🦋 Welcome to the IRC channel of the core developers of the Raku Programming Language (raku.org #rakulang). This channel is logged for the purpose of history keeping about its development | evalbot usage: 'm: say 3;' or /msg camelia m: ... | log inspection situation still under development | For MoarVM see #moarvm
Set by lizmat on 22 May 2021.
raydiak I just rebuilt master and cannot reproduce the error 00:04
ugexe i reverted the problem commit 00:06
`rakubrew build moar d8281722db && rakubrew switch moar-d8281722db` to get back to the problem commit
raydiak ah, got it. should have read a little more carefully :) 00:07
Geth rakudo: 2839cef6bd | (Patrick Böker)++ | src/core.c/CompUnit/Repository/Installation.pm6
Fix CURI probing on Windows

libUVs stat returns unusable writability information on Windows. It can't be trusted for actual writeability. This means `IO::Path.w` is just as unusable. Thus in `CURI.can-install()` on Windows we now actually probe for writeability by trying to create a file.
This fixes installing things on Windows with Rakudo being installed in "C:\Program Files\Rakudo".		 06:38
rakudo: 798878cdd8 | (Patrick Böker)++ (committed using GitHub Web editor) | src/core.c/CompUnit/Repository/Installation.pm6
Merge pull request #4452 from patrickbkr/fix-win-install-write-protected

Fix CURI probing on Windows
rakudo: patrickbkr++ created pull request #4455:
Create checksums for release files		 07:00
sena_kun releasable6, status 11:54
releasable6 sena_kun, Next release in ≈7 days and ≈7 hours. 2 blockers. Changelog for this release was not started yet
sena_kun, Details: gist.github.com/5175cf3729dc80356a...c980bdd2fa
Geth rakudo: 57885f6650 | (Patrick Böker)++ (committed using GitHub Web editor) | tools/templates/moar/Makefile.in
Fix `rakuw.exe` to be a non-console app

The *w.exe binaries are the "windowed" variants that don't open a console window.
  `rakuw.exe` was missing in the respective build system check.		 12:15
rakudo: 6eea26213b | (Patrick Böker)++ (committed using GitHub Web editor) | tools/templates/moar/Makefile.in
Merge pull request #4451 from rakudo/windowed-rakuw

Fix `rakuw.exe` to be a non-console app
Geth nqp: 06014d5657 | (Stefan Seifert)++ | src/core/NQPRoutine.nqp
Fix reproducible build regression

Commit d97f3084addf86a5a187d404294deda82b173b70 introduced new arrays, one of which was assembled depending on hash order. This caused the order of some objects to vary from build to build.		 13:53
rakudo: gfldex++ created pull request #4460:
fix gh##4458		 14:16
patrickb I have two Gimp image files, the backgrounds of the new Windows installer. Together they are roughly 900kb in size. Should I add them to te rakudo repo or leave them out? 14:23
sena_kun patrickb, better to be a separate repo IMO? 14:31
github.com/Raku/marketing is maybe not the right place...
patrickb, note as we build an image for every commit the additional weight may be not so desired 14:32
patrickb The images are specific to the installer, so no generic marketing resources. But it might be the practical solution. 14:33
That's why I asked. ;-)
sena_kun patrickb, we can pull them during building installers from some other repo. I agree marketing is not exactly that, just pointing out possible direction. 14:34
patrickb No. The acutal bmp images I'd keep in the rakudo repo (543kb). I was asking about the Gimp files that are the base of those bmp files... 14:37
sena_kun ah, then ignore me. :/ 14:38
patrickb Hm. I could actually move the entire binary release stuff to some place else. There is no reason for those files to end up in the source releases I'd say. 14:58
But where could they go?
New repo rakudo/release-tools ? 14:59
Hm the releasable stuff is also contained in the source release... 15:00
sena_kun This run started on 2021-07-16T15:28:25Z and finished in 43 minutes. 15:29
patrickb but then the release stuff should stay in sync with rakudo itself. So maybe it's best to just keep it there...
sena_kun yay, a new record for Blin
patrickb yay
Were there changes to the machine? 15:34
sena_kun patrickb, 5950X + Pakku. 15:36
patrickb Weee!
sena_kun with zef the numbers are similar, about 1-2 hours I think
Geth nqp/new-disp: 10131fd17d | (Jonathan Worthington)++ | src/core/dispatchers.nqp
Generify method not found handler in nqp-meth-call

Delegate to the appropriate dispatcher, such that we can have the calling language's method not found handler invoked.		 15:56
nqp/new-disp: 1169dd49f9 | (Jonathan Worthington)++ | t/nqp/097-hll.t
Adapt test to new method not found API
nqp/new-disp: 1d6eefd3be | (Jonathan Worthington)++ | t/nqp/097-hll.t
Make sure test type gets a HLL set
rakudo/new-disp: f47d8b8b84 | (Jonathan Worthington)++ | src/Perl6/bootstrap.c/BOOTSTRAP.nqp
Adapt to method not found handler API change

On MoarVM, after new-disp, it now also receives the arguments. We slurp them, which is conveniently backward compatible with other backends not sending them along.		 15:58
patrickb I just uploaded an MSI of Rakudo 2021.06 to rakudo.org. Testers welcome. 17:22
[Coke]_ patrickb++ 17:28
Altai-man patrickb++ 17:49
ugexe patrickb: ci.appveyor.com/project/ugexe/zef/...s/40025155 21:29
maybe creating a file isn't sufficient and it needs to create a directory? 21:32
just a wild guess
patrickb Is C:\projects\foo writable?
That worked before I assume? 21:33
ugexe it worked before, yes
patrickb There is nothing special about that folder, right? Then why should the probing fail? Wait, does that folder possibly simply doesn't exist yet? 21:34
ugexe it would get created by CURI i believe 21:35
yeah im confused... doesnt seem like it should happen 21:37
patrickb That check is in install(). If no prior call to can-install() happens that error could actually be explained.
Because writeable-path() does not try to create the folder if it doesn't exist. 21:38
ugexe yeah i guess there would be no prior call in that zef example because it doesnt use `auto`, it explicitly overrides that with -to=inst#foo sp yeah
patrickb Wait. No, that's not true. It's the private !writeable-path() that is called. That one does actually call can-install() 21:39
ugexe oh in that appveyor output it shows that message twice 21:40
patrickb Or is there some non-obvious behavior when a private and public method are named the same?
melezhik . 21:43
releasable: status
releasable6 melezhik, Next release in ≈6 days and ≈21 hours. 3 blockers. Changelog for this release was not started yet
melezhik, Details: gist.github.com/3ab4f4e113dcf9fcbf...95932430b5
patrickb that foo folder actually does not exist prior to trying to install to it, IIUC. 21:54
ugexe right
self!prefix-writeable || ?(!$.prefix.e && try { $.prefix.mkdir } && $.prefix.e); 21:56
presumably it doesnt reach the mkdir there
anymore
patrickb I know
the thing caches
the very first time when the dir doesn't exist it remembers the failure 21:57
ugexe self!prefix-writeable is cached, and if its caching a failure i would expect the above to fall into the rhs of the || 21:58
patrickb Right 22:00
But still. 22:05
The negative test is cached and the dir is created afterwards. All following calls will return False. 22:06
Because the writability test fails and the test for non-existance fails as well. 22:07
ugexe maybe s{//=}{||=}
patrickb Ha! I had the same idea! 22:08
Having the check for writability before the check of existence is a performance optimization for the common path of the repo existing, correct? 22:10
ugexe github.com/rakudo/rakudo/commit/49...0af1b0fcb7 22:13
that being said i dont remember exactly why
patrickb Kind of makes sense to not blindly create loads of repos on disk when we might never actually install anything into them... 22:14
Geth rakudo: patrickbkr++ created pull request #4461:
Fix installation to to-be-created CURIS on Windows		 22:17
patrickb ugexe: Shall we just merge and see what the CI thinks of it? (I'd need to reboot and build a Rakudo with a build time > 20 minutes.) 22:19
ugexe ci.appveyor.com/project/ugexe/zef/...s/40025829 22:20
i just made a zef PR that checks out that commit
s/PR/branch/ 22:21
oops my incantation was wrong heh
patrickb I need to go to bed. I'll have a look tomorrow how it went. 22:45
ugexe: Thanks for having a look with me.
Geth rakudo: d668d99292 | (Patrick Böker)++ (committed using GitHub Web editor) | src/core.c/CompUnit/Repository/Installation.pm6
Fix installation to to-be-created CURIS on Windows (#4461)

The negative test for writeability when the repo was still not existing was cached and spoilt all further checks. Now we only cache a successful writeability check.		 22:48
Geth nqp/new-disp: 1c859c3a57 | (Jonathan Worthington)++ | 2 files
Add nqp-find-meth dispatcher

Which will be used to implement findmethod/tryfindmethod/can on the MoarVM backend.		 23:25

Powered by Cro and the Raku® Programming Language.

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

Thank you!