|
Parrot 6.8.0 "Little Lorikeet" | parrot.org/ | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC Set by moderator on 17 September 2014. |
|||
|
01:35
FROGGS joined
01:46
FROGGS_ joined
02:21
sorear joined
02:33
bighugedog joined
02:46
kid51 joined
03:08
wagle joined
05:33
FROGGS joined
06:33
rurban joined
07:09
rurban1 joined
07:14
rurban joined
|
|||
| rurban | FROGGS: I can now confirm your perl6-p --target=pir -e '{ sub prefix:<\\o/>($) { say 42 }; \\o/1 }' >nullpmc.pm.pir problem with \\\\o | 07:22 | |
|
07:23
FROGGS joined
|
|||
| rurban | That's very likely a nqp-p bug | 07:26 | |
| nqp double-quotes the lex name because it is in a string. wrong: .lex "&prefix:<\\\\o/>", $P103 old/good: .lex "&prefix:<\\o/>" | 07:30 | ||
| FROGGS | okay... | 07:32 | |
| rurban | but I'm checking it in parrot first. looks fishy in parrrot also | 07:38 | |
| the parrot testcases only deal with single-quoted names, like .lex 'foo' not .lex "foo" | 07:39 | ||
| nope, it's defineitly a parrot bug with double-quoted names only. single-quoted names, like .lex '&prefix:<\\o/>', $P103 do work fine | 07:45 | ||
| FROGGS | okay... when you can point me to the right places in parrot I might be able to propose a patch... | 07:48 | |
| rurban | I filed a parrot issue #1095 | 07:56 | |
| FROGGS | nice | ||
| rurban | but I have no permissions to view or add to rt.perl.org/Ticket/Display.html?id=116643 | 07:57 | |
| FROGGS | Coke: can you arrange that? | ||
| rurban | probably not. p5p blocked me from RT | 07:58 | |
| FROGGS | O.o | ||
| so you are the guy that posts that russian spam :o) | |||
| rurban | nope, I dared to criticize perl5 | ||
| FROGGS | ahh | 07:59 | |
| rurban | when I start criticizing perl6 they will block me also. The new age | ||
| FROGGS | hehe | ||
| I keep an eye on you then *g* | |||
| rurban | yes, critics are only allowed to do blog posts, so that everyone will see it. | 08:01 | |
| FROGGS | " Reportedly this worked in parrot before, but it could be a nqp change also." that is not true | 08:02 | |
| I said that I can *make* it work by changing the generated .pir file | |||
| rurban | "I get the "Null PMC access" errors only when putting the declaration in an extra block." | 08:04 | |
| the extra block came from nqp | |||
| FROGGS | no | ||
| rurban | not? this causes it to be stored in a .lex now | ||
| FROGGS | "sub prefix:<\\o/>($) { }" works while "{ sub prefix:<\\o/>($) { } }" doesnt | ||
| rurban | yes, because if in a block it needs to be a .lex | 08:05 | |
| And my guess is that this block came from nqp, not parrot. parrot always had this quoting problem is my theory | |||
| FROGGS | so there is no rakudo or nqp change that uncovers this problem | ||
| rurban | and the workaround would be to use single-quotes around .lex names | 08:06 | |
| FROGGS | hmmm | ||
| and then only one backslash? | |||
| rurban | yes | ||
| FROGGS | then we can't use parrot's .escape method anymore :/ | 08:07 | |
| or we fiddle with the result of .escape, but that feels icky | |||
| rurban | in the meantime I will try to get at the bottom of the "\\o" issue, which is a binary character | ||
| FROGGS | and also, I had expected that the second test from issue #1095 would fail, not the third | 08:08 | |
| rurban | me also | ||
| so it's definitely weird | |||
| FROGGS | yeah | ||
| dalek | rrot/rurban/doublequoted-names-gh1095: 2947443 | rurban++ | t/compilers/imcc/syn/clash.t: [test] add testcase for double-quoted lex names GH #1095 in perl6 the correctly double-quoted .lex "foo\\\\o", $P3 name fails to work. see rt.perl.org/Public/Bug/Display.html?id=116643 this might be caused by the switch from globals to lexicals, as this statement is now enclosed in a block. in parrot the binary character \\0 causes problems in lexnames with roundtrips. |
08:14 | |
| FROGGS | eww, I've just seen this: | 08:15 | |
| .lex "&prefix:<\\o/>", $P103 | |||
| $P5004 = "&prefix:<\\\\o/>"($P5003) | |||
| this is from the pir code that works, mind | |||
| so, only .lex seems buggy | |||
| (I guess) | |||
| rurban | yes. before it was a global. with the switch to .lex it started failing was my guess | 08:17 | |
| FROGGS | =item C<void declare_lex_preg(STRING *name, INTVAL preg)> | 08:25 | |
| Declare a lexical variable that is an alias for a PMC register. The PIR | |||
| compiler calls this method in response to a ".lex STRING, PREG" directive. | |||
| where is that pir compiler? | |||
| rurban | hmm, nobody uses that… | 08:27 | |
| oh, nqp does: vm/parrot/pmc/nqplexinfo.pmc | |||
| we probably need to add some testcases for our lexinfo.pmc | 08:29 | ||
| I'm adding a test for this method in parrot now | 08:39 | ||
| hmm, I also don't see nqp nor perl6 using declare_lex_preg | 08:51 | ||
| FROGGS | I just stumbled upon that while searching for how .lex is implemented | 08:54 | |
| rurban | but it works ok | 09:04 | |
| I'm adding more torture tests... | 09:05 | ||
|
09:23
slavorg joined
09:55
dalek joined
|
|||
| dalek | rrot/rurban/lexqnames-gh1095: 234386d | rurban++ | / (4 files): [test] Added LexInfo.declare_lex_preg and more imcc/syn/clash.t tests For GH #1095 global names do work fine, only lexicals not. Note I do not know how to reliable get the correct target register index for declare_lex_preg in pure pir, and how to initialize it. Setting it crashes the ctx. |
10:10 | |
|
10:14
bighugedog joined
|
|||
| dalek | rrot/rurban/make-codas: f750baa | rurban++ | config/ (2 files): [build] Change darwin build flags to enable gdb debugging compile parrot_old static. use this for debugging with gdb. keep -undefined dynamic_lookup only for dynamically loaded bundles. honor cmdline flags for --ld and --link. and move deletion of generated config and makefiles from archclean to realclean/distclean. we want to delete all object and executables only there. |
10:19 | |
| rrot/rurban/make-codas: c959b8a | rurban++ | / (9 files): [cage] add makefile codas also to the Defines.mak |
|||
| rrot/smoke-me/darwin-gdb-gh1096: d2ad859 | rurban++ | config/ (2 files): [build] Change darwin build flags to enable gdb debugging compile parrot_old static. use this for debugging with gdb. keep -undefined dynamic_lookup only for dynamically loaded bundles. honor cmdline flags for --ld and --link. and move deletion of generated config and makefiles from archclean to realclean/distclean. we want to delete all object and executables only there. See [GH #1096] |
10:23 | ||
|
10:28
travis-ci joined
|
|||
| travis-ci | [travis-ci] parrot/parrot#1114 (smoke-me/darwin-gdb-gh1096 - f750baa : Reini Urban): The build has errored. | 10:28 | |
| [travis-ci] Change view : github.com/parrot/parrot/compare/s...gdb-gh1096 | |||
| [travis-ci] Build details : travis-ci.org/parrot/parrot/builds/37380598 | |||
|
10:28
travis-ci left
10:43
travis-ci joined
|
|||
| travis-ci | [travis-ci] parrot/parrot#1115 (smoke-me/darwin-gdb-gh1096 - d2ad859 : Reini Urban): The build passed. | 10:43 | |
| [travis-ci] Change view : github.com/parrot/parrot/commit/d2ad8596cf9f | |||
| [travis-ci] Build details : travis-ci.org/parrot/parrot/builds/37380644 | |||
|
10:43
travis-ci left
|
|||
| rurban | travis had a weird hickup | 10:46 | |
|
11:00
bighugedog joined
11:15
d4l3k_ joined
11:17
dalek joined
|
|||
| dalek | rrot: da59708 | rurban++ | / (9 files): [cage] add makefile codas also to the Defines.mak |
11:18 | |
|
11:18
autark_ joined
11:33
travis-ci joined
|
|||
| travis-ci | [travis-ci] parrot/parrot#1116 (master - da59708 : Reini Urban): The build passed. | 11:33 | |
| [travis-ci] Change view : github.com/parrot/parrot/compare/3...597081356f | |||
| [travis-ci] Build details : travis-ci.org/parrot/parrot/builds/37384375 | |||
|
11:33
travis-ci left
|
|||
| dalek | rrot: 92fd147 | rurban++ | config/gen/makefiles/root.in: [build] opsrenumber.pl is long gone remove the make target |
11:38 | |
|
11:54
travis-ci joined
|
|||
| travis-ci | [travis-ci] parrot/parrot#1117 (master - 92fd147 : Reini Urban): The build passed. | 11:54 | |
| [travis-ci] Change view : github.com/parrot/parrot/compare/d...fd147ee3d2 | |||
| [travis-ci] Build details : travis-ci.org/parrot/parrot/builds/37386236 | |||
|
11:54
travis-ci left
|
|||
| dalek | rrot/smoke-me/darwin-gdb-gh1096: 679ffc7 | rurban++ | t/compilers/imcc/syn/clash.t: [test] add testcase for double-quoted lex names GH #1095 in perl6 the correctly double-quoted .lex "foo\\\\o", $P3 name fails to work. see rt.perl.org/Public/Bug/Display.html?id=116643 this might be caused by the switch from globals to lexicals, as this statement is now enclosed in a block. in parrot the binary character \\0 causes problems in lexnames with roundtrips. |
12:09 | |
| rrot/smoke-me/darwin-gdb-gh1096: 1a0d496 | rurban++ | / (4 files): [test] Added LexInfo.declare_lex_preg and more imcc/syn/clash.t tests For GH #1095 global names do work fine, only lexicals not. Note I do not know how to reliable get the correct target register index for declare_lex_preg in pure pir, and how to initialize it. Setting it crashes the ctx. |
|||
| rrot/smoke-me/darwin-gdb-gh1096: 1d8d02a | rurban++ | config/ (2 files): [build] Change darwin build flags to enable gdb debugging compile parrot_old static. use this for debugging with gdb. keep -undefined dynamic_lookup only for dynamically loaded bundles. honor cmdline flags for --ld and --link. and move deletion of generated config and makefiles from archclean to realclean/distclean. we want to delete all object and executables only there. See [GH #1096] |
|||
|
12:24
travis-ci joined
|
|||
| travis-ci | [travis-ci] parrot/parrot#1118 (smoke-me/darwin-gdb-gh1096 - 1d8d02a : Reini Urban): The build passed. | 12:24 | |
| [travis-ci] Change view : github.com/parrot/parrot/compare/6...8d02ac191e | |||
| [travis-ci] Build details : travis-ci.org/parrot/parrot/builds/37388734 | |||
|
12:24
travis-ci left
|
|||
| Coke | rurban: I can only make people bugadmins on perl6 RT. I don't have any finer grained control. | 14:58 | |
|
15:53
Chirag joined
|
|||
| rurban | Coke: no prob | 16:12 | |
| I am now at the bottom of the imcc parser bugs with quoted names. Updated github.com/parrot/parrot/issues/1095 | 16:13 | ||
| dalek | rrot/rurban/lexqnames-gh1095: a41c105 | rurban++ | t/compilers/imcc/syn/clash.t: [test] add testcase for immc parser bug even with globals Also updated [GH #1095] |
16:15 | |
|
16:30
FROGGS joined
|
|||
| dalek | rrot/rurban/lexqnames-gh1095: 4219e53 | rurban++ | / (2 files): [test] just clarifying comments about the parser |
18:25 | |
| FROGGS | rurban++ | 18:30 | |
|
18:51
rurban joined
|
|||
| FROGGS | Linked: parrot-ops2c | 19:23 | |
| ./ops2c --dynamic src/dynoplibs/bit.ops --quiet | |||
| make: ./ops2c: Kommando nicht gefunden | |||
| make: *** [src/dynoplibs/bit_ops.c] Fehler 127 | |||
| :o( | |||
| hmmm | |||
| maybe I need to reconfigure? | |||
| probably | |||
| I am testing this right now: gist.github.com/FROGGS/8339a718c471294aea47 | 19:47 | ||
| parrot tests are fine, nqp built and rakudo builds right now... | |||
| I am surprised that it compiles at all, so I wonder if that will even fix our problem here | 19:48 | ||
| okay, rakudo also compiles but the bug stays :o) | 19:53 | ||
|
21:00
khisanth_ joined
22:24
kid51 joined
23:59
dalek joined
|
|||