🦋 Welcome to Raku! raku.org/ | evalbot usage: 'p6: say 3;' or /msg camelia p6: ... | irclog: colabti.org/irclogger/irclogger_log/raku
Set by ChanServ on 14 October 2019.
00:01 devmikey left 00:10 xinming left, xinming joined 00:14 molaf joined 00:29 skyl4rk_ joined, skyl4rk left, skyl4rk_ is now known as skyl4rk 00:30 Kaiepi left, Kaiepi joined 00:32 ragekagemage left 00:41 p6steve joined 00:46 p6steve left 00:48 mowcat left 01:03 mowcat joined, titsuki left 01:22 Skarsnik left 01:32 xinming left 01:37 xinming joined 01:49 molaf left 02:00 Sgeo__ joined 02:01 molaf joined, gnufr33dom joined 02:03 xinming left, xinming joined 02:04 Sgeo_ left 02:26 stoned75 left 02:27 pilne left 02:42 p6steve joined 02:46 p6steve left 03:12 xinming left 03:47 _dev_zero left 03:48 holyghost left 03:49 _dev_zero joined 04:05 molaf left 04:08 marcusr joined, brtastic joined 04:09 bdju joined 04:14 mowcat left 04:29 farcas1982regreg joined 04:42 p6steve joined 04:45 stoned75 joined 04:47 p6steve left 05:19 sauvin joined 05:21 rindolf joined
Geth doc/test-for-perl-method: bcf9c26443 | (Stoned Elipot)++ | 2 files
Test for discouraged usage of .perl in sample code

If needed the 'perl' method can be allowed in sample code with the :ok-test<perl> annotation.
05:25
05:29 aborazmeh left
Geth doc: stoned++ created pull request #3359:
Test for discouraged usage of .perl in sample code
05:31
05:42 brtastic left 05:43 holyghost joined 05:54 stoned75_ joined 05:58 stoned75 left 05:59 sjm_uk joined 06:01 p6steve joined 06:06 p6steve left 06:15 |Sno| left, [Sno] joined 06:25 anatofuz_ joined 06:27 anatofuz_ left 06:28 anatofuz joined
Geth doc/test-for-perl-method: 595108c1cc | (Stoned Elipot)++ | writing-docs/EXAMPLES.md
The 'perl' method usage is discouraged

Also mention the alternative, 'raku'.
06:39
doc: bcf9c26443 | (Stoned Elipot)++ | 2 files
Test for discouraged usage of .perl in sample code

If needed the 'perl' method can be allowed in sample code with the :ok-test<perl> annotation.
06:44
doc: 595108c1cc | (Stoned Elipot)++ | writing-docs/EXAMPLES.md
The 'perl' method usage is discouraged

Also mention the alternative, 'raku'.
doc: 35155512ef | (Juan Julián Merelo Guervós)++ (committed using GitHub Web editor) | 2 files
Merge pull request #3359 from Raku/test-for-perl-method

Test for discouraged usage of .perl in sample code Thanks a lot!
06:47 stoned75_ left 06:49 stoned75 joined 06:50 kensanata joined 06:56 andrzejku joined 07:01 brtastic joined 07:03 samcv joined, freezing_ left 07:04 tigerpaws left 07:05 tigerpaws joined, MasterDuke joined 07:06 freezing_ joined 07:09 orinthe left, gdonald left, Ulti left, KindOne left, a3r0 left, Celelibi left, ribasushi left 07:10 brtastic left, ribasushi joined 07:11 brtastic joined 07:13 Geth left, Geth joined 07:14 tigerpaws left 07:15 tigerpaws joined, KindOne joined, orinthe joined, gdonald joined, Ulti joined, a3r0 joined, Celelibi joined, sacomo joined, erdic joined, jjatria joined, sivoais joined, masak joined, ilbelkyr joined, autarch joined, moritz joined, cxreg joined 07:20 tigerpaws left, tigerpaws joined 07:21 sivoais left 07:22 sivoais joined 07:25 tigerpaws left, tigerpaws joined 07:26 farcas1982regreg left 07:34 gnufr33dom left 07:39 expon joined 07:41 mahafyi joined 07:42 SCHAPiE left 07:44 patrickb joined 07:45 p6steve joined 07:48 pecastro joined, sena_kun joined 07:51 stoned75 left 07:53 kst`` left 07:54 SCHAPiE joined 07:56 stoned75 joined 08:07 Kaiepi left 08:08 Kaiepi joined 08:18 xinming joined 08:30 rindolf left 08:34 elcaro left 08:35 elcaro joined
El_Che releasable6: status 08:40
releasable6 El_Che, Release date for Rakudo 2020.05 is listed in “Planned future releases”, but it was already released.
El_Che, and I oop! Backtrace: gist.github.com/ebdccf2240cce4924b...a378f7cd11
El_Che so, is it released? 08:41
08:46 rindolf joined
AlexDaniel sena_kun: ↑ 08:47
08:48 xinming left, xinming joined 08:49 p6steve left 08:51 jjmerelo joined 08:52 sena_kun left, sena_kun joined 08:57 jjmerelo left 09:07 Altai-man_ joined, chloekek joined 09:09 sena_kun left
El_Che is the arrow for the release or the backtrace? 09:32
fyi: now that ubuntu dropper i386 I am moving i386 builds to debian
09:34 jjmerelo joined 09:38 xinming left 09:39 xinming joined 09:44 vike left
lizmat clickbaits rakudoweekly.blog/2020/05/04/2020-...community/ 09:54
Geth ¦ problem-solving: AlexDaniel assigned to jnthn Issue A way to control Rat's behavior when it is running out of range for its denominator github.com/Raku/problem-solving/issues/185 09:55
El_Che lizmat: I am building the new release at we speak 09:58
Can we cheat? :)
lizmat no, it will be in next week's RWN :-)
perhaps with a Rakudo Star to go with it? 09:59
lizmat hopes
El_Che lizmat: I haven't followed the * releases for some time, but I think it's a while from the last release 10:00
lizmat 2020.01 was the last one 10:01
El_Che oh, that's ok, then
seems on schedule
Altai-man_ We don't have tarballs on the website yet 10:03
And I don't know who does that.
lizmat perhaps rba knows? 10:07
El_Che Altai-man_: I am building compiled tar.gz if that helps
(although reflecting on what you said, you mean source tar balls) 10:08
Altai-man_ Oh, no, sources are already there. 10:09
The website just suggests me precompiled ones.
Ok, this means I can send a happy email? I think last time there was a suggestion not to do that until sources or builds are not ready, but I don't remember clearly. :S 10:10
10:24 Black_Ribbon left 10:29 dogbert17 joined, timotimo left, ccamel left, perlmaros left, dogbert11 left, patrickb left, perlmaros joined, camelCaser joined 10:30 dg left, timotimo joined 10:32 dg joined 10:41 farcas1982regreg joined
Altreus is a react block sufficient to keep a lexical variable alive? 10:42
Does it close over it?
e.g.: « my $something-lexical-with-a-supply; start react whenever $something-lexical-with-a-supply.supply { ... } » 10:43
10:43 natrys joined
Altreus will this immediately stop? 10:43
(ironically because of the start)
In reality I'm sending a websocket around the place, but as soon as I try to use it, it seems to be closed - I'm wondering if it's going out of scope without telling me 10:44
lizmat well, it feels wrong to rely on a lexical outside of a start because of possible race conditions
Altreus I just realised the real code is a link away 10:45
github.com/shuppet/p6-api-discord/...n.pm6#L102
10:45 p6steve joined
Altreus I also just realised this is only a problem the /second/ time 10:45
lizmat does it have a DESTROY method? perhaps that can tell you whether it really gets lost or not
well, that feels like a race condition then 10:46
Altreus I agree, but I cannot figure it out at all
I made it lexical so I could be sure /which/ websocket I was working with
jnthn It closes over it, but it will also see any mutations
e.g. it doesn't capture the value, it captures the container 10:47
Altreus So the line above gets run a second time in the same runtime; Discord tells me to reconnect, so I make a new websocket via the same path
Which as far as I can tell is a new everything - new variable, new container, new lexical, new react block etc
lizmat so why aren't the first two lines in !on_ws_connect not inside of the start block ?
Altreus Should they be? I don't have an answer either way 10:48
It didn't seem relevant to the start or the react, so I didn't put it in
What actually happens is when this line runs it is closed and thus crashes -> github.com/shuppet/p6-api-discord/...n.pm6#L181 10:49
but this only happens the second time around
I'm sure it's related to how these threads cross over in the properties of the Connection object itself...
seems like where a race condition should show up 10:50
10:50 p6steve left
Altreus I should probably rethink the semantics of my variables 10:50
10:52 squashable6 left
Altreus afk for groceries :) 10:52
10:53 squashable6 joined 11:03 caterfxo left 11:06 andrzejku left 11:07 sena_kun joined 11:09 Altai-man_ left 11:15 farcas1982regreg left 11:19 andrzejku joined, renormalist joined 11:23 patrickb joined 11:24 caterfxo joined
tbrowder hi, all 11:41
is there any way to trap a core dump when writing a module test? 11:42
could eval-dies-ok be made to work somehow? 11:43
jnthn No, you'd need to run a separate process really 11:45
And then look at the signal
11:50 caterfxo left, caterfxo joined
tbrowder ok, thnx, suspicions confirmed! 12:05
12:17 anatofuz left, epony left 12:19 natrys left
samebchase- Hi, I have a question. I appreciate how every time I try to install something using zef it runs all the tests. I was wondering kind of vetting is done so that no malicious code runs in the tests. 12:29
12:29 chloekek left
jjmerelo samebchase afraid not... 12:30
So it's caveat emptor territory here
12:32 molaf joined
lizmat samebchase-: I think you can install without having the tests run 12:33
--/test
jnthn Yes, but...then you'll want use the module? And it should be just as evil? 12:34
*want to
Um, and could be. :)
samebchase- jnthn: I was just thinking that haha :-)
IIRC there was a bug in utop (OCaml some time ago where a badly formatted bash script or something would end up doing a "rm -rf" 12:35
12:37 molaf left 12:47 jjmerelo left 12:54 plicease joined 13:00 Aussieteen joined 13:01 epony joined 13:03 Aussieteen left 13:04 stoned75 left 13:05 stoned75 joined, aborazmeh joined, aborazmeh left, aborazmeh joined 13:07 Altai-man_ joined
El_Che weekly: new rakudo 2020.05 packages for Linux distributions, addition of Debian 10 i386 packages (Ubuntu dropped i386 from 20.04 onwards): nxadm.github.io/rakudo-pkg/ 13:07
notable6 El_Che, Noted! (weekly)
MasterDuke why would github.com/rakudo/rakudo/blob/mast...m6#L31-L33 allocate a BOOTHash for every call? 13:08
13:09 sena_kun left
MasterDuke p6scalarwithvalue looks pretty simple github.com/rakudo/rakudo/blob/mast...#L778-L791 13:10
jnthn The implicit *%_ (the allocation of that should be elimianted by spesh, but it'll be allocated up until then) 13:12
timotimo we've got enough cases where *@ and *% are still created, but then never used, in my experience 13:13
Altai-man_ El_Che, hopefully, everything builds fine?
timotimo possibly related to the osr thing we stumbled upon the other day
MasterDuke the profile reports 10001000 of them allocated 13:14
same with Scalar 13:16
13:21 chloekek joined
El_Che Altai-man_: everything built very fine 13:21
even 386
(because of the perf test made optional)
Altai-man_++
Altai-man_ still can't believe
El_Che Altai-man_: you should start to :) 13:22
13:24 vike joined
MasterDuke timotimo: what was the OSR thing? i've forgotten. the profile reports only 2 OSRs, but neither in the slow function 13:27
timotimo when we optimize a frame by doing OSR, we're essentially entering the code at a different block from BB 0, so there's an artificial edge between BB 0 and the one where the osrpoint is 13:28
which can give us "phantom" register versions that spesh knows nothing about
which can prevent spesh from optimizing things
MasterDuke hm. think that might be the problem here? even though there's no OSR for the "problem" function? 13:33
Geth ¦ problem-solving: Skarsnik assigned to AlexDaniel Issue Nativecall, native trait library usage rework github.com/Raku/problem-solving/issues/186 13:39
timotimo perhaps not 13:40
MasterDuke seems like it might be something with spesh, given how expensive MVM_spesh_arg_guard_run and resolve_using_guards are. any suggestions for debugging? 13:43
13:44 mowcat joined 13:45 squashable6 left, squashable6 joined
MasterDuke and the difference between what the profile and the spesh log report about whether the function is speshed/jitted or not 13:47
13:49 rainmanjam left 13:53 Skarsnik joined 13:55 stoned75 left, stoned75 joined 14:06 wamba joined 14:07 wamba left 14:15 aborazmeh left 14:16 natrys joined 14:17 farcas1982regreg joined
Geth ¦ problem-solving: AlexDaniel assigned to jnthn Issue Nativecall, native trait library usage rework github.com/Raku/problem-solving/issues/186 14:18
¦ problem-solving: AlexDaniel self-unassigned Nativecall, native trait library usage rework github.com/Raku/problem-solving/issues/186
14:23 caterfxo left, Doc_Holliwood joined
Doc_Holliwood m: say (1..10).grep({ $_ ~~2 ff^ $_ ~~4 }) # can i do this without an explicit block? 14:24
camelia (2 3)
Doc_Holliwood m: say (1..10).grep( * ~~ 2 ff^ * ~~ 4 )
camelia ()
14:48 jjmerelo joined 14:49 patrickb left
samebchase- I am using `mongo-perl6-driver`, and it is taking me 20+ seconds to fetch a single object from the DB. Is this expected? 14:53
or should I try some other library
[Coke] I can't comment on whether or not that's *expected*, but damn is that *slow* 14:54
tadzik is it 20+ seconds for the first object, or every single fetch?
maybe it's just slow to load 14:55
14:59 patrickb joined
samebchase- Two queries one after the other are also taking 20 seconds. Connection initialization seems to be happening fast. What do you all use to profile your programs? 14:59
Or am I running into some timeout, due to a config messup 15:01
$collection.find seems to be taking all the time 15:02
Setting up the client connection, database, and collection happen fast, and then the program hangs for 20seconds to do the $collection.find(); I do have an index on the field I am querying etc. 15:04
15:07 sena_kun joined 15:09 Altai-man_ left 15:18 brtastic left
timotimo samebchase-: you can use raku --profile or using the Telemetry module (comes with rakudo) to time individual stuff 15:23
15:23 stoned75 left
timotimo hold up, Telemetry doesn't time individual sections 15:23
but running your program with -Msnapper can get you some info about what the program is doing over time 15:24
15:24 _jrjsmrtn left
timotimo the defualt --profile doesn't give very reliable results when your program uses multiple threads, which i would think the mongo p6 driver probably uses for something? 15:25
15:25 __jrjsmrtn__ joined
timotimo so you can --profile=/tmp/blah.sql and feed that into moarperf: github.com/timo/moarperf/releases/...6efe9a9874 should work (but needs some other dependencies from zef) 15:26
15:35 Doc_Holliwood left 15:39 stoned75 joined 15:40 caterfxo joined 15:41 caterfxo is now known as Guest66327, caterfxo joined 15:44 brtastic joined 15:45 Guest66327 left
Geth ¦ problem-solving: JJ assigned to jnthn Issue Use a grammar for command line options github.com/Raku/problem-solving/issues/187 15:45
15:46 MilenaDevi joined 15:48 MilenaDevi left 15:49 kensanata left 15:59 Doc_Holliwood joined 16:05 xinming left 16:06 xinming joined 16:07 farcas1982regreg left
samebchase- timotimo: thanks, will try out --profile 16:14
timotimo just running the program with "time" will also tell you if it has been doing a lot of waiting, which you can tell by the user and system times being shorter than the wallclock time it took to run 16:19
16:21 andrzejku left, brtastic left 16:23 KindOne left
samebchase- real 0m22.649s 16:34
user 0m2.960s
sys 0m0.096s
yeah, you are right, there is a huge difference between the various times 16:35
timotimo sys being low is fine, but the user time being so low looks a lot like it's doing a whole load of waiting 16:36
"user" can, at least for single-threaded workloads, be interpreted somewhat like "if there was no reason to wait, the whole thing could have taken this long" 16:37
Geth ¦ problem-solving: JJ assigned to jnthn Issue Should we have named arguments in test routines? github.com/Raku/problem-solving/issues/188 16:41
samebchase- I have filed an issue on Github, with the snippet I am running. Perhaps, the maintainer of the library might be able to point out if I am doing something wrong. I am following the documentation very closely... 16:42
timotimo you could try recording the network traffic with for example wireshark, and see where the pause(s) occur 16:43
16:46 molaf joined
samebchase- I am running the program inside a VM 16:49
:-D
I tried increasing the logging on Mongo, so that I can see the query happening, but couldn't get it working... this will take some more investigation 16:50
jjmerelo samebchase did you try to run it with some other library or mongo CLI to rule that out?
samebchase- yes, Mongo CLI is instant
I chained together five queries, and it takes the same time as a single query 16:51
22 secs
16:53 KindOne joined
timotimo did you put some output between the lines in the raku script? 16:53
it's possible, but i wouldn't expect it to be the case, that compile time is high for some reason
try raku --stagestats blah.rk
samebchase- github.com/MARTIMM/mongo-perl6-driver/issues/31 16:54
you can see the program there. I added a `say` after every line
Everything up to `coll.find` happens instantly
and then it hangs for a long time before printing the output
--stagestats shows that parsing took 1.8s and everything else is negligible 16:56
I will laugh hard if I find there is some 20 second config value/timeout that is messing with things. 16:57
some misconfiguration on my end 17:00
AlexDaniel samebchase-: I did this more often than I'd like to admit 17:01
`sleep 60` … wait why is my program hanging? Ctrl-c
samebchase- 😂 17:02
I am trying to build a quick object dumper which I can then pipe to `jq` or `bat` for pretty-printing and/or further processing 17:03
writing lots of tiny scripts to optimise my workflow at $DAYJOB 17:05
17:07 Altai-man_ joined 17:09 sena_kun left
timotimo there's a cool module called Data::Dump::Tree, but i'm not sure if there's already a frontend that does filtering and such like you'd expect to be able to do with jq 17:10
17:22 softmoth joined 17:23 Xliff joined
Xliff Hello. 17:23
tellable6 2020-05-02T17:00:21Z #raku <AlexDaniel> Xliff: colabti.org/irclogger/irclogger_lo...05-02#l406
softmoth ,o/ 17:24
Xliff AlexDaniel: Tell him to file an issue. Audio::OggVorbis is on github!
.tell AlexDaniel Tell him to file an issue. Audio::OggVorbis is on github! 17:25
tellable6 Xliff, I'll pass your message to AlexDaniel
AlexDaniel .tell tellable6 can you file an issue?
tellable6 AlexDaniel, Thanks for the message
Xliff kicks Alexdaniel with a flipper.
AlexDaniel mmmm 17:26
Xliff AlexDaniel, I don't even know what you were talking about in a link.
AlexDaniel anyway: github.com/Xliff/p6-audio-oggvorbis/issues/2
Xliff Oh. :P
AlexDaniel `zef install Audio::OggVorbis` doesn't work, IIRC
Xliff So the problem is what? CPAN?
AlexDaniel it's not on cpan, so no 17:27
Xliff: maybe github.com/ugexe/Perl6-ecosystems ?
maybe something is wrong with your META6.json
Xliff Yeah. Fixed it. Tests passed. Pushed. 17:30
17:31 patrickb left
Xliff So... what's the fastest (or best) way to convert a Buf to a CArray[uint8]? 17:32
AlexDaniel ohh you have from<native> thingies
Xliff Currently I am using: "CArray[uint8].new( $buf.list )"
AlexDaniel: Thise test fine.
It was the '#' in front of them that were problematic.
AlexDaniel Xliff: yeah they're alright, it's just a bit unusual 17:33
Xliff s/Thiese/These/
AlexDaniel there are like… three modules in total doing this maybe?
17:33 xinming left
AlexDaniel maybe a bit more 17:33
Xliff Wow.
Skarsnik maybe there even stuff in NC itself, this module is a nightmare xD 17:34
17:35 xinming joined, Xliff left, chloekek left
lucs "Precedence issue with ! and :exists, perhaps you meant :!exists?" 17:36
I did! Raku's error messages are really something! raku++
18:05 andrzejku joined
softmoth r: say 'aaaa'.comb('aa'); # I expect two non-overlapping matches, but get 3 overlapping matches? 18:12
camelia (aa aa aa)
softmoth r: say 'aaaa'.comb(/aa/); # This works as I expect
camelia (aa aa)
18:13 brtastic joined
softmoth Is this intentional (just not documented under Str.comb)? Or am I right to be surprised? 18:14
18:16 natrys_ joined, stoned75 left 18:17 xinming left 18:18 xinming joined 18:19 natrys left
Geth doc: 56bd8dda1e | (JJ Merelo)++ | doc/Language/operators.pod6
Adds examples of ~ with Blobs, preparing for #2632
18:19
doc: a580e45cac | (JJ Merelo)++ | doc/Language/operators.pod6
1-arity ~ for blobs #2632

Also, same on Junctions...
linkable6 Link: docs.raku.org/language/operators
DOC#2632 [open]: github.com/Raku/doc/issues/2632 [Hacktoberfest][RFE][big][docs][good first issue][help wanted][new][⚠ Top Priority ⚠] Checklist for 6.d
18:28 sauvin left 18:33 chloekek joined 18:36 ragekagemage joined 18:38 stoned75 joined, ctilmes joined
ctilmes m: use NativeCall; sub foo(size_t $x) { say $x }; foo(27); 18:39
camelia 5===SORRY!5=== Error while compiling <tmp>
Failed to open file /home/camelia/.raku/precomp/CACHEDIR.TAG: Permission denied
at <tmp>:1
Actually thrown at:
in any statement_control at /home/camelia/rakudo-m-inst-2/share/perl6/lib/Perl6/Gra…
ctilmes m: use NativeCall; multi foo(size_t $x) { say $x }; foo(27);
camelia 5===SORRY!5=== Error while compiling <tmp>
Failed to open file /home/camelia/.raku/precomp/CACHEDIR.TAG: Permission denied
at <tmp>:1
Actually thrown at:
in any statement_control at /home/camelia/rakudo-m-inst-2/share/perl6/lib/Perl6/Gra…
ctilmes When I call a sub with a size_t arg, an Int goes into it fine, but with a multi, it barfs and says they don't match 18:40
how come?
[Coke] (can verify that locally, despite camelia here.) 18:43
18:45 andrzejku left
ragekagemage After compiling and installing the new 2020.05 release of MoarVM and NQP, building Rakudo still fails with the error "Stage start : moar://usr/local/lib/libmoar.so.11.0: undefined symbol 'mp_set_double' ld.so: moar: lazy binding failed!" This is one OpenBSD-current, using libtommath 1.2.0, libuv 1.35, and libatomic_ops 7.6.10 18:45
18:52 KindTwo joined 18:54 KindOne left
MasterDuke what if you don't use the system libtommath? 18:58
18:59 wamba joined
MasterDuke ragekagemage: we're using github.com/MoarVM/libtommath/commi...3a688ef274 on top of libtommath v1.2.0 19:00
Skarsnik NativeCall type are ... weird
19:02 KindOne joined 19:03 KindTwo left 19:05 xinming left 19:07 xinming joined 19:08 sena_kun joined, wamba left 19:09 Altai-man_ left 19:18 KindOne left 19:20 KindOne joined, lucasb joined, andrzejku joined 19:32 sjm_uk left 19:36 natrys_ left, xinming left, natrys joined 19:38 xinming joined
ragekagemage MasterDuke: That was the last thing blocking the build, it's working now! \o/ 19:40
19:41 pilne joined
MasterDuke glad to hear it 19:43
El_Che so, happy release day everyone! 19:48
19:48 Grrrr left
ragekagemage Now if only I could figure out how to build it without the --no-jit flag on MoarVM, when I remove it the build for NQP instantly crashes. All my code is running fast enough without it, that's a problem for another day. 19:51
MasterDuke ragekagemage: what hardware are you on? 19:55
El_Che This may be interesting to some (lizmat? jnthn?): github.com/rakudo/rakudo/issues/36...-624270590 <-- speed regression of new release
(I pointed him here, it looks he has his own speed stats) 19:57
ragekagemage MasterDuke: I'm on an AMD 3900X overclocked to 4.125ghz with 16gb of ram. I'm running OpenBSD-current 19:59
MasterDuke probably something with openbsd then, several devs have ryzens and don't have problems. does openbsd disallow self-modifying processes? 20:01
20:04 wtwt5237 joined
MasterDuke El_Che: hm, i can't repro. for me 2020.02 takes 8s and 2020.05 takes 6.1s 20:05
wtwt5237 hi, fellow raku-ers, I noticed the newest version of rakudo (2020.05) is slower than 2020.02: github.com/wtwt5237/raku-for-bioinformatics. Do you have any insight into this?
ragekagemage They have W^X as a security feature, any jit needs extra function calls to mark the memory regions as executeable and unwriteable. I've never had to mess with it but I know it's something that a lot of ports need to have patched before they can run. You can mark a partition as wxallowed to allow binaries to not be held to that requirement. I just haven't had time to mess with it
MasterDuke wtwt5237: i just mentioned this to El_Che, for me 2020.02 takes 8s and 2020.05 takes 6.1s 20:06
wtwt5237 interesting. so it is different for different tasks
MasterDuke ragekagemage: i think there might be an open issue in moarvm for that problem
wtwt5237: you haven't changed OS or hardware between runs? 20:08
wtwt5237 no, i don't think so. but I am running on university server. there could be changes behind the scenes
MasterDuke that's a really dramatic change you're seeing. how do you install raku? 20:10
timotimo hi tao, we just chatted on the github issue tracker 20:12
Doc_Holliwood the heck you mean private multimethods are not supported? i demand to talk to a manager! 20:14
20:16 rindolf left 20:22 jjmerelo left, melezhik joined
tbrowder El_Che: new rakudo-pkg worked fine on the upgrade on my Deb Buster box--thanks! 20:22
melezhik Xliff: FYI - I can see Audio::OggVorbis is successfully installed through rakudist - repo.westus.cloudapp.azure.com/spar...debian/284 20:23
tellable6 melezhik, I'll pass your message to Xliff
wtwt5237 MasterDuke, I really just installed it in the standard way 20:27
except that I found the nqp_configure module is missing, and manually downloaded it and put it into the install package
timotimo, hello again! 20:28
timotimo o/
20:28 stoned75 left
timotimo can you retry with the rakudo-2020.05.tar.gz? it should at the very least not complain about the nqp_configure submodule 20:29
MasterDuke wtwt5237: or maybe you accidentally ran against test3.sam? 20:34
wtwt5237 no. it is test2.sam. I double checked
timotimo which benchmark are we talking about and how do i properly run it? 20:37
wtwt5237 please check: github.com/wtwt5237/raku-for-bioin...C%20sister
test2.sam
El_Che tbrowder: glad to hear 20:38
wtwt5237: I am the other person on the ticket thread
wtwt5237 hello! 20:39
timotimo benchmark1.p6 or benchmark2.p6?
wtwt5237 benchmark2.p6 20:40
timotimo OK
that takes 15 seconds on my machine, huh?
wtwt5237 speed is very similar. 8s 20:41
MasterDuke same user time?
wtwt5237 real 0m8.556s 20:42
20:42 stoned75 joined
timotimo Total Time: 19.69 seconds 20:43
Total CPU Usage: 38.49 seconds
MasterDuke i get wildly varying times in runs immediately after each other. 7.5s, 10.0s, 5.8s 20:44
timotimo OK 20:45
the user time is much more stable
the cpu% value fluctuates noticeably 20:46
so mostly a "system is scheduling stuff differently" issue perhaps
MasterDuke yes. 21s, 20s, 23s user
timotimo about 35 over here :|
MasterDuke timotimo: but notice the last line in the table? github.com/wtwt5237/raku-for-bioin...-on-sister 20:47
gotta get yourself a ryzen
20:47 stoned75 left
timotimo wow can the user time there really be accurate? that's wild 20:47
20:48 stoned75 joined
timotimo like, if the real time only went up 2 seconds, the cpu core usage must have increased a few hundred percent? 20:48
wtwt5237 I got to leave the conversation now. if you are interested in contacting me for any other details, please email me at [email@hidden.address] 20:49
20:49 wtwt5237 left
timotimo the output of --benchmark=blah.html will be only a little part of everything with a .race.map 20:49
--profile* 20:50
20:50 kst joined
Geth doc/hashmap-perl-to-raku: 82c50100bc | (Stoned Elipot)++ | doc/Language/hashmap.pod6
Change .perl to .raku, ref #3309
20:52
doc: stoned++ created pull request #3362:
Change .perl to .raku, ref #3309
linkable6 DOC#3309 [open]: github.com/Raku/doc/issues/3309 [docs][update] Change .perl to .raku in examples
20:55 chloekek left
timotimo does Digest::SHA1::Native miss a dependency on LibraryMake? 20:56
20:57 Redfoxmoon left 20:59 Kaiepi left, Kaeipi joined, Kaeipi left 21:00 Kaeipi joined
melezhik timotimo: Digest::SHA1::Native succeeds in RakuDist - repo.westus.cloudapp.azure.com/spar...debian/286 21:02
dependency gets installed - Testing [OK] for LibraryMake:ver<1.0.0>:auth<github:retupmoca> 21:03
21:03 brtastic left 21:07 Altai-man_ joined 21:09 sena_kun left 21:14 Altai-man_ left, softmoth_ joined 21:16 softmoth__ joined 21:17 softmoth left 21:19 softmoth_ left
timotimo outdated stuff in my zef store has apparently caused trouble 21:26
21:27 softmoth__ left, softmoth__ joined
MasterDuke nuke it from orbit, it's the only way to be sure 21:27
timotimo yup, that's what i did
tungsten rod from geostationary orbit
melezhik ;-) 21:29
you can always use rakudist to check such a things 21:30
timotimo i didn't see a search function there, do i just have to scroll through "recent builds"? 21:31
melezhik I'll add a search function soon, so far it's just a `curl -d os=debian repo.westus.cloudapp.azure.com/raku...1::Native` 21:32
it will queue a new build and you'll get results soon
I am going to add more convenient web UI soon, so that people could trigger builds from html form, rather then from `curl` 21:33
21:33 softmoth joined
melezhik and yeah - recent builds are 500 last results - so you can search through a browser search 21:34
timotimo OK
21:35 softmoth__ left
timotimo aaaaaanyway, i'm looking through the profiler results from the bioperl benchmark 21:36
21:36 softmoth_ joined
MasterDuke seeing anything interesting? 21:37
timotimo not so sure 21:38
21:38 softmoth left
timotimo i'll probably toss out the race multitheading for a moment so the profile gets a lot simpler 21:40
21:42 softmoth_ left
MasterDuke timotimo: btw, did you git pull the repo recently? a micro-optimization was just merged ~40min ago 21:45
timotimo i think i got it after that, but let me look again 21:46
ah, that's a good one 21:47
goes even faster if you're sure 64bit is enough to hold the sum, the nyou can .sum(:wrap) and it'll use native ints
another thing i've done locally is to use comb instead of match :g and use )> <-[I]> instead of <!before I> 21:48
though i'm not sure if that particular last change makes such a big difference 21:49
MasterDuke i don't know the data well enough to know if :wrap is safe, but i'd forgotten about that, good to get a reminder
21:57 thundergnat joined
Geth doc: b2a8ece763 | Coke++ | xt/examples-compilation.t
add ref to ticket
22:00
thundergnat Is anyone else having trouble installing Inline::Perl5:ver<0.48> on Raku 2020.05? I'm not having much success with it. Version 0.47 installs fine but .48 keeps complaining about undefined symbols. 22:01
Fail log: gist.github.com/thundergnat/68ff93...15af6f9171 22:02
22:04 andrzejku left
timotimo the regex matches for the substr lines take about 5 seconds in total 22:06
out of about 30 seconds profiled run time
Match is responsible for a whole lot of allocations (by size * count) 22:07
3 million allocations vs Scalar's 5.7 mil, but 442 megs vs 275 megs 22:08
MasterDuke matching seems expensive by any metric you choose 22:10
timotimo yeah
tearing the first match (well, subst really) apart into stuff like index and substr and such could be relatively easy, i.e. finding the first non-\d char and seeing if it's an S and then just substr-ing the number out and use that 22:11
22:12 Redfoxmoon joined
timotimo also interesting to see that match and sum take the same amount of time per entry (though match is called twice as often because there's two ~~ in there) 22:14
MasterDuke really? why is sum so expensive?
timotimo not sure 22:15
also interesting, the regexes are on line 33 and 31, right?
22:16 KotH left
timotimo the one on 33 takes 2.183s and the one on 31 takes about 1.421s 22:16
MasterDuke i checked the number of numbers found in the match, 208215 times it was just one, 1491 times it was zero, 189 times it was 3, 84 times it was 2, and 21 times it was 6 22:17
huh. adding the end of line anchor slows it down? 22:18
timotimo how much is in the $!CIGAR field? it uses a $, i think we're not yet good at going to the end and starting backwards
it's possible the regex engine is scanning from the beginning every time
how often does it match vs not matching?
MasterDuke what do you mean? 22:19
Geth doc: 82c50100bc | (Stoned Elipot)++ | doc/Language/hashmap.pod6
Change .perl to .raku, ref #3309
doc: 9bcce59e58 | (Will Coleda)++ (committed using GitHub Web editor) | doc/Language/hashmap.pod6
Merge pull request #3362 from Raku/hashmap-perl-to-raku

Change .perl to .raku, ref #3309
linkable6 Link: docs.raku.org/language/hashmap
DOC#3309 [open]: github.com/Raku/doc/issues/3309 [docs][update] Change .perl to .raku in examples
MasterDuke heh, what you if flip the string and the regexes?
22:20 stoned75 left
MasterDuke oh, don't starts-with and ends-with now take regexes? 22:20
timotimo they do i believe
but just a ends-with with the S could be a good start if that has a rejection rate of more than 10%
1078 / 110000 22:21
that is a very, very high rejection rate
hum, but .ends-with("S") doesn't seem to give True, like, ever? what did i do wrong now 22:23
oh haha
removed the S at the end before checking if it exists
Geth doc: f39a270390 | sztanyi++ (committed using GitHub Web editor) | doc/Language/numerics.pod6
change perl method to raku method

change 1e0.perl to 1e0.raku
doc: 907dab8660 | (Will Coleda)++ (committed using GitHub Web editor) | doc/Language/numerics.pod6
Merge pull request #3282 from sztanyi/patch-2

change perl method to raku method
linkable6 Link: docs.raku.org/language/numerics
timotimo the rejection rate of "ends with an S?" seems to be 1:1 for the whole regex 22:24
that's good
22:25 KotH joined
timotimo it does seem a bit faster to check for an S at the end first 22:25
melezhik thundergnat: Inline::Perl5:ver<0.48> on Raku 2020.05 seems fine on RakuDist - repo.westus.cloudapp.azure.com/spar...debian/287
thundergnat Yep. It does seem that way, but it fails to install for me on two different Linux Mint systems. 22:26
melezhik you can always check by just ` curl -d os=debian -d rakudo_version=40a82d8723470a58d2d0cdde4cb14f8d0e845c91 repo.westus.cloudapp.azure.com/raku...ne::Perl5` 22:27
thundergnat with the same failure mode on each
melezhik it's possible to run test against ubuntu
thundergnat And it's very odd because 0.47 installs fine on both systems but 0.48 fails. 22:28
timotimo i did it wrong 22:29
22:29 lucasb left
melezhik thundergnat: I've just run it on ubuntu. let's see - repo.westus.cloudapp.azure.com/sparky/builds 22:30
timotimo i had the match line with the .sum in there twice because i unstashed my changes and forgot to delete the original new line
22:31 natrys left
MasterDuke ha 22:32
timotimo yeah it's faster now in total
gist.github.com/timo/4fecd3851edca...b3cf7c2ef1 my current diff
looks like a .contains for "S" can also reject a whole bunch of lines 22:33
that's worth rather a bit 22:34
cdn.discordapp.com/attachments/633...nknown.png 22:35
^- check out the icicle graph
TWEAK used to be about 3/4 of the width or something 22:36
MasterDuke nice 22:38
timotimo the $!CIGAR strings are all short-ish, so no need to limit how far into the string to search for the existence of any S 22:45
BagHash(1(781) 10(11) 11(11) 19(11) 3(107052) 5(990) 6(814) 7(88) 8(187) 9(55))
Geth doc: 68237e2e92 | Coke++ | 7 files
Use .raku instead of .perl in examples

Part of #3309
Update usage and some output diagnostics which themselves have been updated
22:47
linkable6 DOC#3309 [open]: github.com/Raku/doc/issues/3309 [docs][update] Change .perl to .raku in examples 22:48
melezhik thundergnat: if it helps - Inline::Perl5 succeed for Rakudo version 2020.05, on ubuntu - repo.westus.cloudapp.azure.com/spar...ubuntu/289 22:51
thundergnat melezhik: That's encouraging, though it doesn't really help on _my_ system(s). Thanks for checking though. 22:53
melezhik you are welcome 22:54
thundergnat It isn't the end of the world, I can run .47, it is just a little disconcerting.
22:56 Skarsnik left 23:02 pecastro left 23:03 xinming left, xinming joined
ctilmes m: sub foo(int32 $x) { say $x }; foo(27); 23:06
camelia 27