This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
installman3dir is null
[perl5.git] / pod / perltodo.pod
CommitLineData
7711098a
GS
1=head1 NAME
2
3perltodo - Perl TO-DO List
4
5=head1 DESCRIPTION
e50bb9a1 6
0bdfc961
NC
7This is a list of wishes for Perl. The tasks we think are smaller or easier
8are listed first. Anyone is welcome to work on any of these, but it's a good
9idea to first contact I<perl5-porters@perl.org> to avoid duplication of
10effort. By all means contact a pumpking privately first if you prefer.
e50bb9a1 11
0bdfc961
NC
12Whilst patches to make the list shorter are most welcome, ideas to add to
13the list are also encouraged. Check the perl5-porters archives for past
14ideas, and any discussion about them. One set of archives may be found at:
e50bb9a1 15
0bdfc961 16 http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/
938c8732 17
617eabfa
NC
18What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
19not, but if your patch is incorporated, then we'll add your name to the
20F<AUTHORS> file, which ships in the official distribution. How many other
21programming languages offer you 1 line of immortality?
938c8732 22
4e577f8b 23=head1 The roadmap to 5.10
938c8732 24
4e577f8b
NC
25The roadmap to 5.10 envisages feature based releases, as various items in this
26TODO are completed.
27
256ddcd0
YO
28=head2 Needed for the final 5.10.0 release
29
256ddcd0
YO
30Review perlguts. Significant changes have occured since 5.8, and we can't
31release a new version without making sure these are covered.
32
0bdfc961 33=head1 Tasks that only need Perl knowledge
e50bb9a1 34
0bdfc961 35=head2 common test code for timed bail out
e50bb9a1 36
0bdfc961
NC
37Write portable self destruct code for tests to stop them burning CPU in
38infinite loops. This needs to avoid using alarm, as some of the tests are
39testing alarm/sleep or timers.
e50bb9a1 40
0bdfc961 41=head2 POD -> HTML conversion in the core still sucks
e50bb9a1 42
938c8732 43Which is crazy given just how simple POD purports to be, and how simple HTML
adebf063
NC
44can be. It's not actually I<as> simple as it sounds, particularly with the
45flexibility POD allows for C<=item>, but it would be good to improve the
46visual appeal of the HTML generated, and to avoid it having any validation
47errors. See also L</make HTML install work>, as the layout of installation tree
48is needed to improve the cross-linking.
938c8732 49
dc0fb092
SP
50The addition of C<Pod::Simple> and its related modules may make this task
51easier to complete.
52
aa237293
NC
53=head2 Parallel testing
54
b2e2905c 55(This probably impacts much more than the core: also the Test::Harness
02f21748
RGS
56and TAP::* modules on CPAN.)
57
aa237293
NC
58The core regression test suite is getting ever more comprehensive, which has
59the side effect that it takes longer to run. This isn't so good. Investigate
60whether it would be feasible to give the harness script the B<option> of
61running sets of tests in parallel. This would be useful for tests in
62F<t/op/*.t> and F<t/uni/*.t> and maybe some sets of tests in F<lib/>.
63
64Questions to answer
65
66=over 4
67
68=item 1
69
70How does screen layout work when you're running more than one test?
71
72=item 2
73
74How does the caller of test specify how many tests to run in parallel?
75
76=item 3
77
78How do setup/teardown tests identify themselves?
79
80=back
81
82Pugs already does parallel testing - can their approach be re-used?
83
0bdfc961 84=head2 Make Schwern poorer
e50bb9a1 85
613bd4f7 86We should have tests for everything. When all the core's modules are tested,
0bdfc961
NC
87Schwern has promised to donate to $500 to TPF. We may need volunteers to
88hold him upside down and shake vigorously in order to actually extract the
89cash.
3958b146 90
0bdfc961 91=head2 Improve the coverage of the core tests
e50bb9a1 92
02f21748
RGS
93Use Devel::Cover to ascertain the core modules's test coverage, then add
94tests that are currently missing.
30222c0f 95
0bdfc961 96=head2 test B
e50bb9a1 97
0bdfc961 98A full test suite for the B module would be nice.
e50bb9a1 99
0bdfc961 100=head2 A decent benchmark
e50bb9a1 101
617eabfa 102C<perlbench> seems impervious to any recent changes made to the perl core. It
0bdfc961
NC
103would be useful to have a reasonable general benchmarking suite that roughly
104represented what current perl programs do, and measurably reported whether
105tweaks to the core improve, degrade or don't really affect performance, to
106guide people attempting to optimise the guts of perl. Gisle would welcome
107new tests for perlbench.
6168cf99 108
0bdfc961 109=head2 fix tainting bugs
6168cf99 110
0bdfc961
NC
111Fix the bugs revealed by running the test suite with the C<-t> switch (via
112C<make test.taintwarn>).
e50bb9a1 113
0bdfc961 114=head2 Dual life everything
e50bb9a1 115
0bdfc961
NC
116As part of the "dists" plan, anything that doesn't belong in the smallest perl
117distribution needs to be dual lifed. Anything else can be too. Figure out what
118changes would be needed to package that module and its tests up for CPAN, and
119do so. Test it with older perl releases, and fix the problems you find.
e50bb9a1 120
0bdfc961 121=head2 Improving C<threads::shared>
722d2a37 122
0bdfc961
NC
123Investigate whether C<threads::shared> could share aggregates properly with
124only Perl level changes to shared.pm
722d2a37 125
0bdfc961 126=head2 POSIX memory footprint
e50bb9a1 127
0bdfc961
NC
128Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
129various times worked to cut it down. There is probably still fat to cut out -
130for example POSIX passes Exporter some very memory hungry data structures.
e50bb9a1 131
eed36644
NC
132=head2 embed.pl/makedef.pl
133
134There is a script F<embed.pl> that generates several header files to prefix
135all of Perl's symbols in a consistent way, to provide some semblance of
136namespace support in C<C>. Functions are declared in F<embed.fnc>, variables
137in F<interpvar.h> and F<thrdvar.h>. Quite a few of the functions and variables
138are conditionally declared there, using C<#ifdef>. However, F<embed.pl>
139doesn't understand the C macros, so the rules about which symbols are present
140when is duplicated in F<makedef.pl>. Writing things twice is bad, m'kay.
141It would be good to teach C<embed.pl> to understand the conditional
142compilation, and hence remove the duplication, and the mistakes it has caused.
e50bb9a1 143
0bdfc961 144=head1 Tasks that need a little sysadmin-type knowledge
e50bb9a1 145
0bdfc961
NC
146Or if you prefer, tasks that you would learn from, and broaden your skills
147base...
e50bb9a1 148
cd793d32 149=head2 make HTML install work
e50bb9a1 150
adebf063
NC
151There is an C<installhtml> target in the Makefile. It's marked as
152"experimental". It would be good to get this tested, make it work reliably, and
153remove the "experimental" tag. This would include
154
155=over 4
156
157=item 1
158
159Checking that cross linking between various parts of the documentation works.
160In particular that links work between the modules (files with POD in F<lib/>)
161and the core documentation (files in F<pod/>)
162
163=item 2
164
617eabfa
NC
165Work out how to split C<perlfunc> into chunks, preferably one per function
166group, preferably with general case code that could be used elsewhere.
167Challenges here are correctly identifying the groups of functions that go
168together, and making the right named external cross-links point to the right
169page. Things to be aware of are C<-X>, groups such as C<getpwnam> to
170C<endservent>, two or more C<=items> giving the different parameter lists, such
171as
adebf063
NC
172
173 =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
adebf063 174 =item substr EXPR,OFFSET,LENGTH
adebf063
NC
175 =item substr EXPR,OFFSET
176
177and different parameter lists having different meanings. (eg C<select>)
178
179=back
3a89a73c 180
0bdfc961
NC
181=head2 compressed man pages
182
183Be able to install them. This would probably need a configure test to see how
184the system does compressed man pages (same directory/different directory?
185same filename/different filename), as well as tweaking the F<installman> script
186to compress as necessary.
187
30222c0f
NC
188=head2 Add a code coverage target to the Makefile
189
190Make it easy for anyone to run Devel::Cover on the core's tests. The steps
191to do this manually are roughly
192
193=over 4
194
195=item *
196
197do a normal C<Configure>, but include Devel::Cover as a module to install
198(see F<INSTALL> for how to do this)
199
200=item *
201
202 make perl
203
204=item *
205
206 cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness
207
208=item *
209
210Process the resulting Devel::Cover database
211
212=back
213
214This just give you the coverage of the F<.pm>s. To also get the C level
215coverage you need to
216
217=over 4
218
219=item *
220
221Additionally tell C<Configure> to use the appropriate C compiler flags for
222C<gcov>
223
224=item *
225
226 make perl.gcov
227
228(instead of C<make perl>)
229
230=item *
231
232After running the tests run C<gcov> to generate all the F<.gcov> files.
233(Including down in the subdirectories of F<ext/>
234
235=item *
236
237(From the top level perl directory) run C<gcov2perl> on all the C<.gcov> files
238to get their stats into the cover_db directory.
239
240=item *
241
242Then process the Devel::Cover database
243
244=back
245
246It would be good to add a single switch to C<Configure> to specify that you
247wanted to perform perl level coverage, and another to specify C level
248coverage, and have C<Configure> and the F<Makefile> do all the right things
249automatically.
250
02f21748 251=head2 Make Config.pm cope with differences between built and installed perl
0bdfc961
NC
252
253Quite often vendors ship a perl binary compiled with their (pay-for)
254compilers. People install a free compiler, such as gcc. To work out how to
255build extensions, Perl interrogates C<%Config>, so in this situation
256C<%Config> describes compilers that aren't there, and extension building
257fails. This forces people into choosing between re-compiling perl themselves
258using the compiler they have, or only using modules that the vendor ships.
259
260It would be good to find a way teach C<Config.pm> about the installation setup,
261possibly involving probing at install time or later, so that the C<%Config> in
262a binary distribution better describes the installed machine, when the
263installed machine differs from the build machine in some significant way.
264
728f4ecd
NC
265=head2 linker specification files
266
267Some platforms mandate that you provide a list of a shared library's external
268symbols to the linker, so the core already has the infrastructure in place to
269do this for generating shared perl libraries. My understanding is that the
270GNU toolchain can accept an optional linker specification file, and restrict
271visibility just to symbols declared in that file. It would be good to extend
272F<makedef.pl> to support this format, and to provide a means within
273C<Configure> to enable it. This would allow Unix users to test that the
274export list is correct, and to build a perl that does not pollute the global
275namespace with private symbols.
276
a229ae3b
RGS
277=head2 Cross-compile support
278
279Currently C<Configure> understands C<-Dusecrosscompile> option. This option
280arranges for building C<miniperl> for TARGET machine, so this C<miniperl> is
281assumed then to be copied to TARGET machine and used as a replacement of full
282C<perl> executable.
283
d1307786 284This could be done little differently. Namely C<miniperl> should be built for
a229ae3b 285HOST and then full C<perl> with extensions should be compiled for TARGET.
d1307786
JH
286This, however, might require extra trickery for %Config: we have one config
287first for HOST and then another for TARGET.
0bdfc961
NC
288
289=head1 Tasks that need a little C knowledge
290
291These tasks would need a little C knowledge, but don't need any specific
292background or experience with XS, or how the Perl interpreter works
293
294=head2 Make it clear from -v if this is the exact official release
89007cb3 295
617eabfa
NC
296Currently perl from C<p4>/C<rsync> ships with a F<patchlevel.h> file that
297usually defines one local patch, of the form "MAINT12345" or "RC1". The output
298of perl -v doesn't report that a perl isn't an official release, and this
89007cb3 299information can get lost in bugs reports. Because of this, the minor version
fa11829f 300isn't bumped up until RC time, to minimise the possibility of versions of perl
89007cb3
NC
301escaping that believe themselves to be newer than they actually are.
302
303It would be useful to find an elegant way to have the "this is an interim
304maintenance release" or "this is a release candidate" in the terse -v output,
305and have it so that it's easy for the pumpking to remove this just as the
306release tarball is rolled up. This way the version pulled out of rsync would
307always say "I'm a development release" and it would be safe to bump the
308reported minor version as soon as a release ships, which would aid perl
309developers.
310
0bdfc961
NC
311This task is really about thinking of an elegant way to arrange the C source
312such that it's trivial for the Pumpking to flag "this is an official release"
313when making a tarball, yet leave the default source saying "I'm not the
314official release".
315
0f788cd2
NC
316=head2 Ordering of "global" variables.
317
318F<thrdvar.h> and F<intrpvarh> define the "global" variables that need to be
319per-thread under ithreads, where the variables are actually elements in a
320structure. As C dictates, the variables must be laid out in order of
321declaration. There is a comment
322C</* Important ones in the first cache line (if alignment is done right) */>
323which implies that at some point in the past the ordering was carefully chosen
324(at least in part). However, it's clear that the ordering is less than perfect,
325as currently there are things such as 7 C<bool>s in a row, then something
326typically requiring 4 byte alignment, and then an odd C<bool> later on.
327(C<bool>s are typically defined as C<char>s). So it would be good for someone
328to review the ordering of the variables, to see how much alignment padding can
329be removed.
330
d7939546
NC
331It's also worth checking that all variables are actually used. Perl 5.8.0
332shipped with C<PL_nrs> still defined in F<thrdvar.h>, despite it being unused
333since a change over a year earlier. Had this been spotted before release, it
334could have been removed, but now it has to remain in the 5.8.x releases to
335keep the structure the same size, to retain binary compatibility.
336
c1ab7b38
NC
337It's probably worth checking if all need to be the types they are. For example
338
339 PERLVAR(Ierror_count, I32) /* how many errors so far, max 10 */
340
341might work as well if stored in a signed (or unsigned) 8 bit value, if the
342comment is accurate. C<PL_multi_open> and C<PL_multi_close> can probably
343become C<char>s. Finding variables to downsize coupled with rearrangement
344could shrink the interpreter structure; a size saving which is multiplied by
345the number of threads running.
346
fee0a0f7 347=head2 Profile Perl - am I hot or not?
62403a3c 348
fee0a0f7
NC
349The Perl source code is stable enough that it makes sense to profile it,
350identify and optimise the hotspots. It would be good to measure the
351performance of the Perl interpreter using free tools such as cachegrind,
352gprof, and dtrace, and work to reduce the bottlenecks they reveal.
353
354As part of this, the idea of F<pp_hot.c> is that it contains the I<hot> ops,
355the ops that are most commonly used. The idea is that by grouping them, their
356object code will be adjacent in the executable, so they have a greater chance
357of already being in the CPU cache (or swapped in) due to being near another op
358already in use.
62403a3c
NC
359
360Except that it's not clear if these really are the most commonly used ops. So
fee0a0f7
NC
361as part of exercising your skills with coverage and profiling tools you might
362want to determine what ops I<really> are the most commonly used. And in turn
363suggest evictions and promotions to achieve a better F<pp_hot.c>.
62403a3c 364
98fed0ad
NC
365=head2 Allocate OPs from arenas
366
367Currently all new OP structures are individually malloc()ed and free()d.
368All C<malloc> implementations have space overheads, and are now as fast as
369custom allocates so it would both use less memory and less CPU to allocate
370the various OP structures from arenas. The SV arena code can probably be
371re-used for this.
372
a229ae3b 373=head2 Improve win32/wince.c
0bdfc961 374
a229ae3b 375Currently, numerous functions look virtually, if not completely,
02f21748 376identical in both C<win32/wince.c> and C<win32/win32.c> files, which can't
6d71adcd
NC
377be good.
378
379=head1 Tasks that need a knowledge of XS
380
381These tasks would need C knowledge, and roughly the level of knowledge of
382the perl API that comes from writing modules that use XS to interface to
383C.
384
385=head2 shrink C<PVBM>s
386
387By removing unused elements and careful re-ordering, the structures for C<AV>s,
388C<HV>s, C<CV>s and C<GV>s have recently been shrunk considerably. C<PVIO>s
389probably aren't worth it, as typical programs don't use more than 8, and
390(at least) C<Filter::Util::Call> uses C<SvPVX>/C<SvCUR>/C<SvLEN> on a C<PVIO>,
391so it would mean code changes to modules on CPAN. C<PVBM>s might have some
392savings to win.
393
394=head2 autovivification
395
396Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;
397
398This task is incremental - even a little bit of work on it will help.
399
400=head2 Unicode in Filenames
401
402chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
403opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
404system, truncate, unlink, utime, -X. All these could potentially accept
405Unicode filenames either as input or output (and in the case of system
406and qx Unicode in general, as input or output to/from the shell).
407Whether a filesystem - an operating system pair understands Unicode in
408filenames varies.
409
410Known combinations that have some level of understanding include
411Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
412OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to
413create Unicode filenames, what forms of Unicode are accepted and used
414(UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
415and so on, varies. Finding the right level of interfacing to Perl
416requires some thought. Remember that an OS does not implicate a
417filesystem.
418
419(The Windows -C command flag "wide API support" has been at least
420temporarily retired in 5.8.1, and the -C has been repurposed, see
421L<perlrun>.)
422
423=head2 Unicode in %ENV
424
425Currently the %ENV entries are always byte strings.
426
427=head2 use less 'memory'
428
429Investigate trade offs to switch out perl's choices on memory usage.
430Particularly perl should be able to give memory back.
431
432This task is incremental - even a little bit of work on it will help.
433
434=head2 Re-implement C<:unique> in a way that is actually thread-safe
435
436The old implementation made bad assumptions on several levels. A good 90%
437solution might be just to make C<:unique> work to share the string buffer
438of SvPVs. That way large constant strings can be shared between ithreads,
439such as the configuration information in F<Config>.
440
441=head2 Make tainting consistent
442
443Tainting would be easier to use if it didn't take documented shortcuts and
444allow taint to "leak" everywhere within an expression.
445
446=head2 readpipe(LIST)
447
448system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
449running a shell. readpipe() (the function behind qx//) could be similarly
450extended.
451
452=head2 strcat(), strcpy(), strncat(), strncpy(), sprintf(), vsprintf()
453
454Maybe create a utility that checks after each libperl.a creation that
455none of the above (nor sprintf(), vsprintf(), or *SHUDDER* gets())
456ever creep back to libperl.a.
457
458 nm libperl.a | ./miniperl -alne '$o = $F[0] if /:$/; print "$o $F[1]" if $F[0] eq "U" && $F[1] =~ /^(?:strn?c(?:at|py)|v?sprintf|gets)$/'
459
460Note, of course, that this will only tell whether B<your> platform
461is using those naughty interfaces.
462
463=head2 Audit the code for destruction ordering assumptions
464
465Change 25773 notes
466
467 /* Need to check SvMAGICAL, as during global destruction it may be that
468 AvARYLEN(av) has been freed before av, and hence the SvANY() pointer
469 is now part of the linked list of SV heads, rather than pointing to
470 the original body. */
471 /* FIXME - audit the code for other bugs like this one. */
472
473adding the C<SvMAGICAL> check to
474
475 if (AvARYLEN(av) && SvMAGICAL(AvARYLEN(av))) {
476 MAGIC *mg = mg_find (AvARYLEN(av), PERL_MAGIC_arylen);
477
478Go through the core and look for similar assumptions that SVs have particular
479types, as all bets are off during global destruction.
480
0bdfc961 481=head1 Tasks that need a knowledge of the interpreter
3298bd4d 482
0bdfc961
NC
483These tasks would need C knowledge, and knowledge of how the interpreter works,
484or a willingness to learn.
3298bd4d 485
4fedb12c
RGS
486=head2 Implement $value ~~ 0 .. $range
487
488It would be nice to extend the syntax of the C<~~> operator to also
489understand numeric (and maybe alphanumeric) ranges.
490
d10fc472 491=head2 Attach/detach debugger from running program
1626a787 492
cd793d32
NC
493The old perltodo notes "With C<gdb>, you can attach the debugger to a running
494program if you pass the process ID. It would be good to do this with the Perl
0bdfc961
NC
495debugger on a running Perl program, although I'm not sure how it would be
496done." ssh and screen do this with named pipes in /tmp. Maybe we can too.
1626a787 497
0bdfc961
NC
498=head2 LVALUE functions for lists
499
500The old perltodo notes that lvalue functions don't work for list or hash
501slices. This would be good to fix.
502
503=head2 LVALUE functions in the debugger
504
505The old perltodo notes that lvalue functions don't work in the debugger. This
506would be good to fix.
507
0bdfc961
NC
508=head2 regexp optimiser optional
509
510The regexp optimiser is not optional. It should configurable to be, to allow
511its performance to be measured, and its bugs to be easily demonstrated.
512
02f21748
RGS
513=head2 delete &function
514
515Allow to delete functions. One can already undef them, but they're still
516in the stash.
517
0bdfc961
NC
518=head2 optional optimizer
519
520Make the peephole optimizer optional. Currently it performs two tasks as
521it walks the optree - genuine peephole optimisations, and necessary fixups of
522ops. It would be good to find an efficient way to switch out the
523optimisations whilst keeping the fixups.
524
525=head2 You WANT *how* many
526
527Currently contexts are void, scalar and list. split has a special mechanism in
528place to pass in the number of return values wanted. It would be useful to
529have a general mechanism for this, backwards compatible and little speed hit.
530This would allow proposals such as short circuiting sort to be implemented
531as a module on CPAN.
532
533=head2 lexical aliases
534
535Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>.
536
537=head2 entersub XS vs Perl
538
539At the moment pp_entersub is huge, and has code to deal with entering both
540perl and XS subroutines. Subroutine implementations rarely change between
541perl and XS at run time, so investigate using 2 ops to enter subs (one for
542XS, one for perl) and swap between if a sub is redefined.
2810d901
NC
543
544=head2 Self ties
545
546self ties are currently illegal because they caused too many segfaults. Maybe
547the causes of these could be tracked down and self-ties on all types re-
548instated.
0bdfc961
NC
549
550=head2 Optimize away @_
551
552The old perltodo notes "Look at the "reification" code in C<av.c>".
553
0bdfc961
NC
554=head2 What hooks would assertions need?
555
556Assertions are in the core, and work. However, assertions needed to be added
557as a core patch, rather than an XS module in ext, or a CPAN module, because
558the core has no hooks in the necessary places. It would be useful to
559investigate what hooks would need to be added to make it possible to provide
560the full assertion support from a CPAN module, so that we aren't constraining
561the imagination of future CPAN authors.
562
16fc99ce
NC
563=head2 Properly Unicode safe tokeniser and pads.
564
565The tokeniser isn't actually very UTF-8 clean. C<use utf8;> is a hack -
566variable names are stored in stashes as raw bytes, without the utf-8 flag
567set. The pad API only takes a C<char *> pointer, so that's all bytes too. The
568tokeniser ignores the UTF-8-ness of C<PL_rsfp>, or any SVs returned from
569source filters. All this could be fixed.
570
0bdfc961
NC
571=head1 Big projects
572
573Tasks that will get your name mentioned in the description of the "Highlights
574of 5.10"
575
576=head2 make ithreads more robust
577
4e577f8b 578Generally make ithreads more robust. See also L</iCOW>
0bdfc961
NC
579
580This task is incremental - even a little bit of work on it will help, and
581will be greatly appreciated.
582
6c047da7
YST
583One bit would be to write the missing code in sv.c:Perl_dirp_dup.
584
0bdfc961
NC
585=head2 iCOW
586
587Sarathy and Arthur have a proposal for an improved Copy On Write which
588specifically will be able to COW new ithreads. If this can be implemented
589it would be a good thing.
590
591=head2 (?{...}) closures in regexps
592
593Fix (or rewrite) the implementation of the C</(?{...})/> closures.
594
595=head2 A re-entrant regexp engine
596
597This will allow the use of a regex from inside (?{ }), (??{ }) and
598(?(?{ })|) constructs.
6bda09f9 599
6bda09f9
YO
600=head2 Add class set operations to regexp engine
601
602Apparently these are quite useful. Anyway, Jeffery Friedl wants them.
603
604demerphq has this on his todo list, but right at the bottom.
605
606