Perl 6 - the future is here, just unevenly distributed

IRC log for #perl11, 2016-05-03

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
02:21 ingy joined #perl11
02:28 ingy joined #perl11
04:19 ingy joined #perl11
07:03 rurban joined #perl11
07:21 rurban Beware: p5p removed CxFOREACHDEF() without notice with 5.24. Added with 5.9.3 given/when
07:31 willthechill rurban: when is that used?
07:31 willthechill please give a one-line code example
07:51 sten rurban hi, there are problem in rc1 prevent to auto install with perlbrew, ext/Config/Makefile.PL is readonly
07:52 sten Unsuccessful Makefile.PL(ext/Config): error=Cannot write Makefile.PL:Permission denied at /Users/user/perl5/perlbrew/build/cperl-5.22.2-RC1/cpan/ExtUtils-MakeMaker/lib/ExtUtils/Command.pm line 182.
07:52 sten Generating a Unix-style Makefile
07:52 sten Writing Makefile for Config
07:52 sten make: *** [lib/auto/Config/Config.bundle] Error 2
07:52 sten ##### Brew Failed #####
07:57 sten and schmorp don't do fixes
07:58 sten Coro fail with rc1
08:10 ribasushi sten: Schmorp can't "do fixes"
08:10 ribasushi sten: read here for details: http://blogs.perl.org/users/aristotle/2016/05/coro-vs-5022.html
08:15 lizmat I think aristotle set a new standard for bringing up painful issues in p5p
08:15 rurban ribasushi: nice summary :) ("despite the expertise of Reini Urban") I haven't even tried to workaround the warnhook
08:15 lizmat aristotle++
08:15 rurban I didn't consider it important enough
08:17 willthechill and the world turns... despite Reini, haha!  ;-)
08:17 rurban sten: this is interesting. bulk88 should fix that. he touches Makefile.PL, and apparently this is forbidden with the src tarballs. I'll try to find a workaround. Do you want to file a ticket?
08:18 rurban aristotle did a nice summary of the warnhook and EXTCONST bugs. EXTCONST is a clear p5p bug they refused to fix for several releases now
08:18 rurban warnhook is a dirty workaround for a broken architecture
08:20 rurban willthechill: to my rescue, I was the only who actually did something with Coro. The others just do summaries, but don't fix the bugs
08:23 willthechill well we have a long way to go before history can make a clear judgment... like a few decades
08:23 willthechill but I am glad you are actually addressing the Coro issue!
08:23 willthechill :-)
08:24 sten thanks for link
08:24 rurban well, just a simple one-day patch to make it work. without overrding warnhook, which schmorp correctly analyzed as a big problem
08:25 sten I can fill ticket if you want)
08:30 willthechill rurban++
08:30 willthechill sten++
08:30 rurban sten, done and fixed https://github.com/perl11/cperl/issues/142
08:31 rurban Coro is very important, I even have a ticket for it, but so far no time to fix it properly, at least in cperl. I doubt that p5p will be able to fix that in the upcoming years
08:32 rurban So far they are only blaming schmorp and not themselves
08:33 sten yeah Coro is very important, they should revert broken patch instantly after test fails.
08:34 sten I'm totally don't understand the way choosen by p5p
08:38 rurban sten: RC2 is coming today
08:41 rurban while the warnhook situation is a horrible p5p bug, which should be fixed properly, and not ad hoc.
08:43 rurban sten: Note that RC2 doesn't need to be done for the binary releases, they are not affected. Only the src tarballs
08:50 rurban I commented on http://blogs.perl.org/users/aristotle/2016/05/coro-vs-5022.html but it needs approval
09:28 travis-ci perl11/cperl#1027 (smoke/relprep - acb0440 : Reini Urban): The build passed. https://travis-ci.org/perl11/cperl/builds/127453766
10:55 abra joined #perl11
11:03 rurban https://github.com/perl11/cperl/releases/tag/cperl-5.22.2-RC2
11:03 willthechill rurban++
11:55 travis-ci perl11/cperl#1028 (master - acb0440 : Reini Urban): The build passed. https://travis-ci.org/perl11/cperl/builds/127483066
12:02 bulk88 rurban Coro stopped working on p5p perl, because it decided to patch a core vtable that should be stored in const memory for sharabaility between procs, I consted the vtable, coro stopped working, coro needs to ask for an API in core of what it wants to do, not change undocumented data structures
12:03 bulk88 I've considered writing a work around for Coro that uses OS specific facilities to change perms on the mem page with the vtable from RO to RW, you willl loose 4 KB of memory, but atleast Coro (which I dont care much for) will work
12:09 bulk88 the reason new API is needed IIRC is that perl core upgrades r->R or R->r magic type codes, by the case of the letter that reprepsents the magic type, if %SIG's vtable is replaced with a 3rd party vtable ptr, perl core wont do the magic vtable upgrade, hence why patching the ecore vtable is done
12:12 bulk88 3rd party vtable ptrs are always "U" or "~" and cant get hte upgrade doen to them
12:15 bulk88 http://perl5.git.perl.org/perl.git/blob/HEAD:/mg.c#l470 the upper case magic type gets converted to lower case magic type
12:31 rurban yes, that's what I fixed so that it can work with an unpatched perl5.22. with cperl it's easier
12:34 rurban bulk88: btw. I had to patch XSConfig Makefile.PL a tiny bit. See https://github.com/perl11/cperl/commit/b8a48b5b1e5ca21f7d1443b0ea70144adecf89e3
12:35 bulk88 ill have to take the cperl xsconfig files and sync it soon to CPAn
12:35 bulk88 i know there were other patches doen over hte last 2 mo
12:36 rurban on win32 some new keys were also added, the XSConfig.t tests fail there.
12:36 rurban the @cannedkeys, haven't fixed that. I left it for you
12:36 rurban 5.24 will also have more keys
13:09 sten bulk88 maybe need to do some work with critical cpan modules before publish perl that break cpan?
13:10 sten Coro is widely used
13:12 rurban I always test my releases with most of cpan before. ~2000 modules typically
13:26 rurban1 joined #perl11
13:31 rurban2 joined #perl11
13:47 rurban joined #perl11
13:48 rurban1 joined #perl11
13:49 rurban3 joined #perl11
13:52 rurban3 joined #perl11
15:57 rurban joined #perl11
16:11 rurban1 joined #perl11
16:12 rurban2 joined #perl11
17:06 lizmat joined #perl11
17:08 rurban joined #perl11
20:11 willthechill joined #perl11
20:26 rurban joined #perl11
21:40 travis-ci perl11/cperl#1029 (smoke/gh7-signatures - bc3f3b4 : Reini Urban): The build is still failing. https://travis-ci.org/perl11/cperl/builds/127626837

| Channels | #perl11 index | Today | | Search | Google Search | Plain-Text | summary