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.
00:30 lucasb left
Geth rakudo: vrurg++ created pull request #3154:
Fix for incorrect default parent lookup 		02:45
AlexDaniel vrurg: hm, I wonder, do you have permissions to edit my comments? 02:58
vrurg AlexDaniel: yes, I do. Just didn't want to. 02:59
AlexDaniel ok, whatever works for you. But please feel free to edit any of my comments if you ever feel like
vrurg AlexDaniel: interestingly enough, I did fix the problem with Grammar but t/spec/integration/weird-errors.rakudo.moar segfaults on test 30 again. 03:00
AlexDaniel: ok, thanks! But generally, I dislike doing so. 03:01
Done for today. cu!
AlexDaniel yeah, a lot of people are against comment editing, I guess that's ok too. But I have a wiki mindset for anything that I publish, so I'm totally ok with people touching ~my~ our stuff :) 03:04
05:57 robertle left
nine Class Cro::MediaType::Grammar cannot inherit from itself 07:01
I think we need some reverting
07:20 robertle joined 07:50 Kaiepi left 07:53 Kaiepi joined 08:32 vrurg left, vrurg joined 09:42 pmurias joined
|Tux| Rakudo version 2019.07.1-237-g0455f42e3 - MoarVM version 2019.07.1-163-gb49b5202f
csv-ip5xs0.659 - 0.692
csv-ip5xs-204.814 - 4.982
csv-parser21.944 - 22.010
csv-test-xs-200.435 - 0.439
test6.927 - 7.132
test-t1.699 - 1.720
test-t --race0.794 - 0.804
test-t-2029.276 - 29.452
test-t-20 --race9.115 - 9.417
 10:14
eater how do I compile rakudo with a different STATIC_PERL6_HOME? it seems it's either set to ../share/perl6 or $PREFIX/share/perl6 (here: github.com/rakudo/rakudo/blob/3f69...o.pm#L349) 10:20
10:24 masak is now known as Guest74090 10:25 ilogger2 joined 10:26 eater joined, nebuchadnezzar joined, nebuchadnezzar is now known as Guest90858
eater but e.g. if you set libdir to /usr/lib, the runtime files will reside in /usr/lib/perl6 instead of {PREFIX}/share/perl6 10:28
10:29 commavir joined, bartolin joined, japhb joined 10:33 Voldenet joined, Voldenet left, Voldenet joined
gdonald I also ran into the 'Grammar cannot inherit from itself' a few days ago. 10:35
10:39 |Tux| joined 11:08 epony joined 11:33 pmurias joined
Geth rakudo: the-eater++ created pull request #3155:
use libdir instead of rebuilding path from prefix		 11:58
nine m: say "\r\n".ends-with("\n") 12:05
evalable6 False
nine That's...a little bit surprising
eater \r\n is treated like 1 character 12:07
m: say "\r\n".chars 12:08
evalable6 1
timotimo blame the unicode consortium :)
AlexDaniel m: say "\r\n" ~~ /\n$/ 12:43
evalable6 ｢␦｣
AlexDaniel m: say "\r\n" ~~ /"\n"$/
evalable6 Nil
AlexDaniel wait… what was that?
timotimo \n in a regex is a logical newline 12:44
jnthn Heh, the output inside the square brackets looks very odd in my IRC client :)
timotimo inside of a string it's the literal thing
jnthn It comes out fine locally on my terminal
m: dd "\r\n" ~~ /\n$/
evalable6 Match.new(pos => 1, hash => Map.new(()), orig => "\r\n", from => 0, made => Any, list => ()) 12:45
AlexDaniel yeah that's what I'm wondering. evalable6 does replace control characters with stuff, but… what's going on there exactly?
jnthn Beats me. But it dd's to the right thing, so...guess it's something in that mapping.
AlexDaniel github.com/perl6/whateverable/blob...e.pm6#L175 12:47
hhmmmh
m: print “\r”, “\n”
evalable6 ␦
AlexDaniel m: print “\r”
evalable6 ␍
AlexDaniel m: print “\n” 12:48
evalable6
AlexDaniel it comes out as a single grapheme so .trans doesn't work… 12:50
at least doesn't work as expected
jnthn Well, you'd need a trans entry for \r\n 12:51
AlexDaniel jnthn: is that going to help, really? What about other append/prepend combinations that result in a single grapheme?
m: say “\r\n”.trans: “\r\n” => ‘z’ 12:52
evalable6 z
AlexDaniel m: say “\r\n\c[COMBINING DIARESIS]”.trans: “\r\n” => ‘z’ 12:53
evalable6 (exit code 1) 04===SORRY!04=== Error while compiling /tmp/gTRdRJDD4X
Unrecog…
AlexDaniel, Full output: gist.github.com/8ccb4d2dce07684bc7...5b35ecebcf
AlexDaniel m: say “\r\n\c[COMBINING DIAERESIS]”.trans: “\r\n” => ‘z’ 12:54
evalable6 z̈
AlexDaniel now that works?
ok, then I guess it does work…
jnthn The grapheme break rules indicate that there's always a break after a newline
AlexDaniel oooh 12:55
m: say “\r\n\c[COMBINING DIAERESIS]”.chars
evalable6 2
AlexDaniel ok what about a prepend character… lemme find one…
timotimo that rule is very good for us btw because that means after a newline we can immediately flush a decoder's buffer 13:04
AlexDaniel ok, finally x) 13:05
m: say “\x[0600]b”.chars
evalable6 1
AlexDaniel buuut it doesn't work on \r\n, so I guess we're alright!
m: say “\x[0600]\r\n”.chars
evalable6 2
jnthn yes, iirc the rules are to break either side of a control 13:06
Though my memory is hazy thanks to samcv++ making me not need to look at this stuff for a while :D 13:07
13:07 vrurg joined
vrurg nine: reverting is not needed, I got it fixed yesterday but haven't completed testing of failed modules. 13:10
gdonald: that'd be great if you reported the problem right away.
AlexDaniel anyway, fixed: github.com/perl6/whateverable/comm...009f2fdea5 13:15
13:19 evalable6 joined, ChanServ sets mode: +v evalable6
AlexDaniel m: say "\r\n" ~~ /\n$/ 13:20
evalable6 ｢␍｣
AlexDaniel what
m: say "\r\n"
evalable6 ␍
AlexDaniel ok, that didn't work, but I have to run now…
timotimo didn.t? 13:28
13:51 Guest90858 is now known as nebuchadnezzar 13:56 lucasb joined 14:09 pmurias left 14:31 pmurias joined 14:41 pmurias left
gdonald vrurg: Sorry, I was still ruling out Travis and my own project's recent commits. 14:44
14:47 pmurias joined
AlexDaniel timotimo: yeah, definitely not. Should be ␍␤␤ 14:51
timotimo oh
i can't read the tiny symbols on my terminal
AlexDaniel vrurg: I just realized that the directories are a little bit wrong… 15:02
core.v6c not core.c :S
15:04 pmurias left 15:20 patrickb joined
vrurg AlexDaniel: I followed the pattern of core.d/core.e. Let's leave it as it is, too many changes if we rename it again. 15:59
Besides, I don't see why version is better than revision.
[Coke] yawns. 16:00
AlexDaniel vrurg: leaves more room for potential version bumps 16:12
vrurg AlexDaniel: for example?
AlexDaniel vrurg: v7.d
well I'm thinking more v7.0.0 but you get the point :) 16:13
vrurg AlexDaniel: the core is currently bound to .<rev> scheme. Too many changes. Besides, I'm about to open a problem-solving on versionized metamodel. We could have it all considered together. 16:14
jnthn I don't see why the metamodel has to be versioned 16:19
It's in a separate compilation unit because making a metaclass usable in the same compilation unit as it's written in is a pain (it may just work in Rakudo, but not in NQP) 16:20
AlexDaniel I'm not really proposing anything, but looking at the recent comments it seems like some people still don't get that we had a release, and I can't blame them because our versioning is just weird. Releasing a 7.0.0 somewhere in the future might be a good move (as in we didn't just hide perl with a different word, we're really moving forward)
jnthn There's no reason that a 6.e GrammarHOW can't be written in the 6.e BOOTSTRAP or even 6.e CORE.setting 16:21
That gives an NQP option and a Perl 6 option 16:22
I don't think we need a third place versioned
Not to mention that such things will probably be quite rare
AlexDaniel moreover, modules that `use v6` will get nice back compat even if they don't specify a letter, allowing us to do more changes…
.oO( #perl6-devspeakingoutloud )
 16:25
thinking*
vrurg jnthn: I'll try it in BOOTSTRAP and see how it works. 16:26
jnthn: in either case working on the workaround let me spot a potential problem in optimizer find_symbol_in_setting. :) 16:27
jnthn :) 16:36
vrurg AlexDaniel: I think that releasing v7 would require so much work that renaming the directory would be the least singnificant part of it. Doesn't worth efforts now. 16:41
AlexDaniel weell… 16:42
we're renaming stuff now anyway
so might as well do it properly so that all tools have to adjust once
vrurg AlexDaniel: in either case, let me finish with makefiles. Last rename brought so much headache that I felt like as much rules as possible are to be moved into a shared template. 16:47
AlexDaniel vrurg++
patrickb vrurg: How deep are you in the build file swamp? I have some changes ready soon to add --nqp-home and --perl6-home and remove --libdir from NQP and Rakudo Configure.pl... 16:58
tellable6 gist.github.com/5bd4ab58ca8eab8fef...4c304f2cdd
patrickb vrurg: It's strange how we always manage to colide when doing build system stuff. Bleh... 16:59
vrurg patrickb: I didn't see you for so long. Welcom back! :) I'd like to ask you to postpone your fixes then. The merge results in so many changes, that even minor things could complicate my work a lot. 17:00
Sorry for the inconvinience. 17:01
patrickb OK. I'll wait then. 17:02
vrurg patrickb: may I ask you to finalize fix for github.com/rakudo/rakudo/issues/3136? NQP needs @moar::cincludes@ in gcc command line to build properly. 17:06
patrickb: and also github.com/rakudo/rakudo/issues/3090 17:07
AlexDaniel patrickb: o/
timotimo at some point we should really figure out how modules should find the right compiler and flags to build native code ... 17:09
vrurg timotimo: Preserve the collected configuration vars in the compiler? 17:16
like NQP does.
Geth roast: 9513102a53 | (Ben Davies)++ | 6 files
Don't use localhost; bind to 0.0.0.0, connect to 127.0.0.1 instead

localhost can resolve to more than one address, which causes upcoming changes to MoarVM to break some tests when they shouldn't actually be breaking changes.		 17:30
timotimo vrurg: no, that's bad 17:50
vrurg: when you install a perl6 on windows, you're unlikely to have an msvc installed just like the maintainer of the package
vrurg timotimo: either way, knowning compiler flags might prove useful if one needs to build compatible binaries. 17:52
vrurg is afk, time to prepare for Dorian arrival. 17:53
timotimo best of luck
vrurg timotimo: thanks! o/ 17:54
18:06 robertle joined 18:10 Kaiepi joined 18:13 Ven`` joined 18:14 Ven`` left
patrickb timotimo: What's wrong with just taking the first compiler you can get your hands on for module native code compilation? As long as it's C and we build DLLs all should be fine. Or am I too naive? 18:39
timotimo that would be nice, for sure
also, give every possibility to override this stuff 18:40
with env vars and such
patrickb Similar to what our own Configure.pl does?
timotimo hm, i didn't actually take into account MakefileFromMeta or what it's called
patrickb I think the big question is, who should be responsible for this? Should the module installer take care of it? (Then we'd need to standardize a declarative way to describe a compilation in META6.json. It's gonna be hard to scatch every itch there is.) Or the modules themself using some build time module? (That sounds a lot like what perl5 does.) 18:44
timotimo also, cross-compilation is important to some 18:46
19:02 BeastieBot joined
Geth nqp: patzim++ created pull request #576:
Use cincludes configuration variable from MoarVM		 19:08
rakudo: patzim++ created pull request #3156:
Fixup fix for PR rakudo#3150: Var was missing `moar::` prefix		 19:11
synopsebot RAKUDO#3150 [closed]: github.com/rakudo/rakudo/pull/3150 [build] Use cincludes configuarion variable from MoarVM
Geth nqp: 75ef1a51ee | (Patrick Böker)++ | tools/templates/moar/Makefile.in
Use cincludes configuration variable from MoarVM
 19:12
nqp: 726917eaf9 | (Patrick Böker)++ (committed using GitHub Web editor) | tools/templates/moar/Makefile.in
Merge pull request #576 from patzim/rakudo_3136

Use cincludes configuration variable from MoarVM
rakudo: 3d5c5517bd | (Patrick Böker)++ | tools/templates/moar/Makefile.in
Fixup fix for PR rakudo#3150: Var was missing `moar::` prefix
rakudo: 579ac66e8b | (Patrick Böker)++ (committed using GitHub Web editor) | tools/templates/moar/Makefile.in
Merge pull request #3156 from patzim/rakudo_3136

Fixup fix for PR rakudo#3150: Var was missing `moar::` prefix
nine patrickb: we already have a declarative way to describe compilation and that's the recommended and only really supported way 19:21
modules.perl6.org/dist/Distributio...:cpan:NINE
19:38 Ven`` joined 19:52 Ven`` left 20:01 pmurias joined
timotimo i wish there was a ready-to-copy example makefile and meta6.json 20:06
or a starter template for mi6 or whatever
21:04 Ven`` joined 21:10 Ven`` left 22:32 patrickb left 22:40 evalable6 left 22:43 evalable6 joined, ChanServ sets mode: +v evalable6 22:46 lucasb left 23:02 pmurias left 23:16 Kaiepi 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!