Perl 6 language and compiler development | Logs at colabti.org/irclogger/irclogger_log/perl6-dev | For toolchain/installation stuff see #perl6-toolchain | For MoarVM see #moarvm
Set by Zoffix on 27 July 2018.
ugexe lizmat: fwiw that WHICH thing goes back to pre-6c, before we had a repo chain. So I’m not sure caching in this way makes sense 00:22
because having identical repos should not have identical next-repo 00:23
caching makes sense if we have no repo-chain, e.g. the best match from ALL repos is chosen when loading, not the best match from the FIRST repo containing a match 00:25
the one thing i've felt CUR::Locally is mean to do is normalize file paths (make them absolute + cleanup stuff so ../foo../ and ../ are treated as the same repo) 02:50
github.com/rakudo/rakudo/blob/7a3f...ry.pm6#L34
that $class.?absolutify is from pre 6.c as well, but it seems like the idea was to allow classes to provide the api for normalizing the path spec 02:51
note nothing uses or has *ever* had a method absolutify
see the notes from lizmat here: github.com/rakudo/rakudo/pull/468 02:53
04:01 elcaro joined 05:03 vrurg left 06:49 robertle joined 07:10 patrickb joined 08:15 [TuxCM] joined
[TuxCM] Rakudo version 2019.03.1-310-g3c5a53229 - MoarVM version 2019.05-3-gf3aeae67c
csv-ip5xs0.721 - 0.744
csv-ip5xs-206.004 - 6.011
csv-parser22.373 - 23.015
csv-test-xs-200.428 - 0.434
test6.989 - 7.445
test-t1.660 - 1.774
test-t --race0.802 - 0.805
test-t-2027.935 - 28.602
test-t-20 --race9.033 - 9.129
 09:13
lizmat Files=1262, Tests=107940, 410 wallclock secs (30.15 usr 7.17 sys + 2914.55 cusr 255.38 csys = 3207.25 CPU) 09:42
11:21 Guest55475 left
timotimo m: $_ = 128; with Str { say "str is defined" } else { say "str is not defined. dollar underscore is now"; say $_.perl; } 11:31
camelia str is not defined. dollar underscore is now
Str
timotimo ^- should this be the case? that the $_ in the else takes the value of $_ inside the with block? 11:32
11:37 [TuxCM] left
jnthn timotimo: Yes, that's an intended behavior 12:11
Or at least, I'm pretty sure TimToady++ made it that way on purpose
timotimo OK! 12:12
i guess i can see how it makes sense to have it, if the undefined value can have different types, or interesting values i guess? 12:13
jnthn Especially if it's a Failure 12:19
timotimo ah, indeed 12:26
12:39 lucasb joined 13:24 vrurg joined
nine Looks like it would actually have been a good idea for me to document how certain parts of Inline::Perl5 really work. Because....wow, this stuff is advanced :D 14:33
TimToady timotimo: yes, one of our explicit design goals was to allow for interesting values of both failure and success 15:01
timotimo i have very personal reasons to find that very appealing :P :P 15:02
TimToady succeeds in failing to understand :)
timotimo fails to succeed at the humor 15:04
i deprecated self just for this humor and it wasn't worth Any! 15:05
jnthn Clearly, a joke that's only funny once is self-deprecating humor...
timotimo i'd call that one idempotent, because no matter how often i think of it, i'll always be able to laugh 15:06
15:13 [TuxCM] joined 15:24 patrickb left 15:42 robertle left 16:03 [TuxCM] left 16:44 robertle joined 17:17 ufobat joined 17:32 patrickb joined
Geth_ rakudo/master: 4 commits pushed by skids++, (Patrick Böker)++, (Elizabeth Mattijsen)++ 18:22
18:27 skids joined
skids crosses fingers for limited ecosystem fallout :-) 18:27
patrickb skids, kawaii: Do you think it makes sense to make the release to be done this weekend to be based on the commit before this merge to just give it a little bit more time to be tested out? 18:29
i.e. create a release-2019.05 branch and base it on the commit before the colon_list merge. 18:30
lizmat hmmm... /me assumed a release branch was already made :-( 18:32
skids Makes sense to me. 18:33
lizmat afaik there is no R* release planned based on the 2019.05 release, is there ? 18:34
18:36 brrt joined
patrickb lizmat: That's not really a problem, the release branch can just be based on some older commit. I'd say the thing a release branch mostly prevents the situation where people stop committing stuff in fear of breaking things. 18:38
lizmat indeed :-) 18:39
skids I'm pretty amazed that PR survived nearly a year without conflicts. 18:42
lizmat skids: that just shows you dare to tread where few others do :-)
18:43 ufobat left
Geth_ rakudo: 1db7e4b1ff | (Elizabeth Mattijsen)++ | 2 files
Remove some superstitious parentheses
 18:43
patrickb skids: Could you have a look at github.com/rakudo/rakudo/pull/1170 ? I think it might be easy for you to say whether that other commit I referenced there actually fixes the bug or just hides/partially fixes it. 18:45
Geth_ roast: 45fb3ccdb9 | (Elizabeth Mattijsen)++ | integration/advent2013-day10.t
Unfudge now passing test, skids++
 18:46
vrurg Any news about the release? Is it still planned for May or postponed till June? 19:05
Geth_ rakudo: e9bbefd3c3 | (Elizabeth Mattijsen)++ | 2 files
Allow an IO::Path for CUR::FileSystem

  - fixes #2879
  - moved the $!lock attribute from CUR::Locally to CUR::Installation
   - as it is only needed there
  - added $!abspath attribute to CUR::Locally
   - as this appears to be really needed for all classes consuming CUR::Locally
   - and it is always calculated, so why not keep it then
  - move cache to role lexical, lose the state (potential race condition)
   - also add a lock so that all access/changes are done by a single thread
  - make $!WHICH a ValueObjAt, indicating a CUR::Locally is a value type		 19:07
nine vrurg: AFAIK this weekend 19:08
lizmat vrurg: kawaii said he most likely had time for it this weekend, so I hope to be able to report the release in the next P6W
vrurg Thanks! That's crazy, but I'm hit by #2166 again. Have fix for it, but it depends on #2816 which depends on release... 19:10
vrurg is in despair...
nine So...why don't we switch to release branches as MoarVM already uses? 19:11
patrickb vrurg: My plan to get p6 with spaces on windows to work depends on #2816 as well, so, you're not the only one waiting ;-) 19:12
vrurg patrickb: ok, it's easier to wait together. ;) 19:13
nine: that'd be the best, I think. 19:14
But it suggests less frequent release cycle. Which we tend to fall into anyway. 19:15
nine Why does it suggest that? 19:16
skids patrickb: I'll take a look. 19:17
lizmat vrurg: is #2816 still a "proposal for review" or is it ready as far as you are concernned ? 19:18
vrurg nine: Because to me it comes with different support model where releases are incorporating critical fixes over some period of time.
lizmat: it is ready for as much as I was able to test it on macOS, Linux, and Windows.
nine Not necessarily. That's of course possible when you have the branch. But you could do that just as well by branching after the fact. 19:19
vrurg nine: Yes, but monthly cicle means too many releases and loads of work to maintain them. Plus we have user expectations. I mean, it is not mandatory, but very much expected scenario. 19:21
Either way, branching for releases is what I'd be happy to see as I'm living on master branch for some time now anyway. 19:22
lizmat vrurg: PR #2816 doesn't work for me, output pasted
timotimo R#2816 19:23
vrurg lizmat: thanks! I knew something would be missing. I can't test all possible use cases.
timotimo oh, we don't have a synopsebot at the moment? 19:24
vrurg nine: BTW, I played a bit with make.nqp. It needed some fixing with regard to handling malformed Makefiles. But more important, there are no target variables like $@, $<. 19:29
lizmat: fixed, you can try now. 19:30
patrickb I think 3c5a5322982c2fe6bb5cc87934b9b419fc81d591 might be a good commit for the 2019.05 release. 19:32
nine vrurg: do we use those? 19:35
lizmat vrurg: will do, in about 15 mins
vrurg nine: That's why I mention it.
nine Oh, apparently we now do 19:36
vrurg nine: no rush. I would commmit it to my branch and get back to this when merged.
vrurg wouldn't do any more sharp movements. 19:37
nine Looks like $@ is used in rakudo. I find $< only in MoarVM which needs a bit more anyway
vrurg nine: I wasn't sure which one is needed. 19:40
BTW, I have doubts about make.nqp not having problems with spaces in batch names. You call cmd.exe with /c which might imply same rules about quoting. 19:41
nine Well what I meant when suggesting make.nqp was not as much that it already supports spaces in paths but that it should be vastly easier to fix make.nqp than to work around nmake 19:43
Ok $@ is just the name of the target? That should be easy to implement :) 19:44
Geth_ rakudo: b2e72d6d74 | (Elizabeth Mattijsen)++ | src/core/CompUnit/Repository/Locally.pm6
IO::Path has .absolute, not .abspath
 19:47
vrurg nine: [email@hidden.address] has it with my fixes. 19:48
Perhaps it would make sense to make this repo its home.
patrickb I start loosing faith in cmd being usable for any more complex calling scenarios. I'm seriously pondering replacing the Windows build perl6 runners with binaries as well, just to get rid of bat.
vrurg patrickb: that's about what I was thinking about too last week. :) 19:50
vrurg would be gone for ~1hr 19:51
Any additional testing of #2816 is very much welcome, BTW... 19:52
20:13 brrt left
Geth_ roast: 56dd56ad41 | (Elizabeth Mattijsen)++ | S02-literals/adverbs.t
Add test for R #2863
 20:15
skids Can the evalbots eval against a PR? 20:25
commit: b50aa51e0 role FFFF { method gah {...} }; enum f does FFFF <Fee Fie Foo Fum>; 20:34
committable6 skids, ¦b50aa51: «04===SORRY!04=== Error while compiling /tmp/WarfwQYRkQ␤Method 'gah' must be implemented by f because it is required by roles: FFFF.␤at /tmp/WarfwQYRkQ:1␤ «exit code = 1»»
skids commit: fd5d5bdfb role FFFF { method gah {...} }; enum f does FFFF <Fee Fie Foo Fum>;
committable6 skids, ¦fd5d5bd: «»
skids patrickb: ^^ one problem with that commit. There might be more, I'll have to dig in deeper than I did before to figure it out. 20:35
But I don't know if my PR did anything about the "handles" issue. 20:36
21:14 skids left 21:33 DrForr joined 22:41 patrickb left

Powered by Cro and the Raku® Programming Language.

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

Thank you!