Perl 6 language and compiler development 2.0 | Logs at irclog.perlgeek.de/perl6-dev/today | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by moderator on 10 May 2018.
01:17 Merfont joined 01:21 Kaiepi joined 01:31 unicodable6 joined 01:58 ilbot3 joined
moderator Perl 6 language and compiler development 2.0 | Logs at irclog.perlgeek.de/perl6-dev/today | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
02:36 ufobat_ joined 02:54 |Tux| joined
Geth nqp: ugexe++ created pull request #441:
[js] Add getppid op		 03:04
03:07 committable6 joined 03:59 statisfiable6 joined 04:33 Kaiepi joined 04:39 lizmat joined 05:09 lizmat joined 05:39 lizmat joined 06:45 robertle joined 06:52 Ven`` joined 07:14 Ven`` joined 07:16 Ven` joined 07:32 Ven`` joined 07:33 Ven` joined
lizmat Files=1239, Tests=76339, 318 wallclock secs (15.52 usr 5.32 sys + 2174.99 cusr 223.03 csys = 2418.86 CPU) 07:40
07:53 Ven`` joined 07:59 Ven`` joined
|Tux| Rakudo version 2018.04.1-70-gfebcb9177 - MoarVM version 2018.04.1-98-g1aa02fe45
csv-ip5xs0.872 - 0.873
csv-ip5xs-208.570 - 8.697
csv-parser36.138 - 36.567
csv-test-xs-200.421 - 0.479
test8.878 - 8.896
test-t2.341 - 2.398
test-t --race0.966 - 0.990
test-t-2043.068 - 43.357
test-t-20 --race14.750 - 15.129
 08:06
08:40 jsimonet joined 09:14 dct joined 09:38 dct joined 11:09 Ven`` joined 12:05 Ven`` joined 12:17 brrt joined 12:19 Ven`` joined 12:44 Ven`` joined 13:10 AlexDaniel joined
AlexDaniel releasable6: status 13:24
yoleaux 10 May 2018 05:12Z <jmerelo> AlexDaniel: thanks, although I don't have all the information on how that binary is built. I'll try to anyway.
releasable6 AlexDaniel, Next release in ≈8 days and ≈5 hours. 2 blockers. 0 out of 70 commits logged
AlexDaniel, Details: gist.github.com/188e5cf083b386f3ea...d90426e659
Geth roast: 356b63ce68 | usev6++ | S02-types/array.t
[JVM] Fudge newly added test; R#1571
 13:36
synopsebot R#1571 [open]: github.com/rakudo/rakudo/issues/1571 Flaws in implied sinkage / `&unwanted` helper
13:46 skids joined 14:15 lucasb joined
Geth roast: c2bdb06b66 | usev6++ | S32-str/encode.t
[JVM] Adjust fudging after recent removal of tests
 14:53
roast: 11ff74efe0 | usev6++ | S09-typed-arrays/native.t
[JVM] Fudge some newly added tests
 15:01
16:50 lizmat joined 17:00 ExtraCrispy joined, lizmat joined 17:56 greppable6 joined 18:41 robertle joined
Geth roast: 14eaf91799 | usev6++ | S02-types/capture.t
[JVM] Fudge test for Capture.WHICH

  (And replace tabs with spaces.)		 19:12
bartolin_ Would you mind if I split some long arrays used for testing various combinations of sets in order to make the tests run on the JVM backend? e.g. github.com/perl6/roast/blob/14eaf9...bset.t#L30 19:51
we are hitting the limit for parameters for method calls in java (or so I understand the problem): github.com/rakudo/rakudo/issues/1817 19:52
jnthn Hm, I wonder if there's a way to code-gen things differently to avoid that... 19:53
(Doing so might be good for all backends)
bartolin_ oh, well, that would be best of course. if forgot to mention that alternative, because i have no idea how to do it ... 19:54
jnthn :) 19:56
No objections from me for such a test change, btw. We can always revert it if it's no longer needed.
Is there a Rakudo ticket about the error?
bartolin_ yes, i just created R#1817 19:57
synopsebot R#1817 [open]: github.com/rakudo/rakudo/issues/1817 [JVM] [JVM] Failing tests due to 'IllegalArgumentException: bad parameter count ...'
bartolin_ thanks for your input -- I'll take a look at splitting the tests. maybe it makes sense to add the golfed code as a separate test, too. 19:59
nine Shouldn't the compiler be even allowed to ignore all encapsulation and just generate nqp::push(nqp::getattr(@ok, List, '$!reified'), ...) for those values? Or is there some way that @ok can end up something different than an Array there? 20:00
Ah, of course, if someone were to overwrite infix:<,> in that scope... 20:01
But that could be detected at compile time 20:02
jnthn nine: Yeah, that was my thinking; make sure it's not overriden 20:05
And then nqp::list (which just code-gens a sequence of nqp::push) 20:07
nine Oh, how convenient. So half the job is already done 20:09
Looks like the static optimizer has all the pieces one would need for this 20:10
jnthn Yup :)
nine Oh, replacing the call to infix:<,> does not suffice. I also need to replace the call to STORE 20:33
Got it somewhat working now. Will investigate the test failures tomorrow. There are probably some interesting cases where we shouldn't apply this optimization. 21:39
Like my @y = 41, ($answer ?? 42 !! slip()), 43;
jnthn aye :) 21:40
nine++
nine bartolin_: re "So it looks like we are able to generate code that does not hit the limit of 254/255 parameters for method calls in java in some cases." Those some cases are when we can compile time evaluate all list items. Then we create the List at compile time also. 21:41
timotimo praise the test suite 21:46
nine jnthn: now I wonder if there is a way to deal with Slips at all. After all if we statically resolve the values we already create the Array at compile time 21:48
And if we cant statically resolve, we cant be sure threre are no slips 21:49
jnthn nine: Could just bind the list we create into $!todo, which'll explode the Slips lazily
nine aah, that sound better :) 21:50
Will try that tomorrow then
AlexDaniel releasable6: status 23:24
releasable6 AlexDaniel, Next release in ≈7 days and ≈19 hours. 2 blockers. 0 out of 70 commits logged
AlexDaniel, Details: gist.github.com/40c60be0e939cb27bf...69c537e5fb
23:30 skids joined

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!