercising your skills with coverage and profiling tools you might
454 want to determine what ops I<really> are the most commonly used. And in turn
455 suggest evictions and promotions to achieve a better F<pp_hot.c>.
457 One piece of Perl code that might make a good testbed is F<installman>.
459 =head2 Allocate OPs from arenas
461 Currently all new OP structures are individually malloc()ed and free()d.
462 All C<malloc> implementations have space overheads, and are now as fast as
463 custom allocates so it would both use less memory and less CPU to allocate
464 the various OP structures from arenas. The SV arena code can probably be
467 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
468 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
469 probably superior to the use of OP arenas, esp. from a cache locality
470 standpoint. See L<Profile Perl - am I hot or not?>.
472 =head2 Improve win32/wince.c
474 Currently, numerous functions look virtually, if not completely,
475 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
478 =head2 Use secure CRT functions when building with VC8 on Win32
480 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
481 that they were "unsafe" and introduced differently named secure versions of
482 them as replacements, e.g. instead of writing
484 FILE* f = fopen(__FILE__, "r");
489 errno_t err = fopen_s(&f, __FILE__, "r");
491 Currently, the warnings about these deprecations have been disabled by adding
492 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
493 warning suppressant and actually make use of the new secure CRT functions.
495 There is also a similar issue with POSIX CRT function names like fileno having
496 been deprecated in favour of ISO C++ conformant names like _fileno. These
497 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
498 might be nice to do as Microsoft suggest here too, although, unlike the secure
499 functions issue, there is presumably little or no benefit in this case.
501 =head2 Fix POSIX::access() and chdir() on Win32
503 These functions currently take no account of DACLs and therefore do not behave
504 correctly in situations where access is restricted by DACLs (as opposed to the
505 read-only attribute).
507 Furthermore, POSIX::access() behaves differently for directories having the
508 read-only attribute set depending on what CRT library is being used. For
509 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
510 such directories are not writable, whereas in fact all directories are writable
511 unless access is denied by DACLs. (In the case of directories, the read-only
512 attribute actually only means that the directory cannot be deleted.) This CRT
513 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
514 not actually be writable if access is indeed denied by DACLs).
516 For the chdir() issue, see ActiveState bug #74552:
517 http://bugs.activestate.com/show_bug.cgi?id=74552
519 Therefore, DACLs should be checked both for consistency across CRTs and for
522 (Note that perl's -w operator should not be modified to check DACLs. It has
523 been written so that it reflects the state of the read-only attribute, even
524 for directories (whatever CRT is being used), for symmetry with chmod().)
526 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
528 Maybe create a utility that checks after each libperl.a creation that
529 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
530 ever creep back to libperl.a.
532 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
534 Note, of course, that this will only tell whether B<your> platform
535 is using those naughty interfaces.
537 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
539 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
540 (4.1 onwards?) supports C<-fstack-protector>, both of which give
541 protection against various kinds of buffer overflow problems.
542 These should probably be used for compiling Perl whenever available,
543 Configure and/or hints files should be adjusted to probe for the
544 availability of these features and enable them as appropriate.
546 =head2 Arenas for GPs? For MAGIC?
548 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
549 It might be a speed or memory saving to change to using arenas. Or it might
550 not. It would need some suitable benchmarking first. In particular, C<GP>s
551 can probably be changed with minimal compatibility impact (probably nothing
552 outside of the core, or even outside of F<gv.c> allocates them), but they
553 probably aren't allocated/deallocated often enough for a speed saving. Whereas
554 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
555 more externally visible, so changing the rules here may bite external code.
559 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
560 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
561 sized bodies from the same actual arena, rather than maintaining one arena for
562 each. This could save 4-6K per thread, of memory no longer tied up in the
563 not-yet-allocated part of an arena.
566 =head1 Tasks that need a knowledge of XS
568 These tasks would need C knowledge, and roughly the level of knowledge of
569 the perl API that comes from writing modules that use XS to interface to
572 =head2 Write an XS cookbook
574 Create pod/perlxscookbook.pod with short, task-focused 'recipes' in XS that
575 demonstrate common tasks and good practices. (Some of these might be
576 extracted from perlguts.) The target audience should be XS novices, who need
577 more examples than perlguts but something less overwhelming than perlapi.
578 Recipes should provide "one pretty good way to do it" instead of TIMTOWTDI.
580 Rather than focusing on interfacing Perl to C libraries, such a cookbook
581 should probably focus on how to optimize Perl routines by re-writing them
582 in XS. This will likely be more motivating to those who mostly work in
583 Perl but are looking to take the next step into XS.
585 Deconstructing and explaining some simpler XS modules could be one way to
586 bootstrap a cookbook. (List::Util? Class::XSAccessor? Tree::Ternary_XS?)
587 Another option could be deconstructing the implementation of some simpler
590 =head2 Remove the use of SVs as temporaries in dump.c
592 F<dump.c> contains debugging routines to dump out the contains of perl data
593 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
594 B<uses> C<SV>s for its temporary buffers, which was a logical initial
595 implementation choice, as they provide ready made memory handling.
597 However, they also lead to a lot of confusion when it happens that what you're
598 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
599 a temporary scalar it can use for a temporary buffer. It's also not possible
600 to dump scalars before the interpreter is properly set up, such as during
601 ithreads cloning. It would be good to progressively replace the use of scalars
602 as string accumulation buffers with something much simpler, directly allocated
603 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
604 US-ASCII, so output character sets are not an issue.
606 Producing and proving an internal simple buffer allocation would make it easier
607 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
608 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
611 =head2 safely supporting POSIX SA_SIGINFO
613 Some years ago Jarkko supplied patches to provide support for the POSIX
614 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
616 Unfortunately, it only works with "unsafe" signals, because under safe
617 signals, by the time Perl gets to run the signal handler, the extra
618 information has been lost. Moreover, it's not easy to store it somewhere,
619 as you can't call mutexs, or do anything else fancy, from inside a signal
622 So it strikes me that we could provide safe SA_SIGINFO support
628 Provide global variables for two file descriptors
632 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
633 pipe, store the reader in one, the writer in the other
637 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
638 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
644 serialise signal number, C<struct siginfo_t> (or at least the parts we care
645 about) into a small auto char buff
649 C<write()> that (non-blocking) to the writer fd
655 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
656 to the current per-signal-number counts
660 if it writes 0%, assume the pipe is full. Flag the data as lost?
664 if it writes partially, croak a panic, as your OS is broken.
672 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
673 the pipe", read the data out, deserialise, build the Perl structures on
674 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
679 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
680 of running Perl code inside the signal handler context. (With all the dangers
681 of things like C<malloc> corruption that that currently offers us)
683 For more information see the thread starting with this message:
684 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
686 =head2 autovivification
688 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
690 This task is incremental - even a little bit of work on it will help.
692 =head2 Unicode in Filenames
694 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
695 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
696 system, truncate, unlink, utime, -X. All these could potentially accept
697 Unicode filenames either as input or output (and in the case of system
698 and qx Unicode in general, as input or output to/from the shell).
699 Whether a filesystem - an operating system pair understands Unicode in
702 Known combinations that have some level of understanding include
703 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
704 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
705 create Unicode filenames, what forms of Unicode are accepted and used
706 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
707 and so on, varies. Finding the right level of interfacing to Perl
708 requires some thought. Remember that an OS does not implicate a
711 (The Windows -C command flag "wide API support" has been at least
712 temporarily retired in 5.8.1, and the -C has been repurposed, see
715 Most probably the right way to do this would be this:
716 L</"Virtualize operating system access">.
718 =head2 Unicode in %ENV
720 Currently the %ENV entries are always byte strings.
721 See L</"Virtualize operating system access">.
723 =head2 Unicode and glob()
725 Currently glob patterns and filenames returned from File::Glob::glob()
726 are always byte strings. See L</"Virtualize operating system access">.
728 =head2 Unicode and lc/uc operators
730 Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
731 what the internal encoding of their argument is. That should not be the
732 case. Maybe add a pragma to switch behaviour.
734 =head2 use less 'memory'
736 Investigate trade offs to switch out perl's choices on memory usage.
737 Particularly perl should be able to give memory back.
739 This task is incremental - even a little bit of work on it will help.
741 =head2 Re-implement C<:unique> in a way that is actually thread-safe
743 The old implementation made bad assumptions on several levels. A good 90%
744 solution might be just to make C<:unique> work to share the string buffer
745 of SvPVs. That way large constant strings can be shared between ithreads,
746 such as the configuration information in F<Config>.
748 =head2 Make tainting consistent
750 Tainting would be easier to use if it didn't take documented shortcuts and
751 allow taint to "leak" everywhere within an expression.
753 =head2 readpipe(LIST)
755 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
756 running a shell. readpipe() (the function behind qx//) could be similarly
759 =head2 Audit the code for destruction ordering assumptions
763 /* Need to check SvMAGICAL, as during global destruction it may be that
764 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
765 is now part of the linked list of SV heads, rather than pointing to
766 the original body. */
767 /* FIXME - audit the code for other bugs like this one. */
769 adding the C<SvMAGICAL> check to
771 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
772 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
774 Go through the core and look for similar assumptions that SVs have particular
775 types, as all bets are off during global destruction.
777 =head2 Extend PerlIO and PerlIO::Scalar
779 PerlIO::Scalar doesn't know how to truncate(). Implementing this
780 would require extending the PerlIO vtable.
782 Similarly the PerlIO vtable doesn't know about formats (write()), or
783 about stat(), or chmod()/chown(), utime(), or flock().
785 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
788 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
789 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
792 See also L</"Virtualize operating system access">.
794 =head2 -C on the #! line
796 It should be possible to make -C work correctly if found on the #! line,
797 given that all perl command line options are strict ASCII, and -C changes
798 only the interpretation of non-ASCII characters, and not for the script file
799 handle. To make it work needs some investigation of the ordering of function
800 calls during startup, and (by implication) a bit of tweaking of that order.
802 =head2 Organize error messages
804 Perl's diagnostics (error messages, see L<perldiag>) could use
805 reorganizing and formalizing so that each error message has its
806 stable-for-all-eternity unique id, categorized by severity, type, and
807 subsystem. (The error messages would be listed in a datafile outside
808 of the Perl source code, and the source code would only refer to the
809 messages by the id.) This clean-up and regularizing should apply
810 for all croak() messages.
812 This would enable all sorts of things: easier translation/localization
813 of the messages (though please do keep in mind the caveats of
814 L<Locale::Maketext> about too straightforward approaches to
815 translation), filtering by severity, and instead of grepping for a
816 particular error message one could look for a stable error id. (Of
817 course, changing the error messages by default would break all the
818 existing software depending on some particular error message...)
820 This kind of functionality is known as I<message catalogs>. Look for
821 inspiration for example in the catgets() system, possibly even use it
822 if available-- but B<only> if available, all platforms will B<not>
825 For the really pure at heart, consider extending this item to cover
826 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
828 =head1 Tasks that need a knowledge of the interpreter
830 These tasks would need C knowledge, and knowledge of how the interpreter works,
831 or a willingness to learn.
833 =head2 truncate() prototype
835 The prototype of truncate() is currently C<$$>. It should probably
836 be C<*$> instead. (This is changed in F<opcode.pl>)
838 =head2 decapsulation of smart match argument
840 Currently C<$foo ~~ $object> will die with the message "Smart matching a
841 non-overloaded object breaks encapsulation". It would be nice to allow
842 to bypass this by using explictly the syntax C<$foo ~~ %$object> or
845 =head2 error reporting of [$a ; $b]
847 Using C<;> inside brackets is a syntax error, and we don't propose to change
848 that by giving it any meaning. However, it's not reported very helpfully:
850 $ perl -e '$a = [$b; $c];'
851 syntax error at -e line 1, near "$b;"
852 syntax error at -e line 1, near "$c]"
853 Execution of -e aborted due to compilation errors.
855 It should be possible to hook into the tokeniser or the lexer, so that when a
856 C<;> is parsed where it is not legal as a statement terminator (ie inside
857 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
858 I<';' isn't legal inside an expression - if you need multiple statements use a
859 do {...} block>. See the thread starting at
860 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
862 =head2 lexicals used only once
866 $ perl -we '$pie = 42'
867 Name "main::pie" used only once: possible typo at -e line 1.
871 $ perl -we 'my $pie = 42'
873 Logically all lexicals used only once should warn, if the user asks for
874 warnings. An unworked RT ticket (#5087) has been open for almost seven
875 years for this discrepancy.
879 The handling of Unicode is unclean in many places. For example, the regexp
880 engine matches in Unicode semantics whenever the string or the pattern is
881 flagged as UTF-8, but that should not be dependent on an internal storage
882 detail of the string. Likewise, case folding behaviour is dependent on the
883 UTF8 internal flag being on or off.
885 =head2 Properly Unicode safe tokeniser and pads.
887 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
888 variable names are stored in stashes as raw bytes, without the utf-8 flag
889 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
890 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
891 source filters. All this could be fixed.
893 =head2 state variable initialization in list context
895 Currently this is illegal:
897 state ($a, $b) = foo();
899 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
900 semantics, which is tricky to implement in Perl 5 as currently they produce
901 the same opcode trees. The Perl 6 design is firm, so it would be good to
902 implement the necessary code in Perl 5. There are comments in
903 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
904 constructions involving state variables.
906 =head2 Implement $value ~~ 0 .. $range
908 It would be nice to extend the syntax of the C<~~> operator to also
909 understand numeric (and maybe alphanumeric) ranges.
911 =head2 A does() built-in
913 Like ref(), only useful. It would call the C<DOES> method on objects; it
914 would also tell whether something can be dereferenced as an
915 array/hash/etc., or used as a regexp, etc.
916 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
918 =head2 Tied filehandles and write() don't mix
920 There is no method on tied filehandles to allow them to be called back by
923 =head2 Propagate compilation hints to the debugger
925 Currently a debugger started with -dE on the command-line doesn't see the
926 features enabled by -E. More generally hints (C<$^H> and C<%^H>) aren't
927 propagated to the debugger. Probably it would be a good thing to propagate
928 hints from the innermost non-C<DB::> scope: this would make code eval'ed
929 in the debugger see the features (and strictures, etc.) currently in
932 =head2 Attach/detach debugger from running program
934 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
935 program if you pass the process ID. It would be good to do this with the Perl
936 debugger on a running Perl program, although I'm not sure how it would be
937 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
939 =head2 LVALUE functions for lists
941 The old perltodo notes that lvalue functions don't work for list or hash
942 slices. This would be good to fix.
944 =head2 regexp optimiser optional
946 The regexp optimiser is not optional. It should configurable to be, to allow
947 its performance to be measured, and its bugs to be easily demonstrated.
949 =head2 delete &function
951 Allow to delete functions. One can already undef them, but they're still
954 =head2 C</w> regex modifier
956 That flag would enable to match whole words, and also to interpolate
957 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
959 do { local $"='|'; /\b(?:P)\b/ }
961 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
964 =head2 optional optimizer
966 Make the peephole optimizer optional. Currently it performs two tasks as
967 it walks the optree - genuine peephole optimisations, and necessary fixups of
968 ops. It would be good to find an efficient way to switch out the
969 optimisations whilst keeping the fixups.
971 =head2 You WANT *how* many
973 Currently contexts are void, scalar and list. split has a special mechanism in
974 place to pass in the number of return values wanted. It would be useful to
975 have a general mechanism for this, backwards compatible and little speed hit.
976 This would allow proposals such as short circuiting sort to be implemented
979 =head2 lexical aliases
981 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
983 =head2 entersub XS vs Perl
985 At the moment pp_entersub is huge, and has code to deal with entering both
986 perl and XS subroutines. Subroutine implementations rarely change between
987 perl and XS at run time, so investigate using 2 ops to enter subs (one for
988 XS, one for perl) and swap between if a sub is redefined.
992 Self-ties are currently illegal because they caused too many segfaults. Maybe
993 the causes of these could be tracked down and self-ties on all types
996 =head2 Optimize away @_
998 The old perltodo notes "Look at the "reification" code in C<av.c>".
1000 =head2 Virtualize operating system access
1002 Implement a set of "vtables" that virtualizes operating system access
1003 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
1004 least these interfaces should take SVs as "name" arguments instead of
1005 bare char pointers; probably the most flexible and extensible way
1006 would be for the Perl-facing interfaces to accept HVs. The system
1007 needs to be per-operating-system and per-file-system
1008 hookable/filterable, preferably both from XS and Perl level
1009 (L<perlport/"Files and Filesystems"> is good reading at this point,
1010 in fact, all of L<perlport> is.)
1012 This has actually already been implemented (but only for Win32),
1013 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
1014 variants go through a set of "vtables" for operating system access,
1015 non-Win32 systems currently go straight for the POSIX/UNIX-style
1016 system/library call. Similar system as for Win32 should be
1017 implemented for all platforms. The existing Win32 implementation
1018 probably does not need to survive alongside this proposed new
1019 implementation, the approaches could be merged.
1021 What would this give us? One often-asked-for feature this would
1022 enable is using Unicode for filenames, and other "names" like %ENV,
1023 usernames, hostnames, and so forth.
1024 (See L<perlunicode/"When Unicode Does Not Happen">.)
1026 But this kind of virtualization would also allow for things like
1027 virtual filesystems, virtual networks, and "sandboxes" (though as long
1028 as dynamic loading of random object code is allowed, not very safe
1029 sandboxes since external code of course know not of Perl's vtables).
1030 An example of a smaller "sandbox" is that this feature can be used to
1031 implement per-thread working directories: Win32 already does this.
1033 See also L</"Extend PerlIO and PerlIO::Scalar">.
1035 =head2 Investigate PADTMP hash pessimisation
1037 The peephole optimiser converts constants used for hash key lookups to shared
1038 hash key scalars. Under ithreads, something is undoing this work.
1039 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
1041 =head2 Store the current pad in the OP slab allocator
1044 I hope that I got that "current pad" part correct
1046 Currently we leak ops in various cases of parse failure. I suggested that we
1047 could solve this by always using the op slab allocator, and walking it to
1048 free ops. Dave comments that as some ops are already freed during optree
1049 creation one would have to mark which ops are freed, and not double free them
1050 when walking the slab. He notes that one problem with this is that for some ops
1051 you have to know which pad was current at the time of allocation, which does
1052 change. I suggested storing a pointer to the current pad in the memory allocated
1053 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1054 that this would work.
1056 =head2 repack the optree
1058 Repacking the optree after execution order is determined could allow
1059 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1060 filling. The slab allocator could be reused for this purpose. I think that
1061 the best way to do this is to make it an optional step just before the
1062 completed optree is attached to anything else, and to use the slab allocator
1063 unchanged, so that freeing ops is identical whether or not this step runs.
1064 Note that the slab allocator allocates ops downwards in memory, so one would
1065 have to actually "allocate" the ops in reverse-execution order to get them
1066 contiguous in memory in execution order.
1068 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1070 Note that running this copy, and then freeing all the old location ops would
1071 cause their slabs to be freed, which would eliminate possible memory wastage if
1072 the previous suggestion is implemented, and we swap slabs more frequently.
1074 =head2 eliminate incorrect line numbers in warnings
1082 } elsif ($undef == 0) {
1085 used to produce this output:
1087 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1088 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1090 where the line of the second warning was misreported - it should be line 5.
1091 Rafael fixed this - the problem arose because there was no nextstate OP
1092 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1093 reports that the currently executing line is line 4. The solution was to inject
1094 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1095 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1096 numbers became misreported. (Jenga!)
1098 The problem is more general than C<elsif> (although the C<elsif> case is the
1099 most common and the most confusing). Ideally this code
1109 would produce this output
1111 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1112 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1114 (rather than lines 4 and 5), but this would seem to require every OP to carry
1115 (at least) line number information.
1117 What might work is to have an optional line number in memory just before the
1118 BASEOP structure, with a flag bit in the op to say whether it's present.
1119 Initially during compile every OP would carry its line number. Then add a late
1120 pass to the optimiser (potentially combined with L</repack the optree>) which
1121 looks at the two ops on every edge of the graph of the execution path. If
1122 the line number changes, flags the destination OP with this information.
1123 Once all paths are traced, replace every op with the flag with a
1124 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1125 control on to the true op. All ops would then be replaced by variants that
1126 do not store the line number. (Which, logically, why it would work best in
1127 conjunction with L</repack the optree>, as that is already copying/reallocating
1130 (Although I should note that we're not certain that doing this for the general
1133 =head2 optimize tail-calls
1135 Tail-calls present an opportunity for broadly applicable optimization;
1136 anywhere that C<< return foo(...) >> is called, the outer return can
1137 be replaced by a goto, and foo will return directly to the outer
1138 caller, saving (conservatively) 25% of perl's call&return cost, which
1139 is relatively higher than in C. The scheme language is known to do
1140 this heavily. B::Concise provides good insight into where this
1141 optimization is possible, ie anywhere entersub,leavesub op-sequence
1144 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1146 Bottom line on this is probably a new pp_tailcall function which
1147 combines the code in pp_entersub, pp_leavesub. This should probably
1148 be done 1st in XS, and using B::Generate to patch the new OP into the
1153 Tasks that will get your name mentioned in the description of the "Highlights
1156 =head2 make ithreads more robust
1158 Generally make ithreads more robust. See also L</iCOW>
1160 This task is incremental - even a little bit of work on it will help, and
1161 will be greatly appreciated.
1163 One bit would be to write the missing code in sv.c:Perl_dirp_dup.
1165 Fix Perl_sv_dup, et al so that threads can return objects.
1169 Sarathy and Arthur have a proposal for an improved Copy On Write which
1170 specifically will be able to COW new ithreads. If this can be implemented
1171 it would be a good thing.
1173 =head2 (?{...}) closures in regexps
1175 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1177 =head2 A re-entrant regexp engine
1179 This will allow the use of a regex from inside (?{ }), (??{ }) and
1180 (?(?{ })|) constructs.
1182 =head2 Add class set operations to regexp engine
1184 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1186 demerphq has this on his todo list, but right at the bottom.
1189 =head1 Tasks for microperl
1192 [ Each and every one of these may be obsolete, but they were listed
1193 in the old Todo.micro file]
1196 =head2 make creating uconfig.sh automatic
1198 =head2 make creating Makefile.micro automatic
1200 =head2 do away with fork/exec/wait?
1202 (system, popen should be enough?)
1204 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1206 (uConfigure? :-) native datatype widths and endianness come to mind