japhb AlexDaniel: Yes, the precomp bug is separate from the zef issue (that I worked around by using zef 0.7.4). The precomp bug does appear to be fixed by the bump -- but now something that I expected would happen at run time appears to be happening at precomp time, so I need to figure that out. 01:01
AlexDaniel: It's not clear if the new behavior is a new bug, or if I had previously been depending on an old one.
timotimo japhb: have you seen the new --stagestats? 01:12
it's nested, so it'll show what is causing what to be precomped
japhb timotimo: Ah thanks, saw that in passing but had forgotten about it 01:29
Let me rebuild-all with a working zef (ugexe is having me try newer versions, but I'm still getting issues), and I'll try --stagestats on my precomp problem. 01:30
01:31 squashable6 left 01:32 squashable6 joined
AlexDaniel there we go: github.com/perl6/whateverable/comm...d51319508a 01:42
sorry it took so long :)
01:43 tellable6 left 01:45 tellable6 joined
Geth_ ¦ rakudo: vrurg self-assigned Dynamic variables set via PROCESS:: doesn't have .dynamic set in their container. github.com/rakudo/rakudo/issues/3270 01:50
01:59 toddr joined 02:00 AlexDaniel left 02:01 toddr left 02:16 Kaeipi joined 02:18 redable joined, Kaiepi left 02:29 Kaeipi left 02:30 Kaiepi joined 02:32 redable left 02:34 redable joined 02:43 redable left 02:45 AlexDaniel joined, AlexDaniel left, AlexDaniel joined 03:09 Kaiepi left 03:11 Kaiepi joined, Kaiepi left 03:12 Kaiepi joined 03:17 Kaiepi left 03:34 MasterDuke left 05:30 squashable6 left 05:32 squashable6 joined 06:13 sena_kun joined 06:15 sena_kun left 07:03 japhb joined
ZzZombo What is Truffle? 07:21
08:11 ZzZombo_ joined 08:14 ZzZombo left, ZzZombo_ is now known as ZzZombo 08:45 pmurias joined
pmurias ZzZombo: a piece of tech which generates a JIT that can be plugged in into a JVM from an annotated interpreter written in Java 08:58
ZzZombo: in a way it's an actual performant Parrot 08:59
09:03 ExtraCrispy joined
lizmat Files=1277, Tests=109553, 206 wallclock secs (28.52 usr 7.80 sys + 2910.31 cusr 258.78 csys = 3205.41 CPU) 09:04
09:11 pmurias left 09:12 ExtraCrispy left 09:13 pmurias joined 10:00 pmurias left
tbrowder AlexDaniel: since release seems not so imminent, how do you feel about merging my two PRs? each is a single commit and should be easily reversible, and each was fully spec tested. 11:13
11:59 pmurias joined 12:05 redable joined 12:33 sena_kun joined
nine releasable6: status 12:37
releasable6 nine, Next release will happen when it's ready. 2 blockers. 7 out of 450 commits logged (⚠ 9 warnings)
nine, Details: gist.github.com/3c70e7cfb537866d0f...f1fecaa5d6
nine If we ever want to get a release out, we really need to focus on blockers and regression fixes. If we do that and get back to timely and frequent releases, it won't matter as much anymore when valuable improvements like tbrowder's have to wait for next month's release. 12:44
12:47 ggoebel joined
Geth_ nqp/truffle: 5352e59145 | (Paweł Murias)++ | src/vm/jvm/Truffle.nqp
[truffle] Implement ++$var and --$var (nqp::preinc and nqp::predec)
12:52
nqp/truffle: 6eecd1691b | (Paweł Murias)++ | t/nqp/013-op.t
Test ++$var and --$var
12:57 Altai-man_ joined 12:59 sena_kun left
Geth_ nqp/truffle: acafae770f | (Paweł Murias)++ | 2 files
[truffle] Add a partial and slow nqp::bindattr
13:06
pmurias .tell MasterDuke current the point example is *slower* on truffle then on the original backend, but they way method call/attribute access is implemented is so *unoptimal* we are getting warnings from truffle that things are amiss 13:10
tellable6 pmurias, I'll pass your message to MasterDuke
13:14 redable left 13:34 lucasb joined
tbrowder nine: thanks. i'm not trying to be pushy, it just seems that the blockers are too tough to solve any time soon. i am happy to help any way i can, but i'm afraid all i can do is be a cheer-leader. 13:47
if i can help with testing i will, just let me know. 13:48
13:56 redable joined 14:12 Kaiepi joined
rba Answering now are: design.raku.org/ (no auto-update ask rba) docs.raku.org/ (without content yet; asked jjmerelo) marketing.raku.org/ (share hypnotoad with marketing.perl6.org) modules.raku.org/ (share hypnotoad with modules.perl6.org) and since last Saturday raku.org 14:14
If you find any dead link, you may try to fix it in the repo the website is based on, or just let me know. 14:15
14:17 redable left
pmurias .tell MasterDuke I'll work on replacing our method call implementation with something that's optimizable by truffle 14:25
tellable6 pmurias, I'll pass your message to MasterDuke
pmurias the stable of a type object never changes? 14:37
AlexDaniel .seen Xliff 14:40
tellable6 AlexDaniel, I saw Xliff 2019-10-25T06:53:50Z in #raku-dev: <Xliff> o/
14:55 Kaiepi left 14:59 Altai-man_ left 15:03 Kaiepi joined 15:29 Kaiepi left 15:38 Kaiepi joined
AlexDaniel nine: I'm currently benchmarking github.com/rakudo/rakudo/issues/3268 15:45
it will take some time to confirm it and bisect it afterwards…
dogbert17 AlexDaniel: I have pushed a PR which should fix M#1208 15:47
AlexDaniel I see it, yes
dogbert17: why not use orig_utf8 instead? 15:50
the code looks convoluted and extra juggling with variables doesn't really help 15:51
dogbert17: I mean orig_utf8[error_pos - 2] and such 15:52
dogbert17 well either changing ~10 lines or one. I could have done it either way though.
there's indeed quite a bit of juggling in that code :) 15:53
AlexDaniel you mean 3 lines vs 1
but it'll read much better
I mean, reading it will be slightly easier if orig_utf8 is used, I guess that's what was meant anyway 15:54
15:54 |Tux| joined
AlexDaniel squints 15:55
dogbert17 can I git amend the change as long as it hasn't been merged?
AlexDaniel dogbert17: yep
dogbert17: what's the logic behind `if (error_pos >= 3) {` ? I don't get it 15:56
15:58 |Tux| left
AlexDaniel dogbert17: ah it was merged :) 15:58
nine oops...I just merged it
dogbert17 nine++
15:58 |Tux| joined
nine dogbert17: FWIW now that I've read your commit message, the error is also quite obvious :) Didn't see it the first time I had a look 15:58
dogbert17 gdb FTW 15:59
AlexDaniel nine: ok then maybe you can tell me what's `if (error_pos >= 3) {`
16:00 |Tux| left
AlexDaniel there are 4 different error messages just for the first 4 characters in the string? 16:00
or what am I missing?
16:00 |Tux| joined
dogbert17 AlexDaniel: you're squinting too hard :) 16:01
the way I interpret it is that the code want to print the first erroneous byte and a couple a bytes preceeding it. Depending on where in the sting the error the number of bytes which can be printed out differs 16:04
AlexDaniel also the whole idea of parsing the string again just to get the line number is uhhh… I don't know… I guess it's faster this way for strings that are not malformed? Is it though?
dogbert17 *Depending on where in the sting the error occurs
haha, my spelling is atrocious 16:05
AlexDaniel dogbert17: so 3 different error messages for the first 3 codepoints
dogbert17 I would say so yes
releasable: next 16:07
releasable6 dogbert17, Next release will happen when it's ready. 1 blocker. 7 out of 450 commits logged (⚠ 9 warnings)
dogbert17, Details: gist.github.com/2a2a41f80717b48625...5a42f5b9b6
dogbert17 there's hope, only one blocker left 16:08
nine yes
AlexDaniel and I'm testing it right now 16:09
dogbert17 should be a pretty stable release methinks
nine And samcv++ optimized UTF-8 decoding a lot and some decisions were based on helping the compiler with autovectoring the code. Pessimising the error case sounds entirely sensible
AlexDaniel yep it'll be awesome
dogbert17 AlexDaniel: how are the bots running nowadays, are they stable? 16:10
AlexDaniel yep no problems at all
dogbert17 and gumbo works as well
AlexDaniel yep
dogbert17 yay
AlexDaniel ok, the more I look at that code, the more I don't get it
dogbert17 the decoding stuff? 16:11
16:12 pmurias left
AlexDaniel ah, nevermind, was squinting too hard indeed :) 16:15
dogbert17: anyway :) github.com/MoarVM/MoarVM/commit/48...9c57ac0877 16:20
dogbert17 AlexDaniel++ 16:21
Geth_ nqp: 780616f4eb | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/templates/MOAR_REVISION
[MoarVM Bump] Brings 3 commits

MoarVM bump brought: github.com/MoarVM/MoarVM/compare/2...g48491b896 48491b896 Make ut8_decode code just a bit easier to understand aba68e41e Merge pull request #1210 from dogbert17/fix-m-1208 da19661e8 Reset utf8 pointer to point at original string before using
rakudo: 9f433da0d6 | (Aleks-Daniel Jakimenko-Aleksejev)++ | tools/templates/NQP_REVISION
[NQP Bump] 780616f4e [MoarVM Bump] Brings 3 co […]

NQP bump brought: github.com/perl6/nqp/compare/2019....g780616f4e
AlexDaniel *utf8* code, of course, dammit :)
dogbert17: spelling :D
dogbert17 sometimes your keyboard works against you :) 16:22
AlexDaniel nine: I have not seen latest blin results, btw :S 16:25
it should be absolutely clean given the output before that, but hmm
kawaii: ping? :)
16:39 Kaiepi left 16:42 Kaiepi joined 16:50 rba[m] joined 17:04 MasterDuke joined 17:22 patrickb joined 17:24 vrurg left 17:38 cognomin_ joined
patrickb When was the last rakudo release? 17:41
releasable spits out commits from as far back as may... 17:42
17:43 cognominal left
AlexDaniel patrickb: they were not part of the release then 17:44
patrickb: 2019.07.1 was the last one
timotimo long-lived branches could be a cause of that
AlexDaniel and you can make a commit with any date, really :) 17:45
timotimo that's also true
you can even get a commit hash based on your wishes 17:46
takes a bit of number crunching of course
patrickb long lived branches it is I guess. I never faked history in such a way...
AlexDaniel well, git history is the only true history :) 17:47
17:52 b2gills left 17:56 b2gills joined
AlexDaniel ta-daaa 17:58
github.com/rakudo/rakudo/issues/32...-547071572
releasable6: status
releasable6 AlexDaniel, Next release will happen when it's ready. There are no known blockers. 7 out of 451 commits logged (⚠ 9 warnings)
AlexDaniel, Details: gist.github.com/d11fb9a9186d9a2dd4...9bde7e7328
patrickb I have just logged all of my commits 18:08
18:53 Kaiepi left 18:54 Kaiepi joined
MasterDuke . 18:59
tellable6 2019-10-28T14:25:43Z #raku-dev <pmurias> MasterDuke I'll work on replacing our method call implementation with something that's optimizable by truffle
hey MasterDuke, you have a message: gist.github.com/05452c25906e3c7d42...9bb21aa99f
AlexDaniel patrickb: niiice! Thank you 19:02
MasterDuke pmurias: cool. fwiw it's not *too* much slower on my machine. for 10000000, just got 13s with regular jvm backend and 19s for truffle (but yeah, lots of complaints)
AlexDaniel patrickb: thoughts on making a raku symlink?
tellable6 MasterDuke, I'll pass your message to pmurias
MasterDuke pmurias: oh, that was without the custom new(). with that the regular jvm backend was 3.5s and truffle was 5.7s 19:05
tellable6 MasterDuke, I'll pass your message to pmurias
patrickb AlexDaniel: Can you elaborate?
AlexDaniel patrickb: currently we create a bunch of perl6-* binaries, but people want to have `raku` in their shebang and use `raku` on the command line 19:06
patrickb: I guess ideally we would like to only generate raku-* binaries in the far future. But right now, should we create a symlink?
patrickb I think there was some talking of actually only having a rakudo binary and making raku und perl6 only symlinks 19:07
*and
AlexDaniel right 19:08
patrickb I dunno if we want to go down that route. 19:09
Given symlinks are not a thing on all platforms we want to serve, we might need to copy on some.
I guess the difficult question is not how to do this from a technical perspective. 19:10
It's rather how we want to have it.
rakudo binary + perl6 symlink + raku symlink?
raku binary + perl6 symlink?
MasterDuke what platform doesn't support symlinks? 19:11
patrickb MasterDuke: I'm not sure about what side effects symlinks have on Windows. 19:12
Also I have no information about the more obscure OSes like AIX 19:13
AlexDaniel “Now in Windows 10 Creators Update, a user (with admin rights) can first enable Developer Mode, and then any user on the machine can run the mklink command without elevating a command-line console. 19:14
What drove this change? The availability and use of symlinks is a big deal to modern developers”
patrickb On linux it's a common thing to just symlink executables. I have rarely seen this on Windows. Of course there are plenty of those on your desktop, but can one actually execute them?
MasterDuke pretty sure you can 19:15
(but also don't have a windows system to test that with)
AlexDaniel I guess you can but that's relatively new
patrickb are we talking about the same thing? I'm talking about *.lnk files
they even have the wrong suffix! :-( 19:16
AlexDaniel patrickb: it feels like these are different
but last windows I actually used was XP 19:17
patrickb reboots into a Windows.
MasterDuke yeah, pretty sure windows does also have real symlinks
19:18 patrickb left
MasterDuke but i won't bet huge amounts on that assertion 19:18
19:28 patrickb joined 19:29 robertle joined 19:51 patrickb left 20:14 patrickb joined
nine .lnk files are more akin to .desktop files. The application needs to understand them. Windows has supported symlinks for decades but only on NTFS. They are transparent to applications but have (had) quite a few restrictions 20:27
patrickb Is it possible to include a CompRepo folder using -I ? 20:31
20:39 patrickb left 20:40 Kaiepi left 20:41 Kaiepi joined
lizmat And yet another Rakudo Weekly hits the Net: rakudoweekly.blog/2019/10/28/2019-...e-toaling/ 20:49
20:53 patrickb joined
lizmat notable6: weekly 20:54
notable6 lizmat, 4 notes: gist.github.com/b6fc89d24cc7036be6...ccd362dc25
20:55 samcv left
lizmat notable6: weekly reset 20:59
notable6 lizmat, Moved existing notes to “weekly_2019-10-28T20:59:26Z”
21:00 samcv joined 21:04 pmurias joined 21:37 samcv left
pmurias MasterDuke: it's not that much slower on machine, but we want the truffle backend to be *faster* ;) 21:44
tellable6 2019-10-28T19:02:54Z #raku-dev <MasterDuke> pmurias: cool. fwiw it's not *too* much slower on my machine. for 10000000, just got 13s with regular jvm backend and 19s for truffle (but yeah, lots of complaints)
2019-10-28T19:05:12Z #raku-dev <MasterDuke> pmurias: oh, that was without the custom new(). with that the regular jvm backend was 3.5s and truffle was 5.7s
21:45 ggoebel left 22:11 pmurias left 22:13 pmurias joined 22:40 samcv_ joined
AlexDaniel samcv_: so there's nothing preventing a moar release currently, I'm still waiting for Blin results but I'm relatively sure it's going to be clean 22:43
japhb lizmat: I can't tell if there's a pun I don't get, but shouldn't the last word of the weekly title be 'Tooling' rather than 'Toaling'? 22:45
MasterDuke pmurias: heh, yeah, i certainly hope it gets faster. i was looking at that pdf you linked recently. it has a bunch of stuff about optimization and how/when/where/what, but i couldn't really tell what was applicable (or how) 23:18
samcv_ AlexDaniel, roger 23:29
23:31 patrickb left
Geth_ nqp/truffle: c01384a028 | (Daniel Green)++ | src/vm/jvm/runtime/org/perl6/nqp/truffle/io/StandardReadHandle.java
[truffle] Remove unused import
23:55