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 Remove macperl references from tests
31 MacPerl is gone. The tests don't need to be there.
33 =head2 Remove duplication of test setup.
35 Schwern notes, that there's duplication of code - lots and lots of tests have
36 some variation on the big block of C<$Is_Foo> checks. We can safely put this
37 into a file, change it to build an C<%Is> hash and require it. Maybe just put
38 it into F<test.pl>. Throw in the handy tainting subroutines.
40 =head2 POD -E<gt> HTML conversion in the core still sucks
42 Which is crazy given just how simple POD purports to be, and how simple HTML
43 can be. It's not actually I<as> simple as it sounds, particularly with the
44 flexibility POD allows for C<=item>, but it would be good to improve the
45 visual appeal of the HTML generated, and to avoid it having any validation
46 errors. See also L</make HTML install work>, as the layout of installation tree
47 is needed to improve the cross-linking.
49 The addition of C<Pod::Simple> and its related modules may make this task
52 =head2 Make ExtUtils::ParseXS use strict;
54 F<lib/ExtUtils/ParseXS.pm> contains this line
56 # use strict; # One of these days...
58 Simply uncomment it, and fix all the resulting issues :-)
60 The more practical approach, to break the task down into manageable chunks, is
61 to work your way though the code from bottom to top, or if necessary adding
62 extra C<{ ... }> blocks, and turning on strict within them.
64 =head2 Parallel testing
66 (This probably impacts much more than the core: also the Test::Harness
67 and TAP::* modules on CPAN.)
69 All of the tests in F<t/> can now be run in parallel, if C<$ENV{TEST_JOBS}>
70 is set. However, tests within each directory in F<ext> and F<lib> are still
71 run in series, with directories run in parallel. This is an adequate
72 heuristic, but it might be possible to relax it further, and get more
73 throughput. Specifically, it would be good to audit all of F<lib/*.t>, and
74 make them use C<File::Temp>.
76 =head2 Make Schwern poorer
78 We should have tests for everything. When all the core's modules are tested,
79 Schwern has promised to donate to $500 to TPF. We may need volunteers to
80 hold him upside down and shake vigorously in order to actually extract the
83 =head2 Improve the coverage of the core tests
85 Use Devel::Cover to ascertain the core modules's test coverage, then add
86 tests that are currently missing.
90 A full test suite for the B module would be nice.
92 =head2 A decent benchmark
94 C<perlbench> seems impervious to any recent changes made to the perl core. It
95 would be useful to have a reasonable general benchmarking suite that roughly
96 represented what current perl programs do, and measurably reported whether
97 tweaks to the core improve, degrade or don't really affect performance, to
98 guide people attempting to optimise the guts of perl. Gisle would welcome
99 new tests for perlbench.
101 =head2 fix tainting bugs
103 Fix the bugs revealed by running the test suite with the C<-t> switch (via
104 C<make test.taintwarn>).
106 =head2 Dual life everything
108 As part of the "dists" plan, anything that doesn't belong in the smallest perl
109 distribution needs to be dual lifed. Anything else can be too. Figure out what
110 changes would be needed to package that module and its tests up for CPAN, and
111 do so. Test it with older perl releases, and fix the problems you find.
113 To make a minimal perl distribution, it's useful to look at
114 F<t/lib/commonsense.t>.
116 =head2 Bundle dual life modules in ext/
118 For maintenance (and branch merging) reasons, it would be useful to move
119 some architecture-independent dual-life modules from lib/ to ext/, if this
120 has no negative impact on the build of perl itself.
122 =head2 POSIX memory footprint
124 Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
125 various times worked to cut it down. There is probably still fat to cut out -
126 for example POSIX passes Exporter some very memory hungry data structures.
128 =head2 embed.pl/makedef.pl
130 There is a script F<embed.pl> that generates several header files to prefix
131 all of Perl's symbols in a consistent way, to provide some semblance of
132 namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
133 in F<interpvar.h>. Quite a few of the functions and variables
134 are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
135 doesn't understand the C macros, so the rules about which symbols are present
136 when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
137 It would be good to teach C<embed.pl> to understand the conditional
138 compilation, and hence remove the duplication, and the mistakes it has caused.
140 =head2 use strict; and AutoLoad
142 Currently if you write
145 use AutoLoader 'AUTOLOAD';
150 print join (' ', No, strict, here), "!\n";
153 then C<use strict;> isn't in force within the autoloaded subroutines. It would
154 be more consistent (and less surprising) to arrange for all lexical pragmas
155 in force at the __END__ block to be in force within each autoloaded subroutine.
157 There's a similar problem with SelfLoader.
159 =head2 profile installman
161 The F<installman> script is slow. All it is doing text processing, which we're
162 told is something Perl is good at. So it would be nice to know what it is doing
163 that is taking so much CPU, and where possible address it.
166 =head1 Tasks that need a little sysadmin-type knowledge
168 Or if you prefer, tasks that you would learn from, and broaden your skills
171 =head2 make HTML install work
173 There is an C<installhtml> target in the Makefile. It's marked as
174 "experimental". It would be good to get this tested, make it work reliably, and
175 remove the "experimental" tag. This would include
181 Checking that cross linking between various parts of the documentation works.
182 In particular that links work between the modules (files with POD in F<lib/>)
183 and the core documentation (files in F<pod/>)
187 Work out how to split C<perlfunc> into chunks, preferably one per function
188 group, preferably with general case code that could be used elsewhere.
189 Challenges here are correctly identifying the groups of functions that go
190 together, and making the right named external cross-links point to the right
191 page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
192 C<endservent>, two or more C<=items> giving the different parameter lists, such
195 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
196 =item substr EXPR,OFFSET,LENGTH
197 =item substr EXPR,OFFSET
199 and different parameter lists having different meanings. (eg C<select>)
203 =head2 compressed man pages
205 Be able to install them. This would probably need a configure test to see how
206 the system does compressed man pages (same directory/different directory?
207 same filename/different filename), as well as tweaking the F<installman> script
208 to compress as necessary.
210 =head2 Add a code coverage target to the Makefile
212 Make it easy for anyone to run Devel::Cover on the core's tests. The steps
213 to do this manually are roughly
219 do a normal C<Configure>, but include Devel::Cover as a module to install
220 (see F<INSTALL> for how to do this)
228 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
232 Process the resulting Devel::Cover database
236 This just give you the coverage of the F<.pm>s. To also get the C level
243 Additionally tell C<Configure> to use the appropriate C compiler flags for
250 (instead of C<make perl>)
254 After running the tests run C<gcov> to generate all the F<.gcov> files.
255 (Including down in the subdirectories of F<ext/>
259 (From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
260 to get their stats into the cover_db directory.
264 Then process the Devel::Cover database
268 It would be good to add a single switch to C<Configure> to specify that you
269 wanted to perform perl level coverage, and another to specify C level
270 coverage, and have C<Configure> and the F<Makefile> do all the right things
273 =head2 Make Config.pm cope with differences between built and installed perl
275 Quite often vendors ship a perl binary compiled with their (pay-for)
276 compilers. People install a free compiler, such as gcc. To work out how to
277 build extensions, Perl interrogates C<%Config>, so in this situation
278 C<%Config> describes compilers that aren't there, and extension building
279 fails. This forces people into choosing between re-compiling perl themselves
280 using the compiler they have, or only using modules that the vendor ships.
282 It would be good to find a way teach C<Config.pm> about the installation setup,
283 possibly involving probing at install time or later, so that the C<%Config> in
284 a binary distribution better describes the installed machine, when the
285 installed machine differs from the build machine in some significant way.
287 =head2 linker specification files
289 Some platforms mandate that you provide a list of a shared library's external
290 symbols to the linker, so the core already has the infrastructure in place to
291 do this for generating shared perl libraries. My understanding is that the
292 GNU toolchain can accept an optional linker specification file, and restrict
293 visibility just to symbols declared in that file. It would be good to extend
294 F<makedef.pl> to support this format, and to provide a means within
295 C<Configure> to enable it. This would allow Unix users to test that the
296 export list is correct, and to build a perl that does not pollute the global
297 namespace with private symbols.
299 =head2 Cross-compile support
301 Currently C<Configure> understands C<-Dusecrosscompile> option. This option
302 arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
303 assumed then to be copied to TARGET machine and used as a replacement of full
306 This could be done little differently. Namely C<miniperl> should be built for
307 HOST and then full C<perl> with extensions should be compiled for TARGET.
308 This, however, might require extra trickery for %Config: we have one config
309 first for HOST and then another for TARGET. Tools like MakeMaker will be
310 mightily confused. Having around two different types of executables and
311 libraries (HOST and TARGET) makes life interesting for Makefiles and
312 shell (and Perl) scripts. There is $Config{run}, normally empty, which
313 can be used as an execution wrapper. Also note that in some
314 cross-compilation/execution environments the HOST and the TARGET do
315 not see the same filesystem(s), the $Config{run} may need to do some
316 file/directory copying back and forth.
320 Make F<pod/roffitall> be updated by F<pod/buildtoc>.
322 =head2 Split "linker" from "compiler"
324 Right now, Configure probes for two commands, and sets two variables:
328 =item * C<cc> (in F<cc.U>)
330 This variable holds the name of a command to execute a C compiler which
331 can resolve multiple global references that happen to have the same
332 name. Usual values are F<cc> and F<gcc>.
333 Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
335 =item * C<ld> (in F<dlsrc.U>)
337 This variable indicates the program to be used to link
338 libraries for dynamic loading. On some systems, it is F<ld>.
339 On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
340 the hint file setting.
344 There is an implicit historical assumption from around Perl5.000alpha
345 something, that C<$cc> is also the correct command for linking object files
346 together to make an executable. This may be true on Unix, but it's not true
347 on other platforms, and there are a maze of work arounds in other places (such
348 as F<Makefile.SH>) to cope with this.
350 Ideally, we should create a new variable to hold the name of the executable
351 linker program, probe for it in F<Configure>, and centralise all the special
352 case logic there or in hints files.
354 A small bikeshed issue remains - what to call it, given that C<$ld> is already
355 taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
356 for creating dynamically-loadable modules) and C<$link> could be confused with
357 the Unix command line executable of the same name, which does something
358 completely different. Andy Dougherty makes the counter argument "In parrot, I
359 tried to call the command used to link object files and libraries into an
360 executable F<link>, since that's what my vaguely-remembered DOS and VMS
361 experience suggested. I don't think any real confusion has ensued, so it's
362 probably a reasonable name for perl5 to use."
364 "Alas, I've always worried that introducing it would make things worse,
365 since now the module building utilities would have to look for
366 C<$Config{link}> and institute a fall-back plan if it weren't found."
367 Although I can see that as confusing, given that C<$Config{d_link}> is true
368 when (hard) links are available.
370 =head2 Configure Windows using PowerShell
372 Currently, Windows uses hard-coded config files based to build the
373 config.h for compiling Perl. Makefiles are also hard-coded and need to be
374 hand edited prior to building Perl. While this makes it easy to create a perl.exe
375 that works across multiple Windows versions, being able to accurately
376 configure a perl.exe for a specific Windows versions and VS C++ would be
377 a nice enhancement. With PowerShell available on Windows XP and up, this
378 may now be possible. Step 1 might be to investigate whether this is possible
379 and use this to clean up our current makefile situation. Step 2 would be to
380 see if there would be a way to use our existing metaconfig units to configure a
381 Windows Perl or whether we go in a separate direction and make it so. Of
382 course, we all know what step 3 is.
384 =head2 decouple -g and -DDEBUGGING
386 Currently F<Configure> automatically adds C<-DDEBUGGING> to the C compiler
387 flags if it spots C<-g> in the optimiser flags. The pre-processor directive
388 C<DEBUGGING> enables F<perl>'s command line <-D> options, but in the process
389 makes F<perl> slower. It would be good to disentangle this logic, so that
390 C-level debugging with C<-g> and Perl level debugging with C<-D> can easily
391 be enabled independently.
393 =head1 Tasks that need a little C knowledge
395 These tasks would need a little C knowledge, but don't need any specific
396 background or experience with XS, or how the Perl interpreter works
398 =head2 Weed out needless PERL_UNUSED_ARG
400 The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
401 unused arguments. Often the arguments can't be removed, as there is an
402 external constraint that determines the prototype of the function, so this
403 approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
404 could be removed. Specifically
410 The prototypes of (nearly all) static functions can be changed
414 Unused arguments generated by short cut macros are wasteful - the short cut
415 macro used can be changed.
419 =head2 Modernize the order of directories in @INC
421 The way @INC is laid out by default, one cannot upgrade core (dual-life)
422 modules without overwriting files. This causes problems for binary
423 package builders. One possible proposal is laid out in this
425 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
429 Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
430 On these systems, it might be the default compilation mode, and there
431 is currently no guarantee that passing no use64bitall option to the
432 Configure process will build a 32bit perl. Implementing -Duse32bit*
433 options would be nice for perl 5.12.
435 =head2 Profile Perl - am I hot or not?
437 The Perl source code is stable enough that it makes sense to profile it,
438 identify and optimise the hotspots. It would be good to measure the
439 performance of the Perl interpreter using free tools such as cachegrind,
440 gprof, and dtrace, and work to reduce the bottlenecks they reveal.
442 As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
443 the ops that are most commonly used. The idea is that by grouping them, their
444 object code will be adjacent in the executable, so they have a greater chance
445 of already being in the CPU cache (or swapped in) due to being near another op
448 Except that it's not clear if these really are the most commonly used ops. So
449 as part of exercising your skills with coverage and profiling tools you might
450 want to determine what ops I<really> are the most commonly used. And in turn
451 suggest evictions and promotions to achieve a better F<pp_hot.c>.
453 One piece of Perl code that might make a good testbed is F<installman>.
455 =head2 Allocate OPs from arenas
457 Currently all new OP structures are individually malloc()ed and free()d.
458 All C<malloc> implementations have space overheads, and are now as fast as
459 custom allocates so it would both use less memory and less CPU to allocate
460 the various OP structures from arenas. The SV arena code can probably be
463 Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
464 Perl_Slab_alloc() to pack optrees into a contiguous block, which is
465 probably superior to the use of OP arenas, esp. from a cache locality
466 standpoint. See L<Profile Perl - am I hot or not?>.
468 =head2 Improve win32/wince.c
470 Currently, numerous functions look virtually, if not completely,
471 identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
474 =head2 Use secure CRT functions when building with VC8 on Win32
476 Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
477 that they were "unsafe" and introduced differently named secure versions of
478 them as replacements, e.g. instead of writing
480 FILE* f = fopen(__FILE__, "r");
485 errno_t err = fopen_s(&f, __FILE__, "r");
487 Currently, the warnings about these deprecations have been disabled by adding
488 -D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
489 warning suppressant and actually make use of the new secure CRT functions.
491 There is also a similar issue with POSIX CRT function names like fileno having
492 been deprecated in favour of ISO C++ conformant names like _fileno. These
493 warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
494 might be nice to do as Microsoft suggest here too, although, unlike the secure
495 functions issue, there is presumably little or no benefit in this case.
497 =head2 Fix POSIX::access() and chdir() on Win32
499 These functions currently take no account of DACLs and therefore do not behave
500 correctly in situations where access is restricted by DACLs (as opposed to the
501 read-only attribute).
503 Furthermore, POSIX::access() behaves differently for directories having the
504 read-only attribute set depending on what CRT library is being used. For
505 example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
506 such directories are not writable, whereas in fact all directories are writable
507 unless access is denied by DACLs. (In the case of directories, the read-only
508 attribute actually only means that the directory cannot be deleted.) This CRT
509 bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
510 not actually be writable if access is indeed denied by DACLs).
512 For the chdir() issue, see ActiveState bug #74552:
513 http://bugs.activestate.com/show_bug.cgi?id=74552
515 Therefore, DACLs should be checked both for consistency across CRTs and for
518 (Note that perl's -w operator should not be modified to check DACLs. It has
519 been written so that it reflects the state of the read-only attribute, even
520 for directories (whatever CRT is being used), for symmetry with chmod().)
522 =head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
524 Maybe create a utility that checks after each libperl.a creation that
525 none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
526 ever creep back to libperl.a.
528 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
530 Note, of course, that this will only tell whether B<your> platform
531 is using those naughty interfaces.
533 =head2 -D_FORTIFY_SOURCE=2, -fstack-protector
535 Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
536 (4.1 onwards?) supports C<-fstack-protector>, both of which give
537 protection against various kinds of buffer overflow problems.
538 These should probably be used for compiling Perl whenever available,
539 Configure and/or hints files should be adjusted to probe for the
540 availability of these features and enable them as appropriate.
542 =head2 Arenas for GPs? For MAGIC?
544 C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
545 It might be a speed or memory saving to change to using arenas. Or it might
546 not. It would need some suitable benchmarking first. In particular, C<GP>s
547 can probably be changed with minimal compatibility impact (probably nothing
548 outside of the core, or even outside of F<gv.c> allocates them), but they
549 probably aren't allocated/deallocated often enough for a speed saving. Whereas
550 C<MAGIC> is allocated/deallocated more often, but in turn, is also something
551 more externally visible, so changing the rules here may bite external code.
555 Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
556 PVHV, and PVCV and PVFM. It should be possible to allocate and return same
557 sized bodies from the same actual arena, rather than maintaining one arena for
558 each. This could save 4-6K per thread, of memory no longer tied up in the
559 not-yet-allocated part of an arena.
562 =head1 Tasks that need a knowledge of XS
564 These tasks would need C knowledge, and roughly the level of knowledge of
565 the perl API that comes from writing modules that use XS to interface to
568 =head2 Remove the use of SVs as temporaries in dump.c
570 F<dump.c> contains debugging routines to dump out the contains of perl data
571 structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
572 B<uses> C<SV>s for its temporary buffers, which was a logical initial
573 implementation choice, as they provide ready made memory handling.
575 However, they also lead to a lot of confusion when it happens that what you're
576 trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
577 a temporary scalar it can use for a temporary buffer. It's also not possible
578 to dump scalars before the interpreter is properly set up, such as during
579 ithreads cloning. It would be good to progressively replace the use of scalars
580 as string accumulation buffers with something much simpler, directly allocated
581 by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
582 US-ASCII, so output character sets are not an issue.
584 Producing and proving an internal simple buffer allocation would make it easier
585 to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
586 B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
589 =head2 safely supporting POSIX SA_SIGINFO
591 Some years ago Jarkko supplied patches to provide support for the POSIX
592 SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
594 Unfortunately, it only works with "unsafe" signals, because under safe
595 signals, by the time Perl gets to run the signal handler, the extra
596 information has been lost. Moreover, it's not easy to store it somewhere,
597 as you can't call mutexs, or do anything else fancy, from inside a signal
600 So it strikes me that we could provide safe SA_SIGINFO support
606 Provide global variables for two file descriptors
610 When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
611 pipe, store the reader in one, the writer in the other
615 In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
616 the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
622 serialise signal number, C<struct siginfo_t> (or at least the parts we care
623 about) into a small auto char buff
627 C<write()> that (non-blocking) to the writer fd
633 if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
634 to the current per-signal-number counts
638 if it writes 0%, assume the pipe is full. Flag the data as lost?
642 if it writes partially, croak a panic, as your OS is broken.
650 in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
651 the pipe", read the data out, deserialise, build the Perl structures on
652 the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
657 I think that this gets us decent C<SA_SIGINFO> support, without the current risk
658 of running Perl code inside the signal handler context. (With all the dangers
659 of things like C<malloc> corruption that that currently offers us)
661 For more information see the thread starting with this message:
662 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
664 =head2 autovivification
666 Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
668 This task is incremental - even a little bit of work on it will help.
670 =head2 Unicode in Filenames
672 chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
673 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
674 system, truncate, unlink, utime, -X. All these could potentially accept
675 Unicode filenames either as input or output (and in the case of system
676 and qx Unicode in general, as input or output to/from the shell).
677 Whether a filesystem - an operating system pair understands Unicode in
680 Known combinations that have some level of understanding include
681 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
682 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
683 create Unicode filenames, what forms of Unicode are accepted and used
684 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
685 and so on, varies. Finding the right level of interfacing to Perl
686 requires some thought. Remember that an OS does not implicate a
689 (The Windows -C command flag "wide API support" has been at least
690 temporarily retired in 5.8.1, and the -C has been repurposed, see
693 Most probably the right way to do this would be this:
694 L</"Virtualize operating system access">.
696 =head2 Unicode in %ENV
698 Currently the %ENV entries are always byte strings.
699 See L</"Virtualize operating system access">.
701 =head2 Unicode and glob()
703 Currently glob patterns and filenames returned from File::Glob::glob()
704 are always byte strings. See L</"Virtualize operating system access">.
706 =head2 Unicode and lc/uc operators
708 Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
709 what the internal encoding of their argument is. That should not be the
710 case. Maybe add a pragma to switch behaviour.
712 =head2 use less 'memory'
714 Investigate trade offs to switch out perl's choices on memory usage.
715 Particularly perl should be able to give memory back.
717 This task is incremental - even a little bit of work on it will help.
719 =head2 Re-implement C<:unique> in a way that is actually thread-safe
721 The old implementation made bad assumptions on several levels. A good 90%
722 solution might be just to make C<:unique> work to share the string buffer
723 of SvPVs. That way large constant strings can be shared between ithreads,
724 such as the configuration information in F<Config>.
726 =head2 Make tainting consistent
728 Tainting would be easier to use if it didn't take documented shortcuts and
729 allow taint to "leak" everywhere within an expression.
731 =head2 readpipe(LIST)
733 system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
734 running a shell. readpipe() (the function behind qx//) could be similarly
737 =head2 Audit the code for destruction ordering assumptions
741 /* Need to check SvMAGICAL, as during global destruction it may be that
742 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
743 is now part of the linked list of SV heads, rather than pointing to
744 the original body. */
745 /* FIXME - audit the code for other bugs like this one. */
747 adding the C<SvMAGICAL> check to
749 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
750 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
752 Go through the core and look for similar assumptions that SVs have particular
753 types, as all bets are off during global destruction.
755 =head2 Extend PerlIO and PerlIO::Scalar
757 PerlIO::Scalar doesn't know how to truncate(). Implementing this
758 would require extending the PerlIO vtable.
760 Similarly the PerlIO vtable doesn't know about formats (write()), or
761 about stat(), or chmod()/chown(), utime(), or flock().
763 (For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
766 PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
767 opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
770 See also L</"Virtualize operating system access">.
772 =head2 -C on the #! line
774 It should be possible to make -C work correctly if found on the #! line,
775 given that all perl command line options are strict ASCII, and -C changes
776 only the interpretation of non-ASCII characters, and not for the script file
777 handle. To make it work needs some investigation of the ordering of function
778 calls during startup, and (by implication) a bit of tweaking of that order.
780 =head2 Organize error messages
782 Perl's diagnostics (error messages, see L<perldiag>) could use
783 reorganizing and formalizing so that each error message has its
784 stable-for-all-eternity unique id, categorized by severity, type, and
785 subsystem. (The error messages would be listed in a datafile outside
786 of the Perl source code, and the source code would only refer to the
787 messages by the id.) This clean-up and regularizing should apply
788 for all croak() messages.
790 This would enable all sorts of things: easier translation/localization
791 of the messages (though please do keep in mind the caveats of
792 L<Locale::Maketext> about too straightforward approaches to
793 translation), filtering by severity, and instead of grepping for a
794 particular error message one could look for a stable error id. (Of
795 course, changing the error messages by default would break all the
796 existing software depending on some particular error message...)
798 This kind of functionality is known as I<message catalogs>. Look for
799 inspiration for example in the catgets() system, possibly even use it
800 if available-- but B<only> if available, all platforms will B<not>
803 For the really pure at heart, consider extending this item to cover
804 also the warning messages (see L<perllexwarn>, C<warnings.pl>).
806 =head1 Tasks that need a knowledge of the interpreter
808 These tasks would need C knowledge, and knowledge of how the interpreter works,
809 or a willingness to learn.
811 =head2 truncate() prototype
813 The prototype of truncate() is currently C<$$>. It should probably
814 be C<*$> instead. (This is changed in F<opcode.pl>)
816 =head2 decapsulation of smart match argument
818 Currently C<$foo ~~ $object> will die with the message "Smart matching a
819 non-overloaded object breaks encapsulation". It would be nice to allow
820 to bypass this by using explictly the syntax C<$foo ~~ %$object> or
823 =head2 error reporting of [$a ; $b]
825 Using C<;> inside brackets is a syntax error, and we don't propose to change
826 that by giving it any meaning. However, it's not reported very helpfully:
828 $ perl -e '$a = [$b; $c];'
829 syntax error at -e line 1, near "$b;"
830 syntax error at -e line 1, near "$c]"
831 Execution of -e aborted due to compilation errors.
833 It should be possible to hook into the tokeniser or the lexer, so that when a
834 C<;> is parsed where it is not legal as a statement terminator (ie inside
835 C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
836 I<';' isn't legal inside an expression - if you need multiple statements use a
837 do {...} block>. See the thread starting at
838 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
840 =head2 lexicals used only once
844 $ perl -we '$pie = 42'
845 Name "main::pie" used only once: possible typo at -e line 1.
849 $ perl -we 'my $pie = 42'
851 Logically all lexicals used only once should warn, if the user asks for
852 warnings. An unworked RT ticket (#5087) has been open for almost seven
853 years for this discrepancy.
857 The handling of Unicode is unclean in many places. For example, the regexp
858 engine matches in Unicode semantics whenever the string or the pattern is
859 flagged as UTF-8, but that should not be dependent on an internal storage
860 detail of the string. Likewise, case folding behaviour is dependent on the
861 UTF8 internal flag being on or off.
863 =head2 Properly Unicode safe tokeniser and pads.
865 The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
866 variable names are stored in stashes as raw bytes, without the utf-8 flag
867 set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
868 tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
869 source filters. All this could be fixed.
871 =head2 state variable initialization in list context
873 Currently this is illegal:
875 state ($a, $b) = foo();
877 In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
878 semantics, which is tricky to implement in Perl 5 as currently they produce
879 the same opcode trees. The Perl 6 design is firm, so it would be good to
880 implement the necessary code in Perl 5. There are comments in
881 C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
882 constructions involving state variables.
884 =head2 Implement $value ~~ 0 .. $range
886 It would be nice to extend the syntax of the C<~~> operator to also
887 understand numeric (and maybe alphanumeric) ranges.
889 =head2 A does() built-in
891 Like ref(), only useful. It would call the C<DOES> method on objects; it
892 would also tell whether something can be dereferenced as an
893 array/hash/etc., or used as a regexp, etc.
894 L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
896 =head2 Tied filehandles and write() don't mix
898 There is no method on tied filehandles to allow them to be called back by
901 =head2 Attach/detach debugger from running program
903 The old perltodo notes "With C<gdb>, you can attach the debugger to a running
904 program if you pass the process ID. It would be good to do this with the Perl
905 debugger on a running Perl program, although I'm not sure how it would be
906 done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
908 =head2 LVALUE functions for lists
910 The old perltodo notes that lvalue functions don't work for list or hash
911 slices. This would be good to fix.
913 =head2 regexp optimiser optional
915 The regexp optimiser is not optional. It should configurable to be, to allow
916 its performance to be measured, and its bugs to be easily demonstrated.
918 =head2 delete &function
920 Allow to delete functions. One can already undef them, but they're still
923 =head2 C</w> regex modifier
925 That flag would enable to match whole words, and also to interpolate
926 arrays as alternations. With it, C</P/w> would be roughly equivalent to:
928 do { local $"='|'; /\b(?:P)\b/ }
930 See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
933 =head2 optional optimizer
935 Make the peephole optimizer optional. Currently it performs two tasks as
936 it walks the optree - genuine peephole optimisations, and necessary fixups of
937 ops. It would be good to find an efficient way to switch out the
938 optimisations whilst keeping the fixups.
940 =head2 You WANT *how* many
942 Currently contexts are void, scalar and list. split has a special mechanism in
943 place to pass in the number of return values wanted. It would be useful to
944 have a general mechanism for this, backwards compatible and little speed hit.
945 This would allow proposals such as short circuiting sort to be implemented
948 =head2 lexical aliases
950 Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
952 =head2 entersub XS vs Perl
954 At the moment pp_entersub is huge, and has code to deal with entering both
955 perl and XS subroutines. Subroutine implementations rarely change between
956 perl and XS at run time, so investigate using 2 ops to enter subs (one for
957 XS, one for perl) and swap between if a sub is redefined.
961 Self-ties are currently illegal because they caused too many segfaults. Maybe
962 the causes of these could be tracked down and self-ties on all types
965 =head2 Optimize away @_
967 The old perltodo notes "Look at the "reification" code in C<av.c>".
969 =head2 Virtualize operating system access
971 Implement a set of "vtables" that virtualizes operating system access
972 (open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
973 least these interfaces should take SVs as "name" arguments instead of
974 bare char pointers; probably the most flexible and extensible way
975 would be for the Perl-facing interfaces to accept HVs. The system
976 needs to be per-operating-system and per-file-system
977 hookable/filterable, preferably both from XS and Perl level
978 (L<perlport/"Files and Filesystems"> is good reading at this point,
979 in fact, all of L<perlport> is.)
981 This has actually already been implemented (but only for Win32),
982 take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
983 variants go through a set of "vtables" for operating system access,
984 non-Win32 systems currently go straight for the POSIX/UNIX-style
985 system/library call. Similar system as for Win32 should be
986 implemented for all platforms. The existing Win32 implementation
987 probably does not need to survive alongside this proposed new
988 implementation, the approaches could be merged.
990 What would this give us? One often-asked-for feature this would
991 enable is using Unicode for filenames, and other "names" like %ENV,
992 usernames, hostnames, and so forth.
993 (See L<perlunicode/"When Unicode Does Not Happen">.)
995 But this kind of virtualization would also allow for things like
996 virtual filesystems, virtual networks, and "sandboxes" (though as long
997 as dynamic loading of random object code is allowed, not very safe
998 sandboxes since external code of course know not of Perl's vtables).
999 An example of a smaller "sandbox" is that this feature can be used to
1000 implement per-thread working directories: Win32 already does this.
1002 See also L</"Extend PerlIO and PerlIO::Scalar">.
1004 =head2 Investigate PADTMP hash pessimisation
1006 The peephole optimiser converts constants used for hash key lookups to shared
1007 hash key scalars. Under ithreads, something is undoing this work.
1008 See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
1010 =head2 Store the current pad in the OP slab allocator
1013 I hope that I got that "current pad" part correct
1015 Currently we leak ops in various cases of parse failure. I suggested that we
1016 could solve this by always using the op slab allocator, and walking it to
1017 free ops. Dave comments that as some ops are already freed during optree
1018 creation one would have to mark which ops are freed, and not double free them
1019 when walking the slab. He notes that one problem with this is that for some ops
1020 you have to know which pad was current at the time of allocation, which does
1021 change. I suggested storing a pointer to the current pad in the memory allocated
1022 for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1023 that this would work.
1025 =head2 repack the optree
1027 Repacking the optree after execution order is determined could allow
1028 removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1029 filling. The slab allocator could be reused for this purpose. I think that
1030 the best way to do this is to make it an optional step just before the
1031 completed optree is attached to anything else, and to use the slab allocator
1032 unchanged, so that freeing ops is identical whether or not this step runs.
1033 Note that the slab allocator allocates ops downwards in memory, so one would
1034 have to actually "allocate" the ops in reverse-execution order to get them
1035 contiguous in memory in execution order.
1037 See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1039 Note that running this copy, and then freeing all the old location ops would
1040 cause their slabs to be freed, which would eliminate possible memory wastage if
1041 the previous suggestion is implemented, and we swap slabs more frequently.
1043 =head2 eliminate incorrect line numbers in warnings
1051 } elsif ($undef == 0) {
1054 used to produce this output:
1056 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1057 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1059 where the line of the second warning was misreported - it should be line 5.
1060 Rafael fixed this - the problem arose because there was no nextstate OP
1061 between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1062 reports that the currently executing line is line 4. The solution was to inject
1063 a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1064 OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1065 numbers became misreported. (Jenga!)
1067 The problem is more general than C<elsif> (although the C<elsif> case is the
1068 most common and the most confusing). Ideally this code
1078 would produce this output
1080 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1081 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1083 (rather than lines 4 and 5), but this would seem to require every OP to carry
1084 (at least) line number information.
1086 What might work is to have an optional line number in memory just before the
1087 BASEOP structure, with a flag bit in the op to say whether it's present.
1088 Initially during compile every OP would carry its line number. Then add a late
1089 pass to the optimiser (potentially combined with L</repack the optree>) which
1090 looks at the two ops on every edge of the graph of the execution path. If
1091 the line number changes, flags the destination OP with this information.
1092 Once all paths are traced, replace every op with the flag with a
1093 nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1094 control on to the true op. All ops would then be replaced by variants that
1095 do not store the line number. (Which, logically, why it would work best in
1096 conjunction with L</repack the optree>, as that is already copying/reallocating
1099 (Although I should note that we're not certain that doing this for the general
1102 =head2 optimize tail-calls
1104 Tail-calls present an opportunity for broadly applicable optimization;
1105 anywhere that C<< return foo(...) >> is called, the outer return can
1106 be replaced by a goto, and foo will return directly to the outer
1107 caller, saving (conservatively) 25% of perl's call&return cost, which
1108 is relatively higher than in C. The scheme language is known to do
1109 this heavily. B::Concise provides good insight into where this
1110 optimization is possible, ie anywhere entersub,leavesub op-sequence
1113 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1115 Bottom line on this is probably a new pp_tailcall function which
1116 combines the code in pp_entersub, pp_leavesub. This should probably
1117 be done 1st in XS, and using B::Generate to patch the new OP into the
1122 Tasks that will get your name mentioned in the description of the "Highlights
1125 =head2 make ithreads more robust
1127 Generally make ithreads more robust. See also L</iCOW>
1129 This task is incremental - even a little bit of work on it will help, and
1130 will be greatly appreciated.
1132 One bit would be to write the missing code in sv.c:Perl_dirp_dup.
1134 Fix Perl_sv_dup, et al so that threads can return objects.
1138 Sarathy and Arthur have a proposal for an improved Copy On Write which
1139 specifically will be able to COW new ithreads. If this can be implemented
1140 it would be a good thing.
1142 =head2 (?{...}) closures in regexps
1144 Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1146 =head2 A re-entrant regexp engine
1148 This will allow the use of a regex from inside (?{ }), (??{ }) and
1149 (?(?{ })|) constructs.
1151 =head2 Add class set operations to regexp engine
1153 Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1155 demerphq has this on his todo list, but right at the bottom.
1158 =head1 Tasks for microperl
1161 [ Each and every one of these may be obsolete, but they were listed
1162 in the old Todo.micro file]
1165 =head2 make creating uconfig.sh automatic
1167 =head2 make creating Makefile.micro automatic
1169 =head2 do away with fork/exec/wait?
1171 (system, popen should be enough?)
1173 =head2 some of the uconfig.sh really needs to be probed (using cc) in buildtime:
1175 (uConfigure? :-) native datatype widths and endianness come to mind