| dalek | ast: 72fe409 | TimToady++ | S32-num/power.t: some tests for *ⁿ |
00:44 | |
| timotimo | oh, i just thought about that today! | 00:46 | |
| b2gills | m: say (^5).hyper(:1batch).map: *+1 | 01:36 | |
| camelia | rakudo-moar 6c9837: OUTPUT«HyperSeq.new» | ||
| b2gills | meant to do that in a private msg | 01:37 | |
| MasterDuke | lizmat timotimo: i just tried all the ones i initially found, only those in Exception.pm could be converted ( github.com/rakudo/rakudo/pull/961 ) | 05:04 | |
| [Tux] | This is Rakudo version 2016.12-3-g6c9837297 built on MoarVM version 2016.12 | 09:42 | |
| csv-ip5xs 3.282 | |||
| test 13.959 | |||
| test-t 6.122 | |||
| csv-parser 13.968 | |||
| dalek | kudo/nom: f812bce | MasterDuke17++ | src/core/Exception.pm: Convert some bindattrs to direct binding See irclog.perlgeek.de/perl6-dev/2016-...i_13759010 for reference. |
10:02 | |
| kudo/nom: ee2e8a5 | lizmat++ | src/core/Exception.pm: Merge pull request #961 from MasterDuke17/Exception.pm_bindattrs_to_direct_binding Convert some bindattrs to direct binding |
|||
| gfldex | in the last week Pod::To::BigPage went from 44s to 42s run time | 10:06 | |
| speed++ | 10:07 | ||
| samcv | nice | 10:12 | |
| gfldex, do you know why that was? | 10:13 | ||
| gfldex | many small improvements | ||
| dalek | kudo/nom: 9672bc0 | (Steve Mynott)++ | tools/create-release-announcement.pl: Using Perl 6 book no longer shipped with R* |
10:16 | |
| kudo/nom: 00c1fed | lizmat++ | tools/create-release-announcement.pl: Merge pull request #962 from LondonPM/nom Using Perl 6 book no longer shipped with R* |
|||
| RabidGravy | I think the headline thing is that the test-t in [Tux] benchmark above is *half* of what it was this time last year :) | ||
| at this rate Term::Cap might be usable before long (at the last testing it took 20s to parse a full termcap file) | 10:18 | ||
| samcv | how big are the termcap files? | 10:21 | |
| RabidGravy | ~ 200K the BSD one I was testing with | 10:23 | |
| there's a gnu one that is larger | 10:26 | ||
| dalek | kudo/nom: 5ec6b32 | lizmat++ | src/core/Block.pm: Replace bindattr with direct attribute usage |
10:30 | |
| MasterDuke | lizmat: huh, i thought i tried changing that Block bindattr | 13:05 | |
| it seems like none that used attributes declared in BOOTSTRAP.nqp could be converted, which is most of the ones i found | 13:09 | ||
| dalek | kudo/nom: fb38190 | moritz++ | src/core/Proc.pm: #128594: die in a controlled way rather than with a core dump |
14:17 | |
| synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=128594 | ||
| Zoffix | Hi. | 15:04 | |
| timotimo | hi! | ||
| Zoffix | Is it normal for a BOOTCode to throw "Cannot invoke this object (REPR: Null; VMNull)" when attempting to call it? | ||
| timotimo | it's not a BOOTCode, it's VMNull :) | ||
| did you accidentally mis-spell something? because nqp will just totally accept anything and give you VMNull (rather, NQPMu) for it | 15:05 | ||
| Zoffix | It's produced by World.compile_in_context | ||
| timotimo | oh, that isn't fun | ||
| Zoffix | If it's compiling `where` $_ = 1, it works: gist.github.com/zoffixznet/73c4161...fa7fd1ced5 but if it's compiling `where` $_ == 1, it throws that exception: gist.github.com/zoffixznet/d8fac04...ef9a603721 | 15:06 | |
| And now I'm wondering if it's normal? Like it's not meant to be called during that time or something? | |||
| timotimo | huh, it's calling ACCEPTS on the result of infix:<==> ? | 15:07 | |
| why aren't you using where * == 1 :) | |||
| Zoffix | Because that's not where the bug happens. | 15:08 | |
| timotimo | OK | ||
| Zoffix | m: subset InfTan of Numeric where $_ == 2; multi tan2 (InfTan) {"hah"}; say tan2(2e0) | 15:09 | |
| camelia | rakudo-moar fb3819: OUTPUT«hah» | ||
| Zoffix | m: subset InfTan of Numeric where $_ == 2; multi tan (InfTan) {"hah"}; say tan(2e0) | ||
| camelia | rakudo-moar fb3819: OUTPUT«-2.18503986326152» | ||
| Zoffix | m: subset InfTan of Numeric where $_ == 2; multi tan (InfTan) {"hah"}; say tan(2) | ||
| camelia | rakudo-moar fb3819: OUTPUT«Cannot invoke this object (REPR: Null; VMNull) in any accepts_type at gen/moar/Metamodel.nqp line 3444 in sub tan at <tmp> line 1 in block <unit> at <tmp> line 1» | ||
| Zoffix | And that goes away if you turn off the optimizer. | ||
| timotimo | aha! | ||
| Zoffix | And the optimizer tries to call the block made by World.compile_in_context | 15:10 | |
| timotimo | can you get the --ll-exception of that? | ||
| i don't know under what circumstance compile_in_context can return null, but the optimizer should probably check for that and pretend it never tried to do anything | |||
| Zoffix | it returns a BOOTCode | 15:11 | |
| If I say $result.HOW.name($result) it says BOOTCode | |||
| timotimo | hm, so the code it invokes does the attempt to call a null | ||
| Zoffix | Here's --ll-exception gist.github.com/zoffixznet/61122e7...bc3a899681 | ||
| <timotimo> huh, it's calling ACCEPTS on the result of infix:<==> ? | 15:12 | ||
| It's supposed to use it as a block, no? | |||
| Like a block that calls == on something | |||
| timotimo | ah, that's the post-optimizer thing | ||
| anyway | |||
| looks like binding the parameter to call the thing makes an explode inside ACCEPTS | 15:13 | ||
| i.e. the where is crashing because we're apparently calling something in the wrong way | |||
| i kind of think bind_one_param ought to catch that exception | 15:14 | ||
| MasterDuke | timotimo: i've been burned before by nqp accepting misspellings and giving VMNulls/NQPMus. is there any way to make nqp a little less forgiving? | 15:17 | |
| timotimo | yeah, should be possible | ||
| find the --target=parse of something simple like MisterMcDoesntExist | |||
| that ought to lead you to the point where some fallback is saying "we didn't find it, so give NQPMu" | |||
| Zoffix | Here's a piece that gives BOOTCode for .HOW.name, and then throws the VMNull: gist.github.com/zoffixznet/f417abc...69c4ffc4df | 15:18 | |
| (it's in the try {} block) | |||
| raiph | Q on SO about forking and piping unanwsered for a day (beyond my paygrade): stackoverflow.com/questions/4120237...n-blocking | 17:03 | |
| timotimo | raiph: i'll run the script through strace and see if i notice anything in particular | 17:08 | |
| need to build my whole perl6 first, though | 17:09 | ||
| raiph | timotimo: thx | 17:11 | |
| gfldex | #130371 can be closed | 17:22 | |
| synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130371 | ||
| timotimo | raiph: answered | 17:23 | |
| raiph | timotimo: <3 | 17:32 | |
| timotimo | that one was actually enjoyable to hunt down | ||
| raiph | :) | 17:34 | |
| lizmat | MasterDuke: re Block attribute: I was triggered by the fact that the further down it *does* use the short-hand :-) | 18:11 | |
| dalek | p: e424132 | (Pawel Murias)++ | src/vm/jvm/ (2 files): [jvm] Improve nqp::tc and nqp::tclc a bit. Still lack support for most codepoints that don't have a 1 to 1 mapping. |
18:34 | |
| ast: 0465872 | (Zoffix Znet)++ | S03-operators/overflow.t: Fix incorrect test The test was added 11 days ago[^1] and is not part of 6.c-errata. Negative exponents underflow, not overflow. [1] github.com/perl6/roast/commit/714a...f999a4d2d1 |
18:35 | ||
| kudo/nom: 7a642f8 | (Zoffix Znet)++ | src/core/Int.pm: Throw on underflow in infix:<**>(Int, Int) Fixes RT#130369: rt.perl.org/Ticket/Display.html?id=130369 The behaviour in the ticket is observed due to the power being small enough that overflow doesn't occur when we raise a number to absolute power, yet large enough that when we divide 1 by it, it underflows. - Add extra check for overflow condition to throw underflow exception if the power is negative - Check for zero result and throw underflow if the base is not zero |
18:38 | ||
| synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130369 | ||
| ast: 522a9c7 | (Zoffix Znet)++ | S03-operators/overflow.t: Test throwage with infix:<**>(Int, Int) with largish negative powers RT#130369: rt.perl.org/Ticket/Display.html?id=130369 Rakudo fix: github.com/rakudo/rakudo/commit/7a642f865a |
18:40 | ||
| synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=130369 | ||
| p: 9abf968 | (Pawel Murias)++ | src/vm/js/ (2 files): [js] Partial implementations of nqp::tc and nqp::tclc (doesn't support polytomic greek and ligatures yet). |
18:42 | ||
| p: 5538cbe | (Pawel Murias)++ | t/nqp/059-nqpop.t: Test nqp::tc and nqp::tclc a bit. |
|||
| ast: 860dc71 | (Zoffix Znet)++ | S02-types/nan.t: Test subclass of NaN !=== a NaN RT#129002: rt.perl.org/Ticket/Display.html?id=129002 |
20:04 | ||
| synopsebot6 | Link: rt.perl.org/rt3//Public/Bug/Displa...?id=129002 | ||
| samcv | pmurias++ | 20:28 | |
| dalek | p: 6b5a281 | (Pawel Murias)++ | src/vm/js/nqp-runtime/ (3 files): [js] Repossess objects in the places where Moar and JVM do that. |
20:44 | |
| b2gills | Where would be a good place to test ｢.lazy-if｣? | 20:47 | |
| lizmat | good night, #perl6-dev! | 21:11 | |
| timotimo | gnite lizmat |