| Geth | rakudo: 34889bebc3 | (Elizabeth Mattijsen)++ | src/core/Rakudo/Internals/HyperToIterator.pm6 Add HyperToIterator.skip-at-least For slight benefit in the .hyper.foo.skip case. |
00:03 | |
| lizmat | . | ||
| yoleaux | 16 Mar 2018 18:33Z <Zoffix> lizmat: is it intentional that a bunch of subs lost their purity in this commit? They do come out faster on HEAD, but I wonder if `is pure` would give even more benefit github.com/rakudo/rakudo/commit/cd...7b598e0bfb | ||
| 16 Mar 2018 18:40Z <Zoffix> lizmat: oh, I think it's: R#1574 . Tho purity won't affect those subs, would it? 'cause method calls won't be folded, only the subs that delegate to methods. Can those `is pure`s go back, do you know? | |||
| synopsebot | R#1574 [closed]: github.com/rakudo/rakudo/issues/1574 Should we allow "is pure" on proto's in the setting | ||
| lizmat | .tell Zoffix no, that was not intentional | ||
| yoleaux | lizmat: I'll pass your message to Zoffix. | ||
| lizmat | .tell Zoffix not sure what the logic is behind which ones do or don't have "is pure" atm | 00:04 | |
| yoleaux | lizmat: I'll pass your message to Zoffix. | ||
| AlexDaniel | lizmat: fwiw github.com/perl6/whateverable/issu...-373876594 | 00:07 | |
| Zoffix | . | 00:16 | |
| yoleaux | 00:03Z <lizmat> Zoffix: no, that was not intentional | ||
| 00:04Z <lizmat> Zoffix: not sure what the logic is behind which ones do or don't have "is pure" atm | |||
| Zoffix | I think it's OK to leave them as is for the release, lest we infect more routines with R#1566 bug. | 00:17 | |
| synopsebot | R#1566 [open]: github.com/rakudo/rakudo/issues/1566 [regression] multi built-ins are not assignable to Callable | ||
| lizmat | yeah | ||
| sounds like a plan | |||
| afk again& | 00:21 | ||
| pmurias | what is stopping github.com/rakudo/rakudo/blob/mast...u.pm6#L113 from showing up in Backtrace.list? | 09:06 | |
| nine | pmurias: it does show up in a simple test: perl6 --ll-exception -e 'class Foo { method BUILD() { die } }.new' | 09:14 | |
| lizmat | Files=1236, Tests=76271, 317 wallclock secs (14.73 usr 5.39 sys + 2169.54 cusr 218.41 csys = 2408.07 CPU) | 12:15 | |
| Geth | rakudo: 3e8ee99116 | (Elizabeth Mattijsen)++ | src/core/IterationBuffer.pm6 Give IterationBuffer an .append for IterationBuffers This should come in handy when re-assembling HyperWorkBatches in the .push-all case, and probably some other cases. |
12:24 | |
| roast: dogbert17++ created pull request #403: Test rt 12851x |
12:35 | ||
| dogbert17 | lizmat: o/ are you trying to speed up [Tux] ccsv tests? | 12:37 | |
| lizmat | not specifically, no | ||
| dogbert17 | ok | ||
| lizmat | trying to speed up any .hyper / .race action | 12:38 | |
| dogbert17 | is it going well? | ||
| lizmat | depends on how you look at it | ||
| dogbert17 | :) | ||
| lizmat | the bits before and after paralellization become very important | 12:39 | |
| because there's only 1 thread doing it | |||
| so if you want to see better wallclock, that's a point of focus | |||
| dogbert17 | I see. Do you see any obvious opportunities in the code? | 12:40 | |
| Geth | roast: 8887624ceb | (Jan-Olof Hendig)++ | S32-str/utf8-c8.t Add tests for RT #128511 |
12:41 | |
| roast: 78b89c9979 | (Jan-Olof Hendig)++ | S32-str/utf8-c8.t Add test for RT #128512 |
|||
| roast: d0d46093e5 | (Elizabeth Mattijsen)++ (committed using GitHub Web editor) | S32-str/utf8-c8.t Merge pull request #403 from dogbert17/test-rt-12851x Test rt 12851x |
|||
| synopsebot | RT#128511 [open]: rt.perl.org/Ticket/Display.html?id=128511 [BUG] utf8-c8 generates spurious NUL | ||
| synopsebot | RT#128512 [open]: rt.perl.org/Ticket/Display.html?id=128512 [BUG] utf8-c8 mangles by NFC | ||
| dogbert17 | lizmat: thx | ||
| lizmat | yw :-) | ||
| dogbert17 will now close two RT's yay :) | 12:42 | ||
| lizmat | ok, seems that doing a fast push-all on HyperToIterator is going to be more involved than I thought | 12:56 | |
| so I'll keep that for after the release | |||
| afk until later this evening& | 12:57 | ||
| timotimo | hum. passing in a prepared "spare" array to the mergesort functions means we'd always allocate a second array even ifthe array to sort is small enough to not have to mergesort | 13:41 | |
| but maybe sinceit's internal we can get away with calling nqp::create on the type and using that | |||
| though i guess cloning could be fnie | 13:42 | ||
| AlexDaniel | releasable6: status | 13:48 | |
| releasable6 | AlexDaniel, Next release in ≈5 hours. Blockers: github.com/rakudo/rakudo/issues?q=...%9A%A0%22. Changelog for this release was not started yet | ||
| AlexDaniel, Details: gist.github.com/d3e4fb8ca563ce700e...2b432617fb | |||
| Geth | rakudo: 511bec0a2c | (Timo Paulssen)++ | src/core/Rakudo/Sorting.pm6 make sure MERGESORT-* don't leak low-level types depending on how many items were in the list, you could end up getting a BOOTIntArray instead of an intarray, for example with 65 elements. This was caused by `$A` and `$B` switching places after every "round" of mergesort application, and if you got the right number of rounds, you'd get `$B` instead, which was just a nqp::list_i (or _n or _s). |
14:16 | |
| timotimo | breakfast time, ideally i'll write a test for roast for this bug, too | 14:27 | |
| didn't really find what file has both sort and native arrays | |||
| AlexDaniel | Geth: ver github.com/rakudo/rakudo/commit/97c6f9c8 | 16:05 | |
| Geth | AlexDaniel, version bump brought in these changes: github.com/perl6/nqp/compare/2018....8-g9716266 | ||
| AlexDaniel | Geth: ver github.com/perl6/nqp/commit/971626...5e4b283e3b | ||
| Geth | AlexDaniel, version bump brought in these changes: github.com/MoarVM/MoarVM/compare/2...1-geee5be4 | ||
| AlexDaniel | jnthn: fwiw if you want github.com/rakudo/rakudo/commit/21997b625695 changelogged, please do it: github.com/rakudo/rakudo/wiki/ChangeLog-Draft | 16:39 | |
| jnthn: I can't come up with a short description :) | |||
| jnthn | AlexDaniel: Fix code generation bug affecting private methods calls in roles where the target private method used a role parameter | 16:52 | |
| I'm not sure I can get it shorter than that. It was an obscure bug ;) | |||
| AlexDaniel | OK | ||
| Geth | roast: 61211a4b9c | usev6++ | 2 files [JVM] Unfudge some passing tests |
17:14 | |
| AlexDaniel | I get these warnings on rakudo-jvm: gist.githubusercontent.com/AlexDan...tfile1.txt | 17:55 | |
| like, every time I run anything (e.g. perl6-j -e 'say 42') | 17:57 | ||
| bartolin: ↑ any comments? | 17:58 | ||
| bartolin | that is with Java 9, isn't it? | ||
| I think there is a bug report somewhere, let me look | 17:59 | ||
| bartolin is testing with Java 8 only | |||
| ah, yes, R #1431 | 18:00 | ||
| AlexDaniel | yes, 9 | ||
| R#1431 | |||
| synopsebot | R#1431 [open]: github.com/rakudo/rakudo/issues/1431 [JVM] illegal reflective access | ||
| AlexDaniel | ok, I see | ||
| bartolin | I think the warning might be caused by using sun.misc.Unsafe here: github.com/rakudo/rakudo/blob/511b...ec.java#L4 | 18:07 | |
| bartolin makes a note to install a VM with Java 9 | |||
| Geth | rakudo: b75d9b1e14 | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/build/NQP_REVISION [NQP Bump] Brings 7 commits NQP bump brought: github.com/perl6/nqp/compare/2018....g199dd708b 199dd708b [js] Implement getcodelocation db6801437 [js] Throw an exception when accessing a missing variable with nqp::atkey 00b1f84f3 Remove partially commented out timing debug code 23835345a [js] Add $$ prefix to avoid interation over an internal field of a ctx 0191550a5 Test reaching the topmost caller with nqp::ctxcaller 15dfe66ef [js] Make nqp::ctxcaller return nqp::null when we have no caller 002097112 [js] Wrap native file and line values before putting in a hash |
18:58 | |
| ¦ rakudo: version bump brought these changes: github.com/perl6/nqp/compare/2018....g199dd708b | |||
| AlexDaniel | so! From my side everything looks great | ||
| no blockers, CI status is good | |||
| did a fresh toast of modules, everything is good | 18:59 | ||
| releasable gets to the last step without any issues | |||
| changelog is also ready, please feel free to tweak it: github.com/rakudo/rakudo/wiki/ChangeLog-Draft | 19:03 |