Commit | Line | Data |
7711098a |
1 | =head1 NAME |
2 | |
3 | perltodo - Perl TO-DO List |
4 | |
5 | =head1 DESCRIPTION |
e50bb9a1 |
6 | |
722d2a37 |
7 | This is a list of wishes for Perl. Send updates to |
e50bb9a1 |
8 | I<perl5-porters@perl.org>. If you want to work on any of these |
9 | projects, be sure to check the perl5-porters archives for past ideas, |
10 | flames, and propaganda. This will save you time and also prevent you |
11 | from implementing something that Larry has already vetoed. One set |
12 | of archives may be found at: |
13 | |
14 | http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/ |
15 | |
cd793d32 |
16 | =head1 assertions |
e50bb9a1 |
17 | |
cd793d32 |
18 | Clean up and finish support for assertions. See L<assertions>. |
e50bb9a1 |
19 | |
cd793d32 |
20 | =head1 iCOW |
e50bb9a1 |
21 | |
cd793d32 |
22 | Sarathy and Arthur have a proposal for an improved Copy On Write which |
23 | specifically will be able to COW new ithreads. If this can be implemented |
24 | it would be a good thing. |
e50bb9a1 |
25 | |
cd793d32 |
26 | =head1 (?{...}) closures in regexps |
4b3b956a |
27 | |
cd793d32 |
28 | Fix (or rewrite) the implementation of the C</(?{...})/> closures. |
4b3b956a |
29 | |
cee7ab84 |
30 | =head1 A re-entrant regexp engine |
31 | |
32 | This will allow the use of a regex from inside (?{ }), (??{ }) and |
33 | (?(?{ })|) constructs. |
34 | |
cd793d32 |
35 | =head1 pragmata |
e50bb9a1 |
36 | |
cd793d32 |
37 | =head2 lexical pragmas |
0562c0e3 |
38 | |
cd793d32 |
39 | Reimplement the mechanism of lexical pragmas to be more extensible. Fix |
40 | current pragmas that don't work well (or at all) with lexical scopes or in |
41 | run-time eval(STRING) (C<sort>, C<re>, C<encoding> for example). MJD has a |
42 | preliminary patch that implements this. |
0562c0e3 |
43 | |
cd793d32 |
44 | =head2 use less 'memory' |
f35392ae |
45 | |
cd793d32 |
46 | Investigate trade offs to switch out perl's choices on memory usage. |
47 | Particularly perl should be able to give memory back. |
f35392ae |
48 | |
cd793d32 |
49 | =head1 prototypes and functions |
c5fc23ff |
50 | |
cd793d32 |
51 | =head2 _ prototype character |
e50bb9a1 |
52 | |
cd793d32 |
53 | Study the possibility of adding a new prototype character, C<_>, meaning |
54 | "this argument defaults to $_". |
e50bb9a1 |
55 | |
cd793d32 |
56 | =head2 inlining autoloaded constants |
776f8809 |
57 | |
cd793d32 |
58 | Currently the optimiser can inline constants when expressed as subroutines |
59 | with prototype ($) that return a constant. Likewise, many packages wrapping |
60 | C libraries export lots of constants as subroutines which are AUTOLOADed on |
61 | demand. However, these have no prototypes, so can't be seen as constants by |
62 | the optimiser. Some way of cheaply (low syntax, low memory overhead) to the |
63 | perl compiler that a name is a constant would be great, so that it knows to |
64 | call the AUTOLOAD routine at compile time, and then inline the constant. |
776f8809 |
65 | |
cd793d32 |
66 | =head2 Finish off lvalue functions |
e50bb9a1 |
67 | |
cd793d32 |
68 | The old perltodo notes "They don't work in the debugger, and they don't work for |
69 | list or hash slices." |
e50bb9a1 |
70 | |
cd793d32 |
71 | =head1 Unicode and UTF8 |
e50bb9a1 |
72 | |
cd793d32 |
73 | =head2 Implicit Latin 1 => Unicode translation |
e50bb9a1 |
74 | |
cd793d32 |
75 | Conversions from byte strings to UTF-8 currently map high bit characters |
76 | to Unicode without translation (or, depending on how you look at it, by |
77 | implicitly assuming that the byte strings are in Latin-1). As perl assumes |
78 | the C locale by default, upgrading a string to UTF-8 may change the |
79 | meaning of its contents regarding character classes, case mapping, etc. |
80 | This should probably emit a warning (at least). |
e50bb9a1 |
81 | |
cd793d32 |
82 | =head2 UTF8 caching code |
e50bb9a1 |
83 | |
cd793d32 |
84 | The string position/offset cache is not optional. It should be. |
e50bb9a1 |
85 | |
938c8732 |
86 | =head2 Unicode in Filenames |
87 | |
88 | chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open, |
89 | opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen, |
90 | system, truncate, unlink, utime, -X. All these could potentially accept |
91 | Unicode filenames either as input or output (and in the case of system |
92 | and qx Unicode in general, as input or output to/from the shell). |
93 | Whether a filesystem - an operating system pair understands Unicode in |
94 | filenames varies. |
95 | |
96 | Known combinations that have some level of understanding include |
97 | Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac |
98 | OS X), NFS v4 is rumored to be Unicode, and of course Plan 9. How to |
99 | create Unicode filenames, what forms of Unicode are accepted and used |
100 | (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used, |
101 | and so on, varies. Finding the right level of interfacing to Perl |
102 | requires some thought. Remember that an OS does not implicate a |
103 | filesystem. |
104 | |
105 | (The Windows -C command flag "wide API support" has been at least |
106 | temporarily retired in 5.8.1, and the -C has been repurposed, see |
107 | L<perlrun>.) |
108 | |
109 | =head2 Unicode in %ENV |
110 | |
111 | Currently the %ENV entries are always byte strings. |
112 | |
cd793d32 |
113 | =head1 Regexps |
e50bb9a1 |
114 | |
cd793d32 |
115 | =head2 regexp optimiser optional |
e50bb9a1 |
116 | |
cd793d32 |
117 | The regexp optimiser is not optional. It should configurable to be, to allow |
118 | its performance to be measured, and its bugs to be easily demonstrated. |
e50bb9a1 |
119 | |
cd793d32 |
120 | =head1 POD |
e50bb9a1 |
121 | |
cd793d32 |
122 | =head2 POD -> HTML conversion still sucks |
e50bb9a1 |
123 | |
938c8732 |
124 | Which is crazy given just how simple POD purports to be, and how simple HTML |
adebf063 |
125 | can be. It's not actually I<as> simple as it sounds, particularly with the |
126 | flexibility POD allows for C<=item>, but it would be good to improve the |
127 | visual appeal of the HTML generated, and to avoid it having any validation |
128 | errors. See also L</make HTML install work>, as the layout of installation tree |
129 | is needed to improve the cross-linking. |
938c8732 |
130 | |
cd793d32 |
131 | =head1 Misc medium sized projects |
e50bb9a1 |
132 | |
cd793d32 |
133 | =head2 UNITCHECK |
e50bb9a1 |
134 | |
cd793d32 |
135 | Introduce a new special block, UNITCHECK, which is run at the end of a |
136 | compilation unit (module, file, eval(STRING) block). This will correspond to |
137 | the Perl 6 CHECK. Perl 5's CHECK cannot be changed or removed because the |
138 | O.pm/B.pm backend framework depends on it. |
e50bb9a1 |
139 | |
cd793d32 |
140 | =head2 optional optimizer |
e50bb9a1 |
141 | |
cd793d32 |
142 | Make the peephole optimizer optional. |
e50bb9a1 |
143 | |
0c13e809 |
144 | =head2 You WANT *how* many |
145 | |
60cb11a8 |
146 | Currently contexts are void, scalar and list. split has a special mechanism in |
0c13e809 |
147 | place to pass in the number of return values wanted. It would be useful to |
148 | have a general mechanism for this, backwards compatible and little speed hit. |
149 | This would allow proposals such as short circuiting sort to be implemented |
150 | as a module on CPAN. |
151 | |
cd793d32 |
152 | =head2 lexical aliases |
e50bb9a1 |
153 | |
cd793d32 |
154 | Allow lexical aliases (maybe via the syntax C<my \$alias = \$foo>. |
e50bb9a1 |
155 | |
cd793d32 |
156 | =head2 no 6 |
e50bb9a1 |
157 | |
cd793d32 |
158 | Make C<no 6> and C<no v6> work (opposite of C<use 5.005>, etc.). |
e50bb9a1 |
159 | |
cd793d32 |
160 | =head2 IPv6 |
3958b146 |
161 | |
cd793d32 |
162 | Clean this up. Check everything in core works |
e50bb9a1 |
163 | |
cd793d32 |
164 | =head2 entersub XS vs Perl |
e50bb9a1 |
165 | |
cd793d32 |
166 | At the moment pp_entersub is huge, and has code to deal with entering both |
167 | perl and and XS subroutines. Subroutine implementations rarely change between |
168 | perl and XS at run time, so investigate using 2 ops to enter subs (one for |
169 | XS, one for perl) and swap between if a sub is redefined. |
e50bb9a1 |
170 | |
cd793d32 |
171 | =head2 @INC source filter to Filter::Simple |
e50bb9a1 |
172 | |
cd793d32 |
173 | The second return value from a sub in @INC can be a source filter. This isn't |
174 | documented. It should be changed to use Filter::Simple, tested and documented. |
e50bb9a1 |
175 | |
cd793d32 |
176 | =head2 bincompat functions |
e50bb9a1 |
177 | |
cd793d32 |
178 | There are lots of functions which are retained for binary compatibility. |
179 | Clean these up. Move them to mathom.c, and don't compile for blead? |
e50bb9a1 |
180 | |
6168cf99 |
181 | =head2 Constant folding |
182 | |
183 | The peephole optimiser should trap errors during constant folding, and give |
184 | up on the folding, rather than bailing out at compile time. It is quite |
185 | possible that the unfoldable constant is in unreachable code, eg something |
186 | akin to C<$a = 0/0 if 0;> |
187 | |
cd793d32 |
188 | =head1 Tests |
e50bb9a1 |
189 | |
cd793d32 |
190 | =head2 Make Schwern poorer |
e50bb9a1 |
191 | |
cd793d32 |
192 | Tests for everything, At which point Schwern coughs up $500 to TPF. |
e50bb9a1 |
193 | |
cd793d32 |
194 | =head2 test B |
722d2a37 |
195 | |
cd793d32 |
196 | A test suite for the B module would be nice. |
722d2a37 |
197 | |
cd793d32 |
198 | =head2 common test code for timed bailout |
e50bb9a1 |
199 | |
cd793d32 |
200 | Write portable self destruct code for tests to stop them burning CPU in |
201 | infinite loops. Needs to avoid using alarm, as some of the tests are testing |
202 | alarm/sleep or timers. |
e50bb9a1 |
203 | |
cd793d32 |
204 | =head1 Installation |
e50bb9a1 |
205 | |
cd793d32 |
206 | =head2 compressed man pages |
e50bb9a1 |
207 | |
adebf063 |
208 | Be able to install them. This would probably need a configure test to see how |
209 | the system does compressed man pages (same directory/different directory? |
210 | same filename/different filename), as well as tweaking the F<installman> script |
211 | to compress as necessary. |
e50bb9a1 |
212 | |
cd793d32 |
213 | =head2 Make Config.pm cope with differences between build and installed perl |
e50bb9a1 |
214 | |
adebf063 |
215 | Quite often vendors ship a perl binary compiled with their (pay-for) |
216 | compilers. People install a free compiler, such as gcc. To work out how to |
217 | build extensions, Perl interrogates C<%Config>, so in this situation |
218 | C<%Config> describes compilers that aren't there, and extension building |
219 | fails. This forces people into chosing between re-compiling perl themselves |
220 | using the compiler they have, or only using modules that the vendor ships. |
221 | |
222 | It would be good to find a way teach C<Config.pm> about the installation setup, |
223 | possibly involving probing at install time or later, so that the C<%Config> in |
224 | a binary distruction better describes the installed machine, when the installed |
225 | machine differs from the build machine in some significant way. |
226 | |
cd793d32 |
227 | =head2 Relocatable perl |
e50bb9a1 |
228 | |
cd793d32 |
229 | Make it possible to create a relocatable perl binary. Will need some collusion |
230 | with Config.pm. We could use a syntax of ... for location of current binary? |
e50bb9a1 |
231 | |
cd793d32 |
232 | =head2 make HTML install work |
e50bb9a1 |
233 | |
adebf063 |
234 | There is an C<installhtml> target in the Makefile. It's marked as |
235 | "experimental". It would be good to get this tested, make it work reliably, and |
236 | remove the "experimental" tag. This would include |
237 | |
238 | =over 4 |
239 | |
240 | =item 1 |
241 | |
242 | Checking that cross linking between various parts of the documentation works. |
243 | In particular that links work between the modules (files with POD in F<lib/>) |
244 | and the core documentation (files in F<pod/>) |
245 | |
246 | =item 2 |
247 | |
248 | Work out how to split perlfunc into chunks, preferably one per function group, |
249 | preferably with general case code that could be used elsewhere. Challenges |
250 | here are correctly identifying the groups of functions that go together, and |
251 | making the right named external cross-links point to the right page. Things to |
252 | be aware of are C<-X>, groups such as C<getpwnam> to C<endservent>, two or |
253 | more C<=items> giving the different parameter lists, such as |
254 | |
255 | =item substr EXPR,OFFSET,LENGTH,REPLACEMENT |
256 | |
257 | =item substr EXPR,OFFSET,LENGTH |
258 | |
259 | =item substr EXPR,OFFSET |
260 | |
261 | and different parameter lists having different meanings. (eg C<select>) |
262 | |
263 | =back |
3a89a73c |
264 | |
89007cb3 |
265 | =head2 put patchlevel in -v |
266 | |
267 | Currently perl from p4/rsync ships with a patchlevel.h file that usually |
268 | defines one local patch, of the form "MAINT12345" or "RC1". The output of |
269 | perl -v doesn't report that a perl isn't an official release, and this |
270 | information can get lost in bugs reports. Because of this, the minor version |
fa11829f |
271 | isn't bumped up until RC time, to minimise the possibility of versions of perl |
89007cb3 |
272 | escaping that believe themselves to be newer than they actually are. |
273 | |
274 | It would be useful to find an elegant way to have the "this is an interim |
275 | maintenance release" or "this is a release candidate" in the terse -v output, |
276 | and have it so that it's easy for the pumpking to remove this just as the |
277 | release tarball is rolled up. This way the version pulled out of rsync would |
278 | always say "I'm a development release" and it would be safe to bump the |
279 | reported minor version as soon as a release ships, which would aid perl |
280 | developers. |
281 | |
cd793d32 |
282 | =head1 Incremental things |
e50bb9a1 |
283 | |
cd793d32 |
284 | Some tasks that don't need to get done in one big hit. |
e50bb9a1 |
285 | |
cd793d32 |
286 | =head2 autovivification |
e50bb9a1 |
287 | |
cd793d32 |
288 | Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict; |
e50bb9a1 |
289 | |
cd793d32 |
290 | =head2 fix tainting bugs |
e50bb9a1 |
291 | |
cd793d32 |
292 | Fix the bugs revealed by running the test suite with the C<-t> switch (via |
293 | C<make test.taintwarn>). |
e50bb9a1 |
294 | |
cd793d32 |
295 | =head2 Make tainting consistent |
e50bb9a1 |
296 | |
cd793d32 |
297 | Tainting would be easier to use if it didn't take documented shortcuts and allow |
298 | taint to "leak" everywhere within an expression. |
e50bb9a1 |
299 | |
969e704b |
300 | =head2 Dual life everything |
301 | |
302 | As part of the "dists" plan, anything that doesn't belong in the smallest perl |
303 | distribution needs to be dual lifed. Anything else can be too. |
304 | |
cd793d32 |
305 | =head1 Vague things |
e50bb9a1 |
306 | |
cd793d32 |
307 | Some more nebulous ideas |
e50bb9a1 |
308 | |
cd793d32 |
309 | =head2 threads |
e50bb9a1 |
310 | |
0abe3f7c |
311 | =over 4 |
312 | |
313 | =item * |
314 | |
315 | Re-implement C<:unique> in a way that is actualy thread-safe |
316 | |
317 | =item * |
318 | |
319 | Make C<threads::shared> share aggregates properly |
320 | |
321 | (these two may actually share approach, if not implementation |
322 | |
323 | =back |
324 | |
325 | Generally make threads more robust. See also L<iCOW> |
e50bb9a1 |
326 | |
cd793d32 |
327 | =head2 POSIX memory footprint |
e50bb9a1 |
328 | |
cd793d32 |
329 | Ilya observed that use POSIX; eats memory like there's no tomorrow, and at |
330 | various times worked to cut it down. There is probably still fat to cut out - |
331 | for example POSIX passes Exporter some very memory hungry data structures. |
e50bb9a1 |
332 | |
cd793d32 |
333 | =head2 Optimize away @_ |
f86a8bc5 |
334 | |
80b46460 |
335 | The old perltodo notes "Look at the "reification" code in C<av.c>". |
3298bd4d |
336 | |
cd793d32 |
337 | =head2 switch ops |
3298bd4d |
338 | |
cd793d32 |
339 | The old perltodo notes "Although we have C<Switch.pm> in core, Larry points to |
340 | the dormant C<nswitch> and C<cswitch> ops in F<pp.c>; using these opcodes would |
341 | be much faster." |
0562c0e3 |
342 | |
d10fc472 |
343 | =head2 Attach/detach debugger from running program |
1626a787 |
344 | |
cd793d32 |
345 | The old perltodo notes "With C<gdb>, you can attach the debugger to a running |
346 | program if you pass the process ID. It would be good to do this with the Perl |
347 | debugger on a running Perl program, although I'm not sure how it would be done." |
348 | ssh and screen do this with named pipes in tmp. Maybe we can too. |
1626a787 |
349 | |
d10fc472 |
350 | =head2 A decent benchmark |
351 | |
352 | perlbench seems impervious to any recent changes made to the perl core. It would |
353 | be useful to have a reasonable general benchmarking suite that roughly |
354 | represented what current perl programs do, and measurably reported whether |
355 | tweaks to the core improve, degrade or don't really affect performance, to |
356 | guide people attempting to optimise the guts of perl. |
80b46460 |
357 | |
358 | =head2 readpipe(LIST) |
359 | |
360 | system() accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid |
361 | running a shell. readpipe() (the function behind qx//) could be similarly |
362 | extended. |
2810d901 |
363 | |
364 | =head2 Self ties |
365 | |
366 | self ties are currently illegal because they caused too many segfaults. Maybe |
367 | the causes of these could be tracked down and self-ties on all types re- |
368 | instated. |