This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Todo: decapsulation of smart match argument
[perl5.git] / pod / perltodo.pod
CommitLineData
7711098a
GS
1=head1 NAME
2
3perltodo - Perl TO-DO List
4
5=head1 DESCRIPTION
e50bb9a1 6
049aabcb
NC
7This is a list of wishes for Perl. The most up to date version of this file
8is at http://perl5.git.perl.org/perl.git/blob_plain/HEAD:/pod/perltodo.pod
9
10The tasks we think are smaller or easier are listed first. Anyone is welcome
11to work on any of these, but it's a good idea to first contact
12I<perl5-porters@perl.org> to avoid duplication of effort, and to learn from
13any previous attempts. By all means contact a pumpking privately first if you
14prefer.
e50bb9a1 15
0bdfc961
NC
16Whilst patches to make the list shorter are most welcome, ideas to add to
17the list are also encouraged. Check the perl5-porters archives for past
18ideas, and any discussion about them. One set of archives may be found at:
e50bb9a1 19
0bdfc961 20 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
938c8732 21
617eabfa
NC
22What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
23not, but if your patch is incorporated, then we'll add your name to the
24F<AUTHORS> file, which ships in the official distribution. How many other
25programming languages offer you 1 line of immortality?
938c8732 26
0bdfc961 27=head1 Tasks that only need Perl knowledge
e50bb9a1 28
5a176cbc
NC
29=head2 Remove duplication of test setup.
30
31Schwern notes, that there's duplication of code - lots and lots of tests have
32some variation on the big block of C<$Is_Foo> checks. We can safely put this
33into a file, change it to build an C<%Is> hash and require it. Maybe just put
34it into F<test.pl>. Throw in the handy tainting subroutines.
35
87a942b1 36=head2 POD -E<gt> HTML conversion in the core still sucks
e50bb9a1 37
938c8732 38Which is crazy given just how simple POD purports to be, and how simple HTML
adebf063
NC
39can be. It's not actually I<as> simple as it sounds, particularly with the
40flexibility POD allows for C<=item>, but it would be good to improve the
41visual appeal of the HTML generated, and to avoid it having any validation
42errors. See also L</make HTML install work>, as the layout of installation tree
43is needed to improve the cross-linking.
938c8732 44
dc0fb092
SP
45The addition of C<Pod::Simple> and its related modules may make this task
46easier to complete.
47
0befdfba
NC
48=head2 Make ExtUtils::ParseXS use strict;
49
50F<lib/ExtUtils/ParseXS.pm> contains this line
51
52 # use strict; # One of these days...
53
54Simply uncomment it, and fix all the resulting issues :-)
55
56The more practical approach, to break the task down into manageable chunks, is
57to work your way though the code from bottom to top, or if necessary adding
58extra C<{ ... }> blocks, and turning on strict within them.
59
aa237293
NC
60=head2 Parallel testing
61
b2e2905c 62(This probably impacts much more than the core: also the Test::Harness
02f21748
RGS
63and TAP::* modules on CPAN.)
64
c707cc00
NC
65All of the tests in F<t/> can now be run in parallel, if C<$ENV{TEST_JOBS}>
66is set. However, tests within each directory in F<ext> and F<lib> are still
67run in series, with directories run in parallel. This is an adequate
68heuristic, but it might be possible to relax it further, and get more
69throughput. Specifically, it would be good to audit all of F<lib/*.t>, and
70make them use C<File::Temp>.
aa237293 71
0bdfc961 72=head2 Make Schwern poorer
e50bb9a1 73
613bd4f7 74We should have tests for everything. When all the core's modules are tested,
0bdfc961
NC
75Schwern has promised to donate to $500 to TPF. We may need volunteers to
76hold him upside down and shake vigorously in order to actually extract the
77cash.
3958b146 78
0bdfc961 79=head2 Improve the coverage of the core tests
e50bb9a1 80
02f21748
RGS
81Use Devel::Cover to ascertain the core modules's test coverage, then add
82tests that are currently missing.
30222c0f 83
0bdfc961 84=head2 test B
e50bb9a1 85
0bdfc961 86A full test suite for the B module would be nice.
e50bb9a1 87
0bdfc961 88=head2 A decent benchmark
e50bb9a1 89
617eabfa 90C<perlbench> seems impervious to any recent changes made to the perl core. It
0bdfc961
NC
91would be useful to have a reasonable general benchmarking suite that roughly
92represented what current perl programs do, and measurably reported whether
93tweaks to the core improve, degrade or don't really affect performance, to
94guide people attempting to optimise the guts of perl. Gisle would welcome
95new tests for perlbench.
6168cf99 96
0bdfc961 97=head2 fix tainting bugs
6168cf99 98
0bdfc961
NC
99Fix the bugs revealed by running the test suite with the C<-t> switch (via
100C<make test.taintwarn>).
e50bb9a1 101
0bdfc961 102=head2 Dual life everything
e50bb9a1 103
0bdfc961
NC
104As part of the "dists" plan, anything that doesn't belong in the smallest perl
105distribution needs to be dual lifed. Anything else can be too. Figure out what
106changes would be needed to package that module and its tests up for CPAN, and
107do so. Test it with older perl releases, and fix the problems you find.
e50bb9a1 108
a393eb28
RGS
109To make a minimal perl distribution, it's useful to look at
110F<t/lib/commonsense.t>.
111
c2aba5b8
RGS
112=head2 Bundle dual life modules in ext/
113
114For maintenance (and branch merging) reasons, it would be useful to move
115some architecture-independent dual-life modules from lib/ to ext/, if this
116has no negative impact on the build of perl itself.
117
0bdfc961 118=head2 POSIX memory footprint
e50bb9a1 119
0bdfc961
NC
120Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
121various times worked to cut it down. There is probably still fat to cut out -
122for example POSIX passes Exporter some very memory hungry data structures.
e50bb9a1 123
eed36644
NC
124=head2 embed.pl/makedef.pl
125
126There is a script F<embed.pl> that generates several header files to prefix
127all of Perl's symbols in a consistent way, to provide some semblance of
128namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
907b3e23 129in F<interpvar.h>. Quite a few of the functions and variables
eed36644
NC
130are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
131doesn't understand the C macros, so the rules about which symbols are present
132when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
133It would be good to teach C<embed.pl> to understand the conditional
134compilation, and hence remove the duplication, and the mistakes it has caused.
e50bb9a1 135
801de10e
NC
136=head2 use strict; and AutoLoad
137
138Currently if you write
139
140 package Whack;
141 use AutoLoader 'AUTOLOAD';
142 use strict;
143 1;
144 __END__
145 sub bloop {
146 print join (' ', No, strict, here), "!\n";
147 }
148
149then C<use strict;> isn't in force within the autoloaded subroutines. It would
150be more consistent (and less surprising) to arrange for all lexical pragmas
151in force at the __END__ block to be in force within each autoloaded subroutine.
152
773b3597
RGS
153There's a similar problem with SelfLoader.
154
91d0cbf6
NC
155=head2 profile installman
156
157The F<installman> script is slow. All it is doing text processing, which we're
158told is something Perl is good at. So it would be nice to know what it is doing
159that is taking so much CPU, and where possible address it.
160
161
0bdfc961 162=head1 Tasks that need a little sysadmin-type knowledge
e50bb9a1 163
0bdfc961
NC
164Or if you prefer, tasks that you would learn from, and broaden your skills
165base...
e50bb9a1 166
cd793d32 167=head2 make HTML install work
e50bb9a1 168
adebf063
NC
169There is an C<installhtml> target in the Makefile. It's marked as
170"experimental". It would be good to get this tested, make it work reliably, and
171remove the "experimental" tag. This would include
172
173=over 4
174
175=item 1
176
177Checking that cross linking between various parts of the documentation works.
178In particular that links work between the modules (files with POD in F<lib/>)
179and the core documentation (files in F<pod/>)
180
181=item 2
182
617eabfa
NC
183Work out how to split C<perlfunc> into chunks, preferably one per function
184group, preferably with general case code that could be used elsewhere.
185Challenges here are correctly identifying the groups of functions that go
186together, and making the right named external cross-links point to the right
187page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
188C<endservent>, two or more C<=items> giving the different parameter lists, such
189as
adebf063
NC
190
191 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
adebf063 192 =item substr EXPR,OFFSET,LENGTH
adebf063
NC
193 =item substr EXPR,OFFSET
194
195and different parameter lists having different meanings. (eg C<select>)
196
197=back
3a89a73c 198
0bdfc961
NC
199=head2 compressed man pages
200
201Be able to install them. This would probably need a configure test to see how
202the system does compressed man pages (same directory/different directory?
203same filename/different filename), as well as tweaking the F<installman> script
204to compress as necessary.
205
30222c0f
NC
206=head2 Add a code coverage target to the Makefile
207
208Make it easy for anyone to run Devel::Cover on the core's tests. The steps
209to do this manually are roughly
210
211=over 4
212
213=item *
214
215do a normal C<Configure>, but include Devel::Cover as a module to install
216(see F<INSTALL> for how to do this)
217
218=item *
219
220 make perl
221
222=item *
223
224 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
225
226=item *
227
228Process the resulting Devel::Cover database
229
230=back
231
232This just give you the coverage of the F<.pm>s. To also get the C level
233coverage you need to
234
235=over 4
236
237=item *
238
239Additionally tell C<Configure> to use the appropriate C compiler flags for
240C<gcov>
241
242=item *
243
244 make perl.gcov
245
246(instead of C<make perl>)
247
248=item *
249
250After running the tests run C<gcov> to generate all the F<.gcov> files.
251(Including down in the subdirectories of F<ext/>
252
253=item *
254
255(From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
256to get their stats into the cover_db directory.
257
258=item *
259
260Then process the Devel::Cover database
261
262=back
263
264It would be good to add a single switch to C<Configure> to specify that you
265wanted to perform perl level coverage, and another to specify C level
266coverage, and have C<Configure> and the F<Makefile> do all the right things
267automatically.
268
02f21748 269=head2 Make Config.pm cope with differences between built and installed perl
0bdfc961
NC
270
271Quite often vendors ship a perl binary compiled with their (pay-for)
272compilers. People install a free compiler, such as gcc. To work out how to
273build extensions, Perl interrogates C<%Config>, so in this situation
274C<%Config> describes compilers that aren't there, and extension building
275fails. This forces people into choosing between re-compiling perl themselves
276using the compiler they have, or only using modules that the vendor ships.
277
278It would be good to find a way teach C<Config.pm> about the installation setup,
279possibly involving probing at install time or later, so that the C<%Config> in
280a binary distribution better describes the installed machine, when the
281installed machine differs from the build machine in some significant way.
282
728f4ecd
NC
283=head2 linker specification files
284
285Some platforms mandate that you provide a list of a shared library's external
286symbols to the linker, so the core already has the infrastructure in place to
287do this for generating shared perl libraries. My understanding is that the
288GNU toolchain can accept an optional linker specification file, and restrict
289visibility just to symbols declared in that file. It would be good to extend
290F<makedef.pl> to support this format, and to provide a means within
291C<Configure> to enable it. This would allow Unix users to test that the
292export list is correct, and to build a perl that does not pollute the global
293namespace with private symbols.
294
a229ae3b
RGS
295=head2 Cross-compile support
296
297Currently C<Configure> understands C<-Dusecrosscompile> option. This option
298arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
299assumed then to be copied to TARGET machine and used as a replacement of full
300C<perl> executable.
301
d1307786 302This could be done little differently. Namely C<miniperl> should be built for
a229ae3b 303HOST and then full C<perl> with extensions should be compiled for TARGET.
d1307786 304This, however, might require extra trickery for %Config: we have one config
87a942b1
JH
305first for HOST and then another for TARGET. Tools like MakeMaker will be
306mightily confused. Having around two different types of executables and
307libraries (HOST and TARGET) makes life interesting for Makefiles and
308shell (and Perl) scripts. There is $Config{run}, normally empty, which
309can be used as an execution wrapper. Also note that in some
310cross-compilation/execution environments the HOST and the TARGET do
311not see the same filesystem(s), the $Config{run} may need to do some
312file/directory copying back and forth.
0bdfc961 313
8537f021
RGS
314=head2 roffitall
315
316Make F<pod/roffitall> be updated by F<pod/buildtoc>.
317
98fca0e8
NC
318=head2 Split "linker" from "compiler"
319
320Right now, Configure probes for two commands, and sets two variables:
321
322=over 4
323
b91dd380 324=item * C<cc> (in F<cc.U>)
98fca0e8
NC
325
326This variable holds the name of a command to execute a C compiler which
327can resolve multiple global references that happen to have the same
328name. Usual values are F<cc> and F<gcc>.
329Fervent ANSI compilers may be called F<c89>. AIX has F<xlc>.
330
b91dd380 331=item * C<ld> (in F<dlsrc.U>)
98fca0e8
NC
332
333This variable indicates the program to be used to link
334libraries for dynamic loading. On some systems, it is F<ld>.
335On ELF systems, it should be C<$cc>. Mostly, we'll try to respect
336the hint file setting.
337
338=back
339
8d159ec1
NC
340There is an implicit historical assumption from around Perl5.000alpha
341something, that C<$cc> is also the correct command for linking object files
342together to make an executable. This may be true on Unix, but it's not true
343on other platforms, and there are a maze of work arounds in other places (such
344as F<Makefile.SH>) to cope with this.
98fca0e8
NC
345
346Ideally, we should create a new variable to hold the name of the executable
347linker program, probe for it in F<Configure>, and centralise all the special
348case logic there or in hints files.
349
350A small bikeshed issue remains - what to call it, given that C<$ld> is already
8d159ec1
NC
351taken (arguably for the wrong thing now, but on SunOS 4.1 it is the command
352for creating dynamically-loadable modules) and C<$link> could be confused with
353the Unix command line executable of the same name, which does something
354completely different. Andy Dougherty makes the counter argument "In parrot, I
355tried to call the command used to link object files and libraries into an
356executable F<link>, since that's what my vaguely-remembered DOS and VMS
357experience suggested. I don't think any real confusion has ensued, so it's
358probably a reasonable name for perl5 to use."
98fca0e8
NC
359
360"Alas, I've always worried that introducing it would make things worse,
361since now the module building utilities would have to look for
362C<$Config{link}> and institute a fall-back plan if it weren't found."
8d159ec1
NC
363Although I can see that as confusing, given that C<$Config{d_link}> is true
364when (hard) links are available.
98fca0e8 365
75585ce3
SP
366=head2 Configure Windows using PowerShell
367
368Currently, Windows uses hard-coded config files based to build the
369config.h for compiling Perl. Makefiles are also hard-coded and need to be
370hand edited prior to building Perl. While this makes it easy to create a perl.exe
371that works across multiple Windows versions, being able to accurately
372configure a perl.exe for a specific Windows versions and VS C++ would be
373a nice enhancement. With PowerShell available on Windows XP and up, this
374may now be possible. Step 1 might be to investigate whether this is possible
375and use this to clean up our current makefile situation. Step 2 would be to
376see if there would be a way to use our existing metaconfig units to configure a
377Windows Perl or whether we go in a separate direction and make it so. Of
378course, we all know what step 3 is.
379
ab45a0fa
NC
380=head2 decouple -g and -DDEBUGGING
381
382Currently F<Configure> automatically adds C<-DDEBUGGING> to the C compiler
383flags if it spots C<-g> in the optimiser flags. The pre-processor directive
384C<DEBUGGING> enables F<perl>'s command line <-D> options, but in the process
385makes F<perl> slower. It would be good to disentangle this logic, so that
386C-level debugging with C<-g> and Perl level debugging with C<-D> can easily
387be enabled independently.
388
0bdfc961
NC
389=head1 Tasks that need a little C knowledge
390
391These tasks would need a little C knowledge, but don't need any specific
392background or experience with XS, or how the Perl interpreter works
393
3d826b29
NC
394=head2 Weed out needless PERL_UNUSED_ARG
395
396The C code uses the macro C<PERL_UNUSED_ARG> to stop compilers warning about
397unused arguments. Often the arguments can't be removed, as there is an
398external constraint that determines the prototype of the function, so this
399approach is valid. However, there are some cases where C<PERL_UNUSED_ARG>
400could be removed. Specifically
401
402=over 4
403
404=item *
405
406The prototypes of (nearly all) static functions can be changed
407
408=item *
409
410Unused arguments generated by short cut macros are wasteful - the short cut
411macro used can be changed.
412
413=back
414
fbf638cb
RGS
415=head2 Modernize the order of directories in @INC
416
417The way @INC is laid out by default, one cannot upgrade core (dual-life)
418modules without overwriting files. This causes problems for binary
3d14fd97
AD
419package builders. One possible proposal is laid out in this
420message:
421L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2002-04/msg02380.html>.
fbf638cb 422
bcbaa2d5
RGS
423=head2 -Duse32bit*
424
425Natively 64-bit systems need neither -Duse64bitint nor -Duse64bitall.
426On these systems, it might be the default compilation mode, and there
427is currently no guarantee that passing no use64bitall option to the
428Configure process will build a 32bit perl. Implementing -Duse32bit*
429options would be nice for perl 5.12.
430
fee0a0f7 431=head2 Profile Perl - am I hot or not?
62403a3c 432
fee0a0f7
NC
433The Perl source code is stable enough that it makes sense to profile it,
434identify and optimise the hotspots. It would be good to measure the
435performance of the Perl interpreter using free tools such as cachegrind,
436gprof, and dtrace, and work to reduce the bottlenecks they reveal.
437
438As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
439the ops that are most commonly used. The idea is that by grouping them, their
440object code will be adjacent in the executable, so they have a greater chance
441of already being in the CPU cache (or swapped in) due to being near another op
442already in use.
62403a3c
NC
443
444Except that it's not clear if these really are the most commonly used ops. So
fee0a0f7
NC
445as part of exercising your skills with coverage and profiling tools you might
446want to determine what ops I<really> are the most commonly used. And in turn
447suggest evictions and promotions to achieve a better F<pp_hot.c>.
62403a3c 448
91d0cbf6
NC
449One piece of Perl code that might make a good testbed is F<installman>.
450
98fed0ad
NC
451=head2 Allocate OPs from arenas
452
453Currently all new OP structures are individually malloc()ed and free()d.
454All C<malloc> implementations have space overheads, and are now as fast as
455custom allocates so it would both use less memory and less CPU to allocate
456the various OP structures from arenas. The SV arena code can probably be
457re-used for this.
458
539f2c54
JC
459Note that Configuring perl with C<-Accflags=-DPL_OP_SLAB_ALLOC> will use
460Perl_Slab_alloc() to pack optrees into a contiguous block, which is
461probably superior to the use of OP arenas, esp. from a cache locality
462standpoint. See L<Profile Perl - am I hot or not?>.
463
a229ae3b 464=head2 Improve win32/wince.c
0bdfc961 465
a229ae3b 466Currently, numerous functions look virtually, if not completely,
02f21748 467identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
6d71adcd
NC
468be good.
469
c5b31784
SH
470=head2 Use secure CRT functions when building with VC8 on Win32
471
472Visual C++ 2005 (VC++ 8.x) deprecated a number of CRT functions on the basis
473that they were "unsafe" and introduced differently named secure versions of
474them as replacements, e.g. instead of writing
475
476 FILE* f = fopen(__FILE__, "r");
477
478one should now write
479
480 FILE* f;
481 errno_t err = fopen_s(&f, __FILE__, "r");
482
483Currently, the warnings about these deprecations have been disabled by adding
484-D_CRT_SECURE_NO_DEPRECATE to the CFLAGS. It would be nice to remove that
485warning suppressant and actually make use of the new secure CRT functions.
486
487There is also a similar issue with POSIX CRT function names like fileno having
488been deprecated in favour of ISO C++ conformant names like _fileno. These
26a6faa8 489warnings are also currently suppressed by adding -D_CRT_NONSTDC_NO_DEPRECATE. It
c5b31784
SH
490might be nice to do as Microsoft suggest here too, although, unlike the secure
491functions issue, there is presumably little or no benefit in this case.
492
038ae9a4
SH
493=head2 Fix POSIX::access() and chdir() on Win32
494
495These functions currently take no account of DACLs and therefore do not behave
496correctly in situations where access is restricted by DACLs (as opposed to the
497read-only attribute).
498
499Furthermore, POSIX::access() behaves differently for directories having the
500read-only attribute set depending on what CRT library is being used. For
501example, the _access() function in the VC6 and VC7 CRTs (wrongly) claim that
502such directories are not writable, whereas in fact all directories are writable
503unless access is denied by DACLs. (In the case of directories, the read-only
504attribute actually only means that the directory cannot be deleted.) This CRT
505bug is fixed in the VC8 and VC9 CRTs (but, of course, the directory may still
506not actually be writable if access is indeed denied by DACLs).
507
508For the chdir() issue, see ActiveState bug #74552:
509http://bugs.activestate.com/show_bug.cgi?id=74552
510
511Therefore, DACLs should be checked both for consistency across CRTs and for
512the correct answer.
513
514(Note that perl's -w operator should not be modified to check DACLs. It has
515been written so that it reflects the state of the read-only attribute, even
516for directories (whatever CRT is being used), for symmetry with chmod().)
517
16815324
NC
518=head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
519
520Maybe create a utility that checks after each libperl.a creation that
521none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
522ever creep back to libperl.a.
523
524 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
525
526Note, of course, that this will only tell whether B<your> platform
527is using those naughty interfaces.
528
de96509d
JH
529=head2 -D_FORTIFY_SOURCE=2, -fstack-protector
530
531Recent glibcs support C<-D_FORTIFY_SOURCE=2> and recent gcc
532(4.1 onwards?) supports C<-fstack-protector>, both of which give
533protection against various kinds of buffer overflow problems.
534These should probably be used for compiling Perl whenever available,
535Configure and/or hints files should be adjusted to probe for the
536availability of these features and enable them as appropriate.
16815324 537
8964cfe0
NC
538=head2 Arenas for GPs? For MAGIC?
539
540C<struct gp> and C<struct magic> are both currently allocated by C<malloc>.
541It might be a speed or memory saving to change to using arenas. Or it might
542not. It would need some suitable benchmarking first. In particular, C<GP>s
543can probably be changed with minimal compatibility impact (probably nothing
544outside of the core, or even outside of F<gv.c> allocates them), but they
545probably aren't allocated/deallocated often enough for a speed saving. Whereas
546C<MAGIC> is allocated/deallocated more often, but in turn, is also something
547more externally visible, so changing the rules here may bite external code.
548
3880c8ec
NC
549=head2 Shared arenas
550
551Several SV body structs are now the same size, notably PVMG and PVGV, PVAV and
552PVHV, and PVCV and PVFM. It should be possible to allocate and return same
553sized bodies from the same actual arena, rather than maintaining one arena for
554each. This could save 4-6K per thread, of memory no longer tied up in the
555not-yet-allocated part of an arena.
556
8964cfe0 557
6d71adcd
NC
558=head1 Tasks that need a knowledge of XS
559
560These tasks would need C knowledge, and roughly the level of knowledge of
561the perl API that comes from writing modules that use XS to interface to
562C.
563
318bf708
NC
564=head2 Remove the use of SVs as temporaries in dump.c
565
566F<dump.c> contains debugging routines to dump out the contains of perl data
567structures, such as C<SV>s, C<AV>s and C<HV>s. Currently, the dumping code
568B<uses> C<SV>s for its temporary buffers, which was a logical initial
569implementation choice, as they provide ready made memory handling.
570
571However, they also lead to a lot of confusion when it happens that what you're
572trying to debug is seen by the code in F<dump.c>, correctly or incorrectly, as
573a temporary scalar it can use for a temporary buffer. It's also not possible
574to dump scalars before the interpreter is properly set up, such as during
575ithreads cloning. It would be good to progressively replace the use of scalars
576as string accumulation buffers with something much simpler, directly allocated
577by C<malloc>. The F<dump.c> code is (or should be) only producing 7 bit
578US-ASCII, so output character sets are not an issue.
579
580Producing and proving an internal simple buffer allocation would make it easier
581to re-write the internals of the PerlIO subsystem to avoid using C<SV>s for
582B<its> buffers, use of which can cause problems similar to those of F<dump.c>,
583at similar times.
584
5d96f598
NC
585=head2 safely supporting POSIX SA_SIGINFO
586
587Some years ago Jarkko supplied patches to provide support for the POSIX
588SA_SIGINFO feature in Perl, passing the extra data to the Perl signal handler.
589
590Unfortunately, it only works with "unsafe" signals, because under safe
591signals, by the time Perl gets to run the signal handler, the extra
592information has been lost. Moreover, it's not easy to store it somewhere,
593as you can't call mutexs, or do anything else fancy, from inside a signal
594handler.
595
596So it strikes me that we could provide safe SA_SIGINFO support
597
598=over 4
599
600=item 1
601
602Provide global variables for two file descriptors
603
604=item 2
605
606When the first request is made via C<sigaction> for C<SA_SIGINFO>, create a
607pipe, store the reader in one, the writer in the other
608
609=item 3
610
611In the "safe" signal handler (C<Perl_csighandler()>/C<S_raise_signal()>), if
612the C<siginfo_t> pointer non-C<NULL>, and the writer file handle is open,
613
614=over 8
615
616=item 1
617
618serialise signal number, C<struct siginfo_t> (or at least the parts we care
619about) into a small auto char buff
620
621=item 2
622
623C<write()> that (non-blocking) to the writer fd
624
625=over 12
626
627=item 1
628
629if it writes 100%, flag the signal in a counter of "signals on the pipe" akin
630to the current per-signal-number counts
631
632=item 2
633
634if it writes 0%, assume the pipe is full. Flag the data as lost?
635
636=item 3
637
638if it writes partially, croak a panic, as your OS is broken.
639
640=back
641
642=back
643
644=item 4
645
646in the regular C<PERL_ASYNC_CHECK()> processing, if there are "signals on
647the pipe", read the data out, deserialise, build the Perl structures on
648the stack (code in C<Perl_sighandler()>, the "unsafe" handler), and call as
649usual.
650
651=back
652
653I think that this gets us decent C<SA_SIGINFO> support, without the current risk
654of running Perl code inside the signal handler context. (With all the dangers
655of things like C<malloc> corruption that that currently offers us)
656
657For more information see the thread starting with this message:
658http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-03/msg00305.html
659
6d71adcd
NC
660=head2 autovivification
661
662Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
663
664This task is incremental - even a little bit of work on it will help.
665
666=head2 Unicode in Filenames
667
668chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
669opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
670system, truncate, unlink, utime, -X. All these could potentially accept
671Unicode filenames either as input or output (and in the case of system
672and qx Unicode in general, as input or output to/from the shell).
673Whether a filesystem - an operating system pair understands Unicode in
674filenames varies.
675
676Known combinations that have some level of understanding include
677Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
678OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
679create Unicode filenames, what forms of Unicode are accepted and used
680(UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
681and so on, varies. Finding the right level of interfacing to Perl
682requires some thought. Remember that an OS does not implicate a
683filesystem.
684
685(The Windows -C command flag "wide API support" has been at least
686temporarily retired in 5.8.1, and the -C has been repurposed, see
687L<perlrun>.)
688
87a942b1
JH
689Most probably the right way to do this would be this:
690L</"Virtualize operating system access">.
691
6d71adcd
NC
692=head2 Unicode in %ENV
693
694Currently the %ENV entries are always byte strings.
87a942b1 695See L</"Virtualize operating system access">.
6d71adcd 696
1f2e7916
JD
697=head2 Unicode and glob()
698
699Currently glob patterns and filenames returned from File::Glob::glob()
87a942b1 700are always byte strings. See L</"Virtualize operating system access">.
1f2e7916 701
dbb0c492
RGS
702=head2 Unicode and lc/uc operators
703
704Some built-in operators (C<lc>, C<uc>, etc.) behave differently, based on
705what the internal encoding of their argument is. That should not be the
706case. Maybe add a pragma to switch behaviour.
707
6d71adcd
NC
708=head2 use less 'memory'
709
710Investigate trade offs to switch out perl's choices on memory usage.
711Particularly perl should be able to give memory back.
712
713This task is incremental - even a little bit of work on it will help.
714
715=head2 Re-implement C<:unique> in a way that is actually thread-safe
716
717The old implementation made bad assumptions on several levels. A good 90%
718solution might be just to make C<:unique> work to share the string buffer
719of SvPVs. That way large constant strings can be shared between ithreads,
720such as the configuration information in F<Config>.
721
722=head2 Make tainting consistent
723
724Tainting would be easier to use if it didn't take documented shortcuts and
725allow taint to "leak" everywhere within an expression.
726
727=head2 readpipe(LIST)
728
729system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
730running a shell. readpipe() (the function behind qx//) could be similarly
731extended.
732
6d71adcd
NC
733=head2 Audit the code for destruction ordering assumptions
734
735Change 25773 notes
736
737 /* Need to check SvMAGICAL, as during global destruction it may be that
738 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
739 is now part of the linked list of SV heads, rather than pointing to
740 the original body. */
741 /* FIXME - audit the code for other bugs like this one. */
742
743adding the C<SvMAGICAL> check to
744
745 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
746 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
747
748Go through the core and look for similar assumptions that SVs have particular
749types, as all bets are off during global destruction.
750
749904bf
JH
751=head2 Extend PerlIO and PerlIO::Scalar
752
753PerlIO::Scalar doesn't know how to truncate(). Implementing this
754would require extending the PerlIO vtable.
755
756Similarly the PerlIO vtable doesn't know about formats (write()), or
757about stat(), or chmod()/chown(), utime(), or flock().
758
759(For PerlIO::Scalar it's hard to see what e.g. mode bits or ownership
760would mean.)
761
762PerlIO doesn't do directories or symlinks, either: mkdir(), rmdir(),
763opendir(), closedir(), seekdir(), rewinddir(), glob(); symlink(),
764readlink().
765
94da6c29
JH
766See also L</"Virtualize operating system access">.
767
3236f110
NC
768=head2 -C on the #! line
769
770It should be possible to make -C work correctly if found on the #! line,
771given that all perl command line options are strict ASCII, and -C changes
772only the interpretation of non-ASCII characters, and not for the script file
773handle. To make it work needs some investigation of the ordering of function
774calls during startup, and (by implication) a bit of tweaking of that order.
775
d6c1e11f
JH
776=head2 Organize error messages
777
778Perl's diagnostics (error messages, see L<perldiag>) could use
a8d0aeb9 779reorganizing and formalizing so that each error message has its
d6c1e11f
JH
780stable-for-all-eternity unique id, categorized by severity, type, and
781subsystem. (The error messages would be listed in a datafile outside
c4bd451b
CB
782of the Perl source code, and the source code would only refer to the
783messages by the id.) This clean-up and regularizing should apply
d6c1e11f
JH
784for all croak() messages.
785
786This would enable all sorts of things: easier translation/localization
787of the messages (though please do keep in mind the caveats of
788L<Locale::Maketext> about too straightforward approaches to
789translation), filtering by severity, and instead of grepping for a
790particular error message one could look for a stable error id. (Of
791course, changing the error messages by default would break all the
792existing software depending on some particular error message...)
793
794This kind of functionality is known as I<message catalogs>. Look for
795inspiration for example in the catgets() system, possibly even use it
796if available-- but B<only> if available, all platforms will B<not>
de96509d 797have catgets().
d6c1e11f
JH
798
799For the really pure at heart, consider extending this item to cover
800also the warning messages (see L<perllexwarn>, C<warnings.pl>).
3236f110 801
0bdfc961 802=head1 Tasks that need a knowledge of the interpreter
3298bd4d 803
0bdfc961
NC
804These tasks would need C knowledge, and knowledge of how the interpreter works,
805or a willingness to learn.
3298bd4d 806
dee8984c
RGS
807=head2 forbid labels with keyword names
808
809Currently C<goto keyword> "computes" the label value:
810
811 $ perl -e 'goto print'
812 Can't find label 1 at -e line 1.
813
814It would be nice to forbid labels with keyword names, to avoid confusion.
815
de6375e3
RGS
816=head2 truncate() prototype
817
818The prototype of truncate() is currently C<$$>. It should probably
819be C<*$> instead. (This is changed in F<opcode.pl>)
820
2d0587d8
RGS
821=head2 decapsulation of smart match argument
822
823Currently C<$foo ~~ $object> will die with the message "Smart matching a
824non-overloaded object breaks encapsulation". It would be nice to allow
825to bypass this by using explictly the syntax C<$foo ~~ %$object> or
826C<$foo ~~ @$object>.
827
565590b5
NC
828=head2 error reporting of [$a ; $b]
829
830Using C<;> inside brackets is a syntax error, and we don't propose to change
831that by giving it any meaning. However, it's not reported very helpfully:
832
833 $ perl -e '$a = [$b; $c];'
834 syntax error at -e line 1, near "$b;"
835 syntax error at -e line 1, near "$c]"
836 Execution of -e aborted due to compilation errors.
837
838It should be possible to hook into the tokeniser or the lexer, so that when a
839C<;> is parsed where it is not legal as a statement terminator (ie inside
840C<{}> used as a hashref, C<[]> or C<()>) it issues an error something like
841I<';' isn't legal inside an expression - if you need multiple statements use a
842do {...} block>. See the thread starting at
843http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2008-09/msg00573.html
844
718140ec
NC
845=head2 lexicals used only once
846
847This warns:
848
849 $ perl -we '$pie = 42'
850 Name "main::pie" used only once: possible typo at -e line 1.
851
852This does not:
853
854 $ perl -we 'my $pie = 42'
855
856Logically all lexicals used only once should warn, if the user asks for
d6f4ea2e
SP
857warnings. An unworked RT ticket (#5087) has been open for almost seven
858years for this discrepancy.
718140ec 859
a3d15f9a
RGS
860=head2 UTF-8 revamp
861
862The handling of Unicode is unclean in many places. For example, the regexp
863engine matches in Unicode semantics whenever the string or the pattern is
864flagged as UTF-8, but that should not be dependent on an internal storage
865detail of the string. Likewise, case folding behaviour is dependent on the
866UTF8 internal flag being on or off.
867
868=head2 Properly Unicode safe tokeniser and pads.
869
870The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
871variable names are stored in stashes as raw bytes, without the utf-8 flag
872set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
873tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
874source filters. All this could be fixed.
875
636e63cb
NC
876=head2 state variable initialization in list context
877
878Currently this is illegal:
879
880 state ($a, $b) = foo();
881
a2874905 882In Perl 6, C<state ($a) = foo();> and C<(state $a) = foo();> have different
a8d0aeb9 883semantics, which is tricky to implement in Perl 5 as currently they produce
a2874905 884the same opcode trees. The Perl 6 design is firm, so it would be good to
a8d0aeb9 885implement the necessary code in Perl 5. There are comments in
a2874905
NC
886C<Perl_newASSIGNOP()> that show the code paths taken by various assignment
887constructions involving state variables.
636e63cb 888
4fedb12c
RGS
889=head2 Implement $value ~~ 0 .. $range
890
891It would be nice to extend the syntax of the C<~~> operator to also
892understand numeric (and maybe alphanumeric) ranges.
a393eb28
RGS
893
894=head2 A does() built-in
895
896Like ref(), only useful. It would call the C<DOES> method on objects; it
897would also tell whether something can be dereferenced as an
898array/hash/etc., or used as a regexp, etc.
899L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-03/msg00481.html>
900
901=head2 Tied filehandles and write() don't mix
902
903There is no method on tied filehandles to allow them to be called back by
904formats.
4fedb12c 905
d10fc472 906=head2 Attach/detach debugger from running program
1626a787 907
cd793d32
NC
908The old perltodo notes "With C<gdb>, you can attach the debugger to a running
909program if you pass the process ID. It would be good to do this with the Perl
0bdfc961
NC
910debugger on a running Perl program, although I'm not sure how it would be
911done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
1626a787 912
0bdfc961
NC
913=head2 LVALUE functions for lists
914
915The old perltodo notes that lvalue functions don't work for list or hash
916slices. This would be good to fix.
917
0bdfc961
NC
918=head2 regexp optimiser optional
919
920The regexp optimiser is not optional. It should configurable to be, to allow
921its performance to be measured, and its bugs to be easily demonstrated.
922
02f21748
RGS
923=head2 delete &function
924
925Allow to delete functions. One can already undef them, but they're still
926in the stash.
927
ef36c6a7
RGS
928=head2 C</w> regex modifier
929
930That flag would enable to match whole words, and also to interpolate
931arrays as alternations. With it, C</P/w> would be roughly equivalent to:
932
933 do { local $"='|'; /\b(?:P)\b/ }
934
935See L<http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-01/msg00400.html>
936for the discussion.
937
0bdfc961
NC
938=head2 optional optimizer
939
940Make the peephole optimizer optional. Currently it performs two tasks as
941it walks the optree - genuine peephole optimisations, and necessary fixups of
942ops. It would be good to find an efficient way to switch out the
943optimisations whilst keeping the fixups.
944
945=head2 You WANT *how* many
946
947Currently contexts are void, scalar and list. split has a special mechanism in
948place to pass in the number of return values wanted. It would be useful to
949have a general mechanism for this, backwards compatible and little speed hit.
950This would allow proposals such as short circuiting sort to be implemented
951as a module on CPAN.
952
953=head2 lexical aliases
954
955Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
956
957=head2 entersub XS vs Perl
958
959At the moment pp_entersub is huge, and has code to deal with entering both
960perl and XS subroutines. Subroutine implementations rarely change between
961perl and XS at run time, so investigate using 2 ops to enter subs (one for
962XS, one for perl) and swap between if a sub is redefined.
2810d901 963
de535794 964=head2 Self-ties
2810d901 965
de535794 966Self-ties are currently illegal because they caused too many segfaults. Maybe
a8d0aeb9 967the causes of these could be tracked down and self-ties on all types
de535794 968reinstated.
0bdfc961
NC
969
970=head2 Optimize away @_
971
972The old perltodo notes "Look at the "reification" code in C<av.c>".
973
87a942b1
JH
974=head2 Virtualize operating system access
975
976Implement a set of "vtables" that virtualizes operating system access
977(open(), mkdir(), unlink(), readdir(), getenv(), etc.) At the very
978least these interfaces should take SVs as "name" arguments instead of
979bare char pointers; probably the most flexible and extensible way
e1a3d5d1
JH
980would be for the Perl-facing interfaces to accept HVs. The system
981needs to be per-operating-system and per-file-system
982hookable/filterable, preferably both from XS and Perl level
87a942b1
JH
983(L<perlport/"Files and Filesystems"> is good reading at this point,
984in fact, all of L<perlport> is.)
985
e1a3d5d1
JH
986This has actually already been implemented (but only for Win32),
987take a look at F<iperlsys.h> and F<win32/perlhost.h>. While all Win32
988variants go through a set of "vtables" for operating system access,
989non-Win32 systems currently go straight for the POSIX/UNIX-style
990system/library call. Similar system as for Win32 should be
991implemented for all platforms. The existing Win32 implementation
992probably does not need to survive alongside this proposed new
993implementation, the approaches could be merged.
87a942b1
JH
994
995What would this give us? One often-asked-for feature this would
94da6c29
JH
996enable is using Unicode for filenames, and other "names" like %ENV,
997usernames, hostnames, and so forth.
998(See L<perlunicode/"When Unicode Does Not Happen">.)
999
1000But this kind of virtualization would also allow for things like
1001virtual filesystems, virtual networks, and "sandboxes" (though as long
1002as dynamic loading of random object code is allowed, not very safe
1003sandboxes since external code of course know not of Perl's vtables).
1004An example of a smaller "sandbox" is that this feature can be used to
1005implement per-thread working directories: Win32 already does this.
1006
1007See also L</"Extend PerlIO and PerlIO::Scalar">.
87a942b1 1008
ac6197af
NC
1009=head2 Investigate PADTMP hash pessimisation
1010
9a2f2e6b 1011The peephole optimiser converts constants used for hash key lookups to shared
057163d7 1012hash key scalars. Under ithreads, something is undoing this work.
ac6197af
NC
1013See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2007-09/msg00793.html
1014
057163d7
NC
1015=head2 Store the current pad in the OP slab allocator
1016
1017=for clarification
1018I hope that I got that "current pad" part correct
1019
1020Currently we leak ops in various cases of parse failure. I suggested that we
1021could solve this by always using the op slab allocator, and walking it to
1022free ops. Dave comments that as some ops are already freed during optree
1023creation one would have to mark which ops are freed, and not double free them
1024when walking the slab. He notes that one problem with this is that for some ops
1025you have to know which pad was current at the time of allocation, which does
1026change. I suggested storing a pointer to the current pad in the memory allocated
1027for the slab, and swapping to a new slab each time the pad changes. Dave thinks
1028that this would work.
1029
52960e22
JC
1030=head2 repack the optree
1031
1032Repacking the optree after execution order is determined could allow
057163d7
NC
1033removal of NULL ops, and optimal ordering of OPs with respect to cache-line
1034filling. The slab allocator could be reused for this purpose. I think that
1035the best way to do this is to make it an optional step just before the
1036completed optree is attached to anything else, and to use the slab allocator
1037unchanged, so that freeing ops is identical whether or not this step runs.
1038Note that the slab allocator allocates ops downwards in memory, so one would
1039have to actually "allocate" the ops in reverse-execution order to get them
1040contiguous in memory in execution order.
1041
1042See http://www.nntp.perl.org/group/perl.perl5.porters/2007/12/msg131975.html
1043
1044Note that running this copy, and then freeing all the old location ops would
1045cause their slabs to be freed, which would eliminate possible memory wastage if
1046the previous suggestion is implemented, and we swap slabs more frequently.
52960e22 1047
12e06b6f
NC
1048=head2 eliminate incorrect line numbers in warnings
1049
1050This code
1051
1052 use warnings;
1053 my $undef;
1054
1055 if ($undef == 3) {
1056 } elsif ($undef == 0) {
1057 }
1058
18a16cc5 1059used to produce this output:
12e06b6f
NC
1060
1061 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1062 Use of uninitialized value in numeric eq (==) at wrong.pl line 4.
1063
18a16cc5
NC
1064where the line of the second warning was misreported - it should be line 5.
1065Rafael fixed this - the problem arose because there was no nextstate OP
1066between the execution of the C<if> and the C<elsif>, hence C<PL_curcop> still
1067reports that the currently executing line is line 4. The solution was to inject
1068a nextstate OPs for each C<elsif>, although it turned out that the nextstate
1069OP needed to be a nulled OP, rather than a live nextstate OP, else other line
1070numbers became misreported. (Jenga!)
12e06b6f
NC
1071
1072The problem is more general than C<elsif> (although the C<elsif> case is the
1073most common and the most confusing). Ideally this code
1074
1075 use warnings;
1076 my $undef;
1077
1078 my $a = $undef + 1;
1079 my $b
1080 = $undef
1081 + 1;
1082
1083would produce this output
1084
1085 Use of uninitialized value $undef in addition (+) at wrong.pl line 4.
1086 Use of uninitialized value $undef in addition (+) at wrong.pl line 7.
1087
1088(rather than lines 4 and 5), but this would seem to require every OP to carry
1089(at least) line number information.
1090
1091What might work is to have an optional line number in memory just before the
1092BASEOP structure, with a flag bit in the op to say whether it's present.
1093Initially during compile every OP would carry its line number. Then add a late
1094pass to the optimiser (potentially combined with L</repack the optree>) which
1095looks at the two ops on every edge of the graph of the execution path. If
1096the line number changes, flags the destination OP with this information.
1097Once all paths are traced, replace every op with the flag with a
1098nextstate-light op (that just updates C<PL_curcop>), which in turn then passes
1099control on to the true op. All ops would then be replaced by variants that
1100do not store the line number. (Which, logically, why it would work best in
1101conjunction with L</repack the optree>, as that is already copying/reallocating
1102all the OPs)
1103
18a16cc5
NC
1104(Although I should note that we're not certain that doing this for the general
1105case is worth it)
1106
52960e22
JC
1107=head2 optimize tail-calls
1108
1109Tail-calls present an opportunity for broadly applicable optimization;
1110anywhere that C<< return foo(...) >> is called, the outer return can
1111be replaced by a goto, and foo will return directly to the outer
1112caller, saving (conservatively) 25% of perl's call&return cost, which
1113is relatively higher than in C. The scheme language is known to do
1114this heavily. B::Concise provides good insight into where this
1115optimization is possible, ie anywhere entersub,leavesub op-sequence
1116occurs.
1117
1118 perl -MO=Concise,-exec,a,b,-main -e 'sub a{ 1 }; sub b {a()}; b(2)'
1119
1120Bottom line on this is probably a new pp_tailcall function which
1121combines the code in pp_entersub, pp_leavesub. This should probably
1122be done 1st in XS, and using B::Generate to patch the new OP into the
1123optrees.
1124
0bdfc961
NC
1125=head1 Big projects
1126
1127Tasks that will get your name mentioned in the description of the "Highlights
87a942b1 1128of 5.12"
0bdfc961
NC
1129
1130=head2 make ithreads more robust
1131
4e577f8b 1132Generally make ithreads more robust. See also L</iCOW>
0bdfc961
NC
1133
1134This task is incremental - even a little bit of work on it will help, and
1135will be greatly appreciated.
1136
6c047da7
YST
1137One bit would be to write the missing code in sv.c:Perl_dirp_dup.
1138
59c7f7d5
RGS
1139Fix Perl_sv_dup, et al so that threads can return objects.
1140
0bdfc961
NC
1141=head2 iCOW
1142
1143Sarathy and Arthur have a proposal for an improved Copy On Write which
1144specifically will be able to COW new ithreads. If this can be implemented
1145it would be a good thing.
1146
1147=head2 (?{...}) closures in regexps
1148
1149Fix (or rewrite) the implementation of the C</(?{...})/> closures.
1150
1151=head2 A re-entrant regexp engine
1152
1153This will allow the use of a regex from inside (?{ }), (??{ }) and
1154(?(?{ })|) constructs.
6bda09f9 1155
6bda09f9
YO
1156=head2 Add class set operations to regexp engine
1157
1158Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
1159
1160demerphq has this on his todo list, but right at the bottom.