mingw/html/pod/perltodo.html

   1 <?xml version="1.0" ?>
   2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
   3 <html xmlns="http://www.w3.org/1999/xhtml">
   4 <head>
   5 <title>perltodo - Perl TO-DO List</title>
   6 <meta http-equiv="content-type" content="text/html; charset=utf-8" />
   7 <link rev="made" href="mailto:" />
   8 </head>
   9
  10 <body style="background-color: white">
  11 <table border="0" width="100%" cellspacing="0" cellpadding="3">
  12 <tr><td class="block" style="background-color: #cccccc" valign="middle">
  13 <big><strong><span class="block">&nbsp;perltodo - Perl TO-DO List</span></strong></big>
  14 </td></tr>
  15 </table>
  16
  17 <p><a name="__index__"></a></p>
  18 <!-- INDEX BEGIN -->
  19
  20 <ul>
  21
  22         <li><a href="#name">NAME</a></li>
  23         <li><a href="#description">DESCRIPTION</a></li>
  24         <li><a href="#the_roadmap_to_5_10">The roadmap to 5.10</a></li>
  25         <ul>
  26
  27                 <li><a href="#needed_for_a_5_9_4_release">Needed for a 5.9.4 release</a></li>
  28                 <li><a href="#needed_for_a_5_9_5_release">Needed for a 5.9.5 release</a></li>
  29                 <li><a href="#needed_for_a_5_9_6_release">Needed for a 5.9.6 release</a></li>
  30         </ul>
  31
  32         <li><a href="#tasks_that_only_need_perl_knowledge">Tasks that only need Perl knowledge</a></li>
  33         <ul>
  34
  35                 <li><a href="#common_test_code_for_timed_bail_out">common test code for timed bail out</a></li>
  36                 <li><a href="#pod___html_conversion_in_the_core_still_sucks">POD -&gt; HTML conversion in the core still sucks</a></li>
  37                 <li><a href="#parallel_testing">Parallel testing</a></li>
  38                 <li><a href="#make_schwern_poorer">Make Schwern poorer</a></li>
  39                 <li><a href="#improve_the_coverage_of_the_core_tests">Improve the coverage of the core tests</a></li>
  40                 <li><a href="#test_b">test B</a></li>
  41                 <li><a href="#a_decent_benchmark">A decent benchmark</a></li>
  42                 <li><a href="#fix_tainting_bugs">fix tainting bugs</a></li>
  43                 <li><a href="#dual_life_everything">Dual life everything</a></li>
  44                 <li><a href="#improving_threads__shared">Improving <code>threads::shared</code></a></li>
  45                 <li><a href="#posix_memory_footprint">POSIX memory footprint</a></li>
  46         </ul>
  47
  48         <li><a href="#tasks_that_need_a_little_sysadmintype_knowledge">Tasks that need a little sysadmin-type knowledge</a></li>
  49         <ul>
  50
  51                 <li><a href="#relocatable_perl">Relocatable perl</a></li>
  52                 <li><a href="#make_html_install_work">make HTML install work</a></li>
  53                 <li><a href="#compressed_man_pages">compressed man pages</a></li>
  54                 <li><a href="#add_a_code_coverage_target_to_the_makefile">Add a code coverage target to the Makefile</a></li>
  55                 <li><a href="#make_config_pm_cope_with_differences_between_build_and_installed_perl">Make Config.pm cope with differences between build and installed perl</a></li>
  56                 <li><a href="#make_parallel_builds_work">make parallel builds work</a></li>
  57                 <li><a href="#linker_specification_files">linker specification files</a></li>
  58         </ul>
  59
  60         <li><a href="#tasks_that_need_a_little_c_knowledge">Tasks that need a little C knowledge</a></li>
  61         <ul>
  62
  63                 <li><a href="#make_it_clear_from_v_if_this_is_the_exact_official_release">Make it clear from -v if this is the exact official release</a></li>
  64                 <li><a href="#tidy_up_global_variables">Tidy up global variables</a></li>
  65                 <li><a href="#ordering_of_global_variables_">Ordering of ``global'' variables.</a></li>
  66                 <li><a href="#bincompat_functions">bincompat functions</a></li>
  67                 <li><a href="#am_i_hot_or_not">am I hot or not?</a></li>
  68                 <li><a href="#emulate_the_perthread_memory_pool_on_unix">emulate the per-thread memory pool on Unix</a></li>
  69                 <li><a href="#reduce_duplication_in_sv_setsv_flags">reduce duplication in sv_setsv_flags</a></li>
  70         </ul>
  71
  72         <li><a href="#tasks_that_need_a_knowledge_of_xs">Tasks that need a knowledge of XS</a></li>
  73         <ul>
  74
  75                 <li><a href="#ipv6">IPv6</a></li>
  76                 <li><a href="#shrink_gvs__cvs">shrink <code>GV</code>s, <code>CV</code>s</a></li>
  77                 <li><a href="#merge_perl_sv_2_inpu_v">merge Perl_sv_2[inpu]v</a></li>
  78                 <li><a href="#utf8_caching_code">UTF8 caching code</a></li>
  79                 <li><a href="#implicit_latin_1____unicode_translation">Implicit Latin 1 =&gt; Unicode translation</a></li>
  80                 <li><a href="#autovivification">autovivification</a></li>
  81                 <li><a href="#unicode_in_filenames">Unicode in Filenames</a></li>
  82                 <li><a href="#unicode_in__env">Unicode in %ENV</a></li>
  83                 <li><a href="#use_less__memory_">use less 'memory'</a></li>
  84                 <li><a href="#reimplement__unique_in_a_way_that_is_actually_threadsafe">Re-implement <code>:unique</code> in a way that is actually thread-safe</a></li>
  85                 <li><a href="#make_tainting_consistent">Make tainting consistent</a></li>
  86                 <li><a href="#readpipe_list_"><code>readpipe(LIST)</code></a></li>
  87         </ul>
  88
  89         <li><a href="#tasks_that_need_a_knowledge_of_the_interpreter">Tasks that need a knowledge of the interpreter</a></li>
  90         <ul>
  91
  92                 <li><a href="#lexical_pragmas">lexical pragmas</a></li>
  93                 <li><a href="#attach_detach_debugger_from_running_program">Attach/detach debugger from running program</a></li>
  94                 <li><a href="#constant_folding">Constant folding</a></li>
  95                 <li><a href="#lvalue_functions_for_lists">LVALUE functions for lists</a></li>
  96                 <li><a href="#lvalue_functions_in_the_debugger">LVALUE functions in the debugger</a></li>
  97                 <li><a href="#__prototype_character">_ prototype character</a></li>
  98                 <li><a href="#state_variables">state variables</a></li>
  99                 <li><a href="#_inc_source_filter_to_filter__simple">@INC source filter to Filter::Simple</a></li>
 100                 <li><a href="#regexp_optimiser_optional">regexp optimiser optional</a></li>
 101                 <li><a href="#unitcheck">UNITCHECK</a></li>
 102                 <li><a href="#optional_optimizer">optional optimizer</a></li>
 103                 <li><a href="#you_want__how__many">You WANT *how* many</a></li>
 104                 <li><a href="#lexical_aliases">lexical aliases</a></li>
 105                 <li><a href="#entersub_xs_vs_perl">entersub XS vs Perl</a></li>
 106                 <li><a href="#self_ties">Self ties</a></li>
 107                 <li><a href="#optimize_away___">Optimize away @_</a></li>
 108                 <li><a href="#what_hooks_would_assertions_need">What hooks would assertions need?</a></li>
 109         </ul>
 110
 111         <li><a href="#big_projects">Big projects</a></li>
 112         <ul>
 113
 114                 <li><a href="#make_ithreads_more_robust">make ithreads more robust</a></li>
 115                 <li><a href="#icow">iCOW</a></li>
 116                 <li><a href="#________closures_in_regexps">(?{...}) closures in regexps</a></li>
 117                 <li><a href="#a_reentrant_regexp_engine">A re-entrant regexp engine</a></li>
 118         </ul>
 119
 120 </ul>
 121 <!-- INDEX END -->
 122
 123 <hr />
 124 <p>
 125 </p>
 126 <h1><a name="name">NAME</a></h1>
 127 <p>perltodo - Perl TO-DO List</p>
 128 <p>
 129 </p>
 130 <hr />
 131 <h1><a name="description">DESCRIPTION</a></h1>
 132 <p>This is a list of wishes for Perl. The tasks we think are smaller or easier
 133 are listed first. Anyone is welcome to work on any of these, but it's a good
 134 idea to first contact <em><a href="mailto:perl5-porters@perl.org">perl5-porters@perl.org</a></em> to avoid duplication of
 135 effort. By all means contact a pumpking privately first if you prefer.</p>
 136 <p>Whilst patches to make the list shorter are most welcome, ideas to add to
 137 the list are also encouraged. Check the perl5-porters archives for past
 138 ideas, and any discussion about them. One set of archives may be found at:</p>
 139 <pre>
 140     <a href="http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/">http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/</a></pre>
 141 <p>What can we offer you in return? Fame, fortune, and everlasting glory? Maybe
 142 not, but if your patch is incorporated, then we'll add your name to the
 143 <em>AUTHORS</em> file, which ships in the official distribution. How many other
 144 programming languages offer you 1 line of immortality?</p>
 145 <p>
 146 </p>
 147 <hr />
 148 <h1><a name="the_roadmap_to_5_10">The roadmap to 5.10</a></h1>
 149 <p>The roadmap to 5.10 envisages feature based releases, as various items in this
 150 TODO are completed.</p>
 151 <p>
 152 </p>
 153 <h2><a name="needed_for_a_5_9_4_release">Needed for a 5.9.4 release</a></h2>
 154 <ul>
 155 <li>
 156 <p>Review assertions. Review syntax to combine assertions. Assertions could take
 157 advantage of the lexical pragmas work. <a href="#what_hooks_would_assertions_need">What hooks would assertions need?</a></p>
 158 </li>
 159 </ul>
 160 <p>
 161 </p>
 162 <h2><a name="needed_for_a_5_9_5_release">Needed for a 5.9.5 release</a></h2>
 163 <ul>
 164 <li><strong><a name="item_implement_l_3c_2f__prototype_character_3e">Implement <a href="#__prototype_character">_ prototype character</a></a></strong>
 165
 166 <li><strong><a name="item_implement_l_3c_2fstate_variables_3e">Implement <a href="#state_variables">state variables</a></a></strong>
 167
 168 </ul>
 169 <p>
 170 </p>
 171 <h2><a name="needed_for_a_5_9_6_release">Needed for a 5.9.6 release</a></h2>
 172 <p>Stabilisation. If all goes well, this will be the equivalent of a 5.10-beta.</p>
 173 <p>
 174 </p>
 175 <hr />
 176 <h1><a name="tasks_that_only_need_perl_knowledge">Tasks that only need Perl knowledge</a></h1>
 177 <p>
 178 </p>
 179 <h2><a name="common_test_code_for_timed_bail_out">common test code for timed bail out</a></h2>
 180 <p>Write portable self destruct code for tests to stop them burning CPU in
 181 infinite loops. This needs to avoid using alarm, as some of the tests are
 182 testing alarm/sleep or timers.</p>
 183 <p>
 184 </p>
 185 <h2><a name="pod___html_conversion_in_the_core_still_sucks">POD -&gt; HTML conversion in the core still sucks</a></h2>
 186 <p>Which is crazy given just how simple POD purports to be, and how simple HTML
 187 can be. It's not actually <em>as</em> simple as it sounds, particularly with the
 188 flexibility POD allows for <code>=item</code>, but it would be good to improve the
 189 visual appeal of the HTML generated, and to avoid it having any validation
 190 errors. See also <a href="#make_html_install_work">make HTML install work</a>, as the layout of installation tree
 191 is needed to improve the cross-linking.</p>
 192 <p>The addition of <code>Pod::Simple</code> and its related modules may make this task
 193 easier to complete.</p>
 194 <p>
 195 </p>
 196 <h2><a name="parallel_testing">Parallel testing</a></h2>
 197 <p>The core regression test suite is getting ever more comprehensive, which has
 198 the side effect that it takes longer to run. This isn't so good. Investigate
 199 whether it would be feasible to give the harness script the <strong>option</strong> of
 200 running sets of tests in parallel. This would be useful for tests in
 201 <em>t/op/*.t</em> and <em>t/uni/*.t</em> and maybe some sets of tests in <em>lib/</em>.</p>
 202 <p>Questions to answer</p>
 203 <ol>
 204 <li>
 205 <p>How does screen layout work when you're running more than one test?</p>
 206 </li>
 207 <li>
 208 <p>How does the caller of test specify how many tests to run in parallel?</p>
 209 </li>
 210 <li>
 211 <p>How do setup/teardown tests identify themselves?</p>
 212 </li>
 213 </ol>
 214 <p>Pugs already does parallel testing - can their approach be re-used?</p>
 215 <p>
 216 </p>
 217 <h2><a name="make_schwern_poorer">Make Schwern poorer</a></h2>
 218 <p>We should have for everything. When all the core's modules are tested,
 219 Schwern has promised to donate to $500 to TPF. We may need volunteers to
 220 hold him upside down and shake vigorously in order to actually extract the
 221 cash.</p>
 222 <p>See <em>t/lib/1_compile.t</em> for the 3 remaining modules that need tests.</p>
 223 <p>
 224 </p>
 225 <h2><a name="improve_the_coverage_of_the_core_tests">Improve the coverage of the core tests</a></h2>
 226 <p>Use Devel::Cover to ascertain the core's test coverage, then add tests that
 227 are currently missing.</p>
 228 <p>
 229 </p>
 230 <h2><a name="test_b">test B</a></h2>
 231 <p>A full test suite for the B module would be nice.</p>
 232 <p>
 233 </p>
 234 <h2><a name="a_decent_benchmark">A decent benchmark</a></h2>
 235 <p><code>perlbench</code> seems impervious to any recent changes made to the perl core. It
 236 would be useful to have a reasonable general benchmarking suite that roughly
 237 represented what current perl programs do, and measurably reported whether
 238 tweaks to the core improve, degrade or don't really affect performance, to
 239 guide people attempting to optimise the guts of perl. Gisle would welcome
 240 new tests for perlbench.</p>
 241 <p>
 242 </p>
 243 <h2><a name="fix_tainting_bugs">fix tainting bugs</a></h2>
 244 <p>Fix the bugs revealed by running the test suite with the <code>-t</code> switch (via
 245 <code>make test.taintwarn</code>).</p>
 246 <p>
 247 </p>
 248 <h2><a name="dual_life_everything">Dual life everything</a></h2>
 249 <p>As part of the ``dists'' plan, anything that doesn't belong in the smallest perl
 250 distribution needs to be dual lifed. Anything else can be too. Figure out what
 251 changes would be needed to package that module and its tests up for CPAN, and
 252 do so. Test it with older perl releases, and fix the problems you find.</p>
 253 <p>
 254 </p>
 255 <h2><a name="improving_threads__shared">Improving <code>threads::shared</code></a></h2>
 256 <p>Investigate whether <code>threads::shared</code> could share aggregates properly with
 257 only Perl level changes to shared.pm</p>
 258 <p>
 259 </p>
 260 <h2><a name="posix_memory_footprint">POSIX memory footprint</a></h2>
 261 <p>Ilya observed that use POSIX; eats memory like there's no tomorrow, and at
 262 various times worked to cut it down. There is probably still fat to cut out -
 263 for example POSIX passes Exporter some very memory hungry data structures.</p>
 264 <p>
 265 </p>
 266 <hr />
 267 <h1><a name="tasks_that_need_a_little_sysadmintype_knowledge">Tasks that need a little sysadmin-type knowledge</a></h1>
 268 <p>Or if you prefer, tasks that you would learn from, and broaden your skills
 269 base...</p>
 270 <p>
 271 </p>
 272 <h2><a name="relocatable_perl">Relocatable perl</a></h2>
 273 <p>The C level patches needed to create a relocatable perl binary are done, as
 274 is the work on <em>Config.pm</em>. All that's left to do is the <code>Configure</code> tweaking
 275 to let people specify how they want to do the install.</p>
 276 <p>
 277 </p>
 278 <h2><a name="make_html_install_work">make HTML install work</a></h2>
 279 <p>There is an <code>installhtml</code> target in the Makefile. It's marked as
 280 ``experimental''. It would be good to get this tested, make it work reliably, and
 281 remove the ``experimental'' tag. This would include</p>
 282 <ol>
 283 <li>
 284 <p>Checking that cross linking between various parts of the documentation works.
 285 In particular that links work between the modules (files with POD in <em>lib/</em>)
 286 and the core documentation (files in <em>pod/</em>)</p>
 287 </li>
 288 <li>
 289 <p>Work out how to split <code>perlfunc</code> into chunks, preferably one per function
 290 group, preferably with general case code that could be used elsewhere.
 291 Challenges here are correctly identifying the groups of functions that go
 292 together, and making the right named external cross-links point to the right
 293 page. Things to be aware of are <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item__2dx"><code>-X</code></a>, groups such as <code>getpwnam</code> to
 294 <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_endservent"><code>endservent</code></a>, two or more <code>=items</code> giving the different parameter lists, such
 295 as</p>
 296 <pre>
 297     =item substr EXPR,OFFSET,LENGTH,REPLACEMENT
 298
 299     =item substr EXPR,OFFSET,LENGTH
 300
 301     =item substr EXPR,OFFSET</pre>
 302 <p>and different parameter lists having different meanings. (eg <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_select"><code>select</code></a>)</p>
 303 </li>
 304 </ol>
 305 <p>
 306 </p>
 307 <h2><a name="compressed_man_pages">compressed man pages</a></h2>
 308 <p>Be able to install them. This would probably need a configure test to see how
 309 the system does compressed man pages (same directory/different directory?
 310 same filename/different filename), as well as tweaking the <em>installman</em> script
 311 to compress as necessary.</p>
 312 <p>
 313 </p>
 314 <h2><a name="add_a_code_coverage_target_to_the_makefile">Add a code coverage target to the Makefile</a></h2>
 315 <p>Make it easy for anyone to run Devel::Cover on the core's tests. The steps
 316 to do this manually are roughly</p>
 317 <ul>
 318 <li>
 319 <p>do a normal <code>Configure</code>, but include Devel::Cover as a module to install
 320 (see <em>INSTALL</em> for how to do this)</p>
 321 </li>
 322 <li>
 323 <pre>
 324     make perl</pre>
 325 <li>
 326 <pre>
 327     cd t; HARNESS_PERL_SWITCHES=-MDevel::Cover ./perl -I../lib harness</pre>
 328 <li>
 329 <p>Process the resulting Devel::Cover database</p>
 330 </li>
 331 </ul>
 332 <p>This just give you the coverage of the <em>.pm</em>s. To also get the C level
 333 coverage you need to</p>
 334 <ul>
 335 <li>
 336 <p>Additionally tell <code>Configure</code> to use the appropriate C compiler flags for
 337 <code>gcov</code></p>
 338 </li>
 339 <li>
 340 <pre>
 341     make perl.gcov</pre>
 342 <p>(instead of <code>make perl</code>)</p>
 343 <li>
 344 <p>After running the tests run <code>gcov</code> to generate all the <em>.gcov</em> files.
 345 (Including down in the subdirectories of <em>ext/</em></p>
 346 </li>
 347 <li>
 348 <p>(From the top level perl directory) run <code>gcov2perl</code> on all the <code>.gcov</code> files
 349 to get their stats into the cover_db directory.</p>
 350 </li>
 351 <li>
 352 <p>Then process the Devel::Cover database</p>
 353 </li>
 354 </ul>
 355 <p>It would be good to add a single switch to <code>Configure</code> to specify that you
 356 wanted to perform perl level coverage, and another to specify C level
 357 coverage, and have <code>Configure</code> and the <em>Makefile</em> do all the right things
 358 automatically.</p>
 359 <p>
 360 </p>
 361 <h2><a name="make_config_pm_cope_with_differences_between_build_and_installed_perl">Make Config.pm cope with differences between build and installed perl</a></h2>
 362 <p>Quite often vendors ship a perl binary compiled with their (pay-for)
 363 compilers.  People install a free compiler, such as gcc. To work out how to
 364 build extensions, Perl interrogates <code>%Config</code>, so in this situation
 365 <code>%Config</code> describes compilers that aren't there, and extension building
 366 fails. This forces people into choosing between re-compiling perl themselves
 367 using the compiler they have, or only using modules that the vendor ships.</p>
 368 <p>It would be good to find a way teach <code>Config.pm</code> about the installation setup,
 369 possibly involving probing at install time or later, so that the <code>%Config</code> in
 370 a binary distribution better describes the installed machine, when the
 371 installed machine differs from the build machine in some significant way.</p>
 372 <p>
 373 </p>
 374 <h2><a name="make_parallel_builds_work">make parallel builds work</a></h2>
 375 <p>Currently parallel builds (such as <code>make -j3</code>) don't work reliably. We believe
 376 that this is due to incomplete dependency specification in the <em>Makefile</em>.
 377 It would be good if someone were able to track down the causes of these
 378 problems, so that parallel builds worked properly.</p>
 379 <p>
 380 </p>
 381 <h2><a name="linker_specification_files">linker specification files</a></h2>
 382 <p>Some platforms mandate that you provide a list of a shared library's external
 383 symbols to the linker, so the core already has the infrastructure in place to
 384 do this for generating shared perl libraries. My understanding is that the
 385 GNU toolchain can accept an optional linker specification file, and restrict
 386 visibility just to symbols declared in that file. It would be good to extend
 387 <em>makedef.pl</em> to support this format, and to provide a means within
 388 <code>Configure</code> to enable it. This would allow Unix users to test that the
 389 export list is correct, and to build a perl that does not pollute the global
 390 namespace with private symbols.</p>
 391 <p>
 392 </p>
 393 <hr />
 394 <h1><a name="tasks_that_need_a_little_c_knowledge">Tasks that need a little C knowledge</a></h1>
 395 <p>These tasks would need a little C knowledge, but don't need any specific
 396 background or experience with XS, or how the Perl interpreter works</p>
 397 <p>
 398 </p>
 399 <h2><a name="make_it_clear_from_v_if_this_is_the_exact_official_release">Make it clear from -v if this is the exact official release</a></h2>
 400 <p>Currently perl from <code>p4</code>/<code>rsync</code> ships with a <em>patchlevel.h</em> file that
 401 usually defines one local patch, of the form ``MAINT12345'' or ``RC1''. The output
 402 of perl -v doesn't report that a perl isn't an official release, and this
 403 information can get lost in bugs reports. Because of this, the minor version
 404 isn't bumped up until RC time, to minimise the possibility of versions of perl
 405 escaping that believe themselves to be newer than they actually are.</p>
 406 <p>It would be useful to find an elegant way to have the ``this is an interim
 407 maintenance release'' or ``this is a release candidate'' in the terse -v output,
 408 and have it so that it's easy for the pumpking to remove this just as the
 409 release tarball is rolled up. This way the version pulled out of rsync would
 410 always say ``I'm a development release'' and it would be safe to bump the
 411 reported minor version as soon as a release ships, which would aid perl
 412 developers.</p>
 413 <p>This task is really about thinking of an elegant way to arrange the C source
 414 such that it's trivial for the Pumpking to flag ``this is an official release''
 415 when making a tarball, yet leave the default source saying ``I'm not the
 416 official release''.</p>
 417 <p>
 418 </p>
 419 <h2><a name="tidy_up_global_variables">Tidy up global variables</a></h2>
 420 <p>There's a note in <em>intrpvar.h</em></p>
 421 <pre>
 422   /* These two variables are needed to preserve 5.8.x bincompat because
 423      we can't change function prototypes of two exported functions.
 424      Probably should be taken out of blead soon, and relevant prototypes
 425      changed.  */</pre>
 426 <p>So doing this, and removing any of the unused variables still present would
 427 be good.</p>
 428 <p>
 429 </p>
 430 <h2><a name="ordering_of_global_variables_">Ordering of ``global'' variables.</a></h2>
 431 <p><em>thrdvar.h</em> and <em>intrpvarh</em> define the ``global'' variables that need to be
 432 per-thread under ithreads, where the variables are actually elements in a
 433 structure. As C dictates, the variables must be laid out in order of
 434 declaration. There is a comment
 435 <code>/* Important ones in the first cache line (if alignment is done right) */</code>
 436 which implies that at some point in the past the ordering was carefully chosen
 437 (at least in part). However, it's clear that the ordering is less than perfect,
 438 as currently there are things such as 7 <code>bool</code>s in a row, then something
 439 typically requiring 4 byte alignment, and then an odd <code>bool</code> later on.
 440 (<code>bool</code>s are typically defined as <code>char</code>s). So it would be good for someone
 441 to review the ordering of the variables, to see how much alignment padding can
 442 be removed.</p>
 443 <p>
 444 </p>
 445 <h2><a name="bincompat_functions">bincompat functions</a></h2>
 446 <p>There are lots of functions which are retained for binary compatibility.
 447 Clean these up. Move them to mathom.c, and don't compile for blead?</p>
 448 <p>
 449 </p>
 450 <h2><a name="am_i_hot_or_not">am I hot or not?</a></h2>
 451 <p>The idea of <em>pp_hot.c</em> is that it contains the <em>hot</em> ops, the ops that are
 452 most commonly used. The idea is that by grouping them, their object code will
 453 be adjacent in the executable, so they have a greater chance of already being
 454 in the CPU cache (or swapped in) due to being near another op already in use.</p>
 455 <p>Except that it's not clear if these really are the most commonly used ops. So
 456 anyone feeling like exercising their skill with coverage and profiling tools
 457 might want to determine what ops <em>really</em> are the most commonly used. And in
 458 turn suggest evictions and promotions to achieve a better <em>pp_hot.c</em>.</p>
 459 <p>
 460 </p>
 461 <h2><a name="emulate_the_perthread_memory_pool_on_unix">emulate the per-thread memory pool on Unix</a></h2>
 462 <p>For Windows, ithreads allocates memory for each thread from a separate pool,
 463 which it discards at thread exit. It also checks that memory is free()d to
 464 the correct pool. Neither check is done on Unix, so code developed there won't
 465 be subject to such strictures, so can harbour bugs that only show up when the
 466 code reaches Windows.</p>
 467 <p>It would be good to be able to optionally emulate the Window pool system on
 468 Unix, to let developers who only have access to Unix, or want to use
 469 Unix-specific debugging tools, check for these problems. To do this would
 470 involve figuring out how the <code>PerlMem_*</code> macros wrap <code>malloc()</code> access, and
 471 providing a layer that records/checks the identity of the thread making the
 472 call, and recording all the memory allocated by each thread via this API so
 473 that it can be summarily free()d at thread exit. One implementation idea
 474 would be to increase the size of allocation, and store the <code>my_perl</code> pointer
 475 (to identify the thread) at the start, along with pointers to make a linked
 476 list of blocks for this thread. To avoid alignment problems it would be
 477 necessary to do something like</p>
 478 <pre>
 479   union memory_header_padded {
 480     struct memory_header {
 481       void *thread_id;   /* For my_perl */
 482       void *next;        /* Pointer to next block for this thread */
 483     } data;
 484     long double padding; /* whatever type has maximal alignment constraint */
 485   };</pre>
 486 <p>although <code>long double</code> might not be the only type to add to the padding
 487 union.</p>
 488 <p>
 489 </p>
 490 <h2><a name="reduce_duplication_in_sv_setsv_flags">reduce duplication in sv_setsv_flags</a></h2>
 491 <p><code>Perl_sv_setsv_flags</code> has a comment
 492 <code>/* There's a lot of redundancy below but we're going for speed here */</code></p>
 493 <p>Whilst this was true 10 years ago, the growing disparity between RAM and CPU
 494 speeds mean that the trade offs have changed. In addition, the duplicate code
 495 adds to the maintenance burden. It would be good to see how much of the
 496 redundancy can be pruned, particular in the less common paths. (Profiling
 497 tools at the ready...). For example, why does the test for
 498 ``Can't redefine active sort subroutine'' need to occur in two places?</p>
 499 <p>
 500 </p>
 501 <hr />
 502 <h1><a name="tasks_that_need_a_knowledge_of_xs">Tasks that need a knowledge of XS</a></h1>
 503 <p>These tasks would need C knowledge, and roughly the level of knowledge of
 504 the perl API that comes from writing modules that use XS to interface to
 505 C.</p>
 506 <p>
 507 </p>
 508 <h2><a name="ipv6">IPv6</a></h2>
 509 <p>Clean this up. Check everything in core works</p>
 510 <p>
 511 </p>
 512 <h2><a name="shrink_gvs__cvs">shrink <code>GV</code>s, <code>CV</code>s</a></h2>
 513 <p>By removing unused elements and careful re-ordering, the structures for <code>AV</code>s
 514 and <code>HV</code>s have recently been shrunk considerably. It's probable that the same
 515 approach would find savings in <code>GV</code>s and <code>CV</code>s, if not all the other
 516 larger-than-<code>PVMG</code> types.</p>
 517 <p>
 518 </p>
 519 <h2><a name="merge_perl_sv_2_inpu_v">merge Perl_sv_2[inpu]v</a></h2>
 520 <p>There's a lot of code shared between <code>Perl_sv_2iv_flags</code>,
 521 <code>Perl_sv_2uv_flags</code>, <code>Perl_sv_2nv</code>, and <code>Perl_sv_2pv_flags</code>. It would be
 522 interesting to see if some of it can be merged into common shared static
 523 functions. In particular, <code>Perl_sv_2uv_flags</code> started out as a cut&amp;paste
 524 from <code>Perl_sv_2iv_flags</code> around 5.005_50 time, and it may be possible to
 525 replace both with a single function that returns a value or union which is
 526 split out by the macros in <em>sv.h</em></p>
 527 <p>
 528 </p>
 529 <h2><a name="utf8_caching_code">UTF8 caching code</a></h2>
 530 <p>The string position/offset cache is not optional. It should be.</p>
 531 <p>
 532 </p>
 533 <h2><a name="implicit_latin_1____unicode_translation">Implicit Latin 1 =&gt; Unicode translation</a></h2>
 534 <p>Conversions from byte strings to UTF-8 currently map high bit characters
 535 to Unicode without translation (or, depending on how you look at it, by
 536 implicitly assuming that the byte strings are in Latin-1). As perl assumes
 537 the C locale by default, upgrading a string to UTF-8 may change the
 538 meaning of its contents regarding character classes, case mapping, etc.
 539 This should probably emit a warning (at least).</p>
 540 <p>This task is incremental - even a little bit of work on it will help.</p>
 541 <p>
 542 </p>
 543 <h2><a name="autovivification">autovivification</a></h2>
 544 <p>Make all autovivification consistent w.r.t LVALUE/RVALUE and strict/no strict;</p>
 545 <p>This task is incremental - even a little bit of work on it will help.</p>
 546 <p>
 547 </p>
 548 <h2><a name="unicode_in_filenames">Unicode in Filenames</a></h2>
 549 <p>chdir, chmod, chown, chroot, exec, glob, link, lstat, mkdir, open,
 550 opendir, qx, readdir, readlink, rename, rmdir, stat, symlink, sysopen,
 551 system, truncate, unlink, utime, -X.  All these could potentially accept
 552 Unicode filenames either as input or output (and in the case of system
 553 and qx Unicode in general, as input or output to/from the shell).
 554 Whether a filesystem - an operating system pair understands Unicode in
 555 filenames varies.</p>
 556 <p>Known combinations that have some level of understanding include
 557 Microsoft NTFS, Apple HFS+ (In Mac OS 9 and X) and Apple UFS (in Mac
 558 OS X), NFS v4 is rumored to be Unicode, and of course Plan 9.  How to
 559 create Unicode filenames, what forms of Unicode are accepted and used
 560 (UCS-2, UTF-16, UTF-8), what (if any) is the normalization form used,
 561 and so on, varies.  Finding the right level of interfacing to Perl
 562 requires some thought.  Remember that an OS does not implicate a
 563 filesystem.</p>
 564 <p>(The Windows -C command flag ``wide API support'' has been at least
 565 temporarily retired in 5.8.1, and the -C has been repurposed, see
 566 <a href="file://C|\msysgit\mingw\html/pod/perlrun.html">the perlrun manpage</a>.)</p>
 567 <p>
 568 </p>
 569 <h2><a name="unicode_in__env">Unicode in %ENV</a></h2>
 570 <p>Currently the %ENV entries are always byte strings.</p>
 571 <p>
 572 </p>
 573 <h2><a name="use_less__memory_">use less 'memory'</a></h2>
 574 <p>Investigate trade offs to switch out perl's choices on memory usage.
 575 Particularly perl should be able to give memory back.</p>
 576 <p>This task is incremental - even a little bit of work on it will help.</p>
 577 <p>
 578 </p>
 579 <h2><a name="reimplement__unique_in_a_way_that_is_actually_threadsafe">Re-implement <code>:unique</code> in a way that is actually thread-safe</a></h2>
 580 <p>The old implementation made bad assumptions on several levels. A good 90%
 581 solution might be just to make <code>:unique</code> work to share the string buffer
 582 of SvPVs. That way large constant strings can be shared between ithreads,
 583 such as the configuration information in <em>Config</em>.</p>
 584 <p>
 585 </p>
 586 <h2><a name="make_tainting_consistent">Make tainting consistent</a></h2>
 587 <p>Tainting would be easier to use if it didn't take documented shortcuts and
 588 allow taint to ``leak'' everywhere within an expression.</p>
 589 <p>
 590 </p>
 591 <h2><a name="readpipe_list_"><code>readpipe(LIST)</code></a></h2>
 592 <p><a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_system"><code>system()</code></a> accepts a LIST syntax (and a PROGRAM LIST syntax) to avoid
 593 running a shell. <code>readpipe()</code> (the function behind qx//) could be similarly
 594 extended.</p>
 595 <p>
 596 </p>
 597 <hr />
 598 <h1><a name="tasks_that_need_a_knowledge_of_the_interpreter">Tasks that need a knowledge of the interpreter</a></h1>
 599 <p>These tasks would need C knowledge, and knowledge of how the interpreter works,
 600 or a willingness to learn.</p>
 601 <p>
 602 </p>
 603 <h2><a name="lexical_pragmas">lexical pragmas</a></h2>
 604 <p>Document the new support for lexical pragmas in 5.9.3 and how %^H works.
 605 Maybe <code>re</code>, <code>encoding</code>, maybe other pragmas could be made lexical.</p>
 606 <p>
 607 </p>
 608 <h2><a name="attach_detach_debugger_from_running_program">Attach/detach debugger from running program</a></h2>
 609 <p>The old perltodo notes ``With <code>gdb</code>, you can attach the debugger to a running
 610 program if you pass the process ID. It would be good to do this with the Perl
 611 debugger on a running Perl program, although I'm not sure how it would be
 612 done.'' ssh and screen do this with named pipes in /tmp. Maybe we can too.</p>
 613 <p>
 614 </p>
 615 <h2><a name="constant_folding">Constant folding</a></h2>
 616 <p>The peephole optimiser should trap errors during constant folding, and give
 617 up on the folding, rather than bailing out at compile time.  It is quite
 618 possible that the unfoldable constant is in unreachable code, eg something
 619 akin to <a href="file://C|\msysgit\mingw\html/pod/perlvar.html#item__a"><code>$a = 0/0 if 0;</code></a></p>
 620 <p>
 621 </p>
 622 <h2><a name="lvalue_functions_for_lists">LVALUE functions for lists</a></h2>
 623 <p>The old perltodo notes that lvalue functions don't work for list or hash
 624 slices. This would be good to fix.</p>
 625 <p>
 626 </p>
 627 <h2><a name="lvalue_functions_in_the_debugger">LVALUE functions in the debugger</a></h2>
 628 <p>The old perltodo notes that lvalue functions don't work in the debugger. This
 629 would be good to fix.</p>
 630 <p>
 631 </p>
 632 <h2><a name="__prototype_character">_ prototype character</a></h2>
 633 <p>Study the possibility of adding a new prototype character, <code>_</code>, meaning
 634 ``this argument defaults to $_''.</p>
 635 <p>
 636 </p>
 637 <h2><a name="state_variables">state variables</a></h2>
 638 <p><code>my $foo if 0;</code> is deprecated, and should be replaced with
 639 <code>state $x = &quot;initial value\n&quot;;</code> the syntax from Perl 6.</p>
 640 <p>
 641 </p>
 642 <h2><a name="_inc_source_filter_to_filter__simple">@INC source filter to Filter::Simple</a></h2>
 643 <p>The second return value from a sub in @INC can be a source filter. This isn't
 644 documented. It should be changed to use Filter::Simple, tested and documented.</p>
 645 <p>
 646 </p>
 647 <h2><a name="regexp_optimiser_optional">regexp optimiser optional</a></h2>
 648 <p>The regexp optimiser is not optional. It should configurable to be, to allow
 649 its performance to be measured, and its bugs to be easily demonstrated.</p>
 650 <p>
 651 </p>
 652 <h2><a name="unitcheck">UNITCHECK</a></h2>
 653 <p>Introduce a new special block, UNITCHECK, which is run at the end of a
 654 compilation unit (module, file, <a href="file://C|\msysgit\mingw\html/pod/perlfunc.html#item_eval"><code>eval(STRING)</code></a> block). This will correspond to
 655 the Perl 6 CHECK. Perl 5's CHECK cannot be changed or removed because the
 656 O.pm/B.pm backend framework depends on it.</p>
 657 <p>
 658 </p>
 659 <h2><a name="optional_optimizer">optional optimizer</a></h2>
 660 <p>Make the peephole optimizer optional. Currently it performs two tasks as
 661 it walks the optree - genuine peephole optimisations, and necessary fixups of
 662 ops. It would be good to find an efficient way to switch out the
 663 optimisations whilst keeping the fixups.</p>
 664 <p>
 665 </p>
 666 <h2><a name="you_want__how__many">You WANT *how* many</a></h2>
 667 <p>Currently contexts are void, scalar and list. split has a special mechanism in
 668 place to pass in the number of return values wanted. It would be useful to
 669 have a general mechanism for this, backwards compatible and little speed hit.
 670 This would allow proposals such as short circuiting sort to be implemented
 671 as a module on CPAN.</p>
 672 <p>
 673 </p>
 674 <h2><a name="lexical_aliases">lexical aliases</a></h2>
 675 <p>Allow lexical aliases (maybe via the syntax <code>my \$alias = \$foo</code>.</p>
 676 <p>
 677 </p>
 678 <h2><a name="entersub_xs_vs_perl">entersub XS vs Perl</a></h2>
 679 <p>At the moment pp_entersub is huge, and has code to deal with entering both
 680 perl and XS subroutines. Subroutine implementations rarely change between
 681 perl and XS at run time, so investigate using 2 ops to enter subs (one for
 682 XS, one for perl) and swap between if a sub is redefined.</p>
 683 <p>
 684 </p>
 685 <h2><a name="self_ties">Self ties</a></h2>
 686 <p>self ties are currently illegal because they caused too many segfaults. Maybe
 687 the causes of these could be tracked down and self-ties on all types re-
 688 instated.</p>
 689 <p>
 690 </p>
 691 <h2><a name="optimize_away___">Optimize away @_</a></h2>
 692 <p>The old perltodo notes ``Look at the ''reification`` code in <code>av.c</code>''.</p>
 693 <p>
 694 </p>
 695 <h2><a name="what_hooks_would_assertions_need">What hooks would assertions need?</a></h2>
 696 <p>Assertions are in the core, and work. However, assertions needed to be added
 697 as a core patch, rather than an XS module in ext, or a CPAN module, because
 698 the core has no hooks in the necessary places. It would be useful to
 699 investigate what hooks would need to be added to make it possible to provide
 700 the full assertion support from a CPAN module, so that we aren't constraining
 701 the imagination of future CPAN authors.</p>
 702 <p>
 703 </p>
 704 <hr />
 705 <h1><a name="big_projects">Big projects</a></h1>
 706 <p>Tasks that will get your name mentioned in the description of the ``Highlights
 707 of 5.10''</p>
 708 <p>
 709 </p>
 710 <h2><a name="make_ithreads_more_robust">make ithreads more robust</a></h2>
 711 <p>Generally make ithreads more robust. See also <a href="#icow">iCOW</a></p>
 712 <p>This task is incremental - even a little bit of work on it will help, and
 713 will be greatly appreciated.</p>
 714 <p>
 715 </p>
 716 <h2><a name="icow">iCOW</a></h2>
 717 <p>Sarathy and Arthur have a proposal for an improved Copy On Write which
 718 specifically will be able to COW new ithreads. If this can be implemented
 719 it would be a good thing.</p>
 720 <p>
 721 </p>
 722 <h2><a name="________closures_in_regexps">(?{...}) closures in regexps</a></h2>
 723 <p>Fix (or rewrite) the implementation of the <code>/(?{...})/</code> closures.</p>
 724 <p>
 725 </p>
 726 <h2><a name="a_reentrant_regexp_engine">A re-entrant regexp engine</a></h2>
 727 <p>This will allow the use of a regex from inside (?{ }), (??{ }) and
 728 (?(?{ })|) constructs.</p>
 729 <table border="0" width="100%" cellspacing="0" cellpadding="3">
 730 <tr><td class="block" style="background-color: #cccccc" valign="middle">
 731 <big><strong><span class="block">&nbsp;perltodo - Perl TO-DO List</span></strong></big>
 732 </td></tr>
 733 </table>
 734
 735 </body>
 736
 737 </html>