Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6-dev, 2016-06-19

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:28 AlexDaniel joined #perl6-dev
00:42 Zoffix If I add RAKUDO_MODULE_DEBUG I get a different error :/   Failed to find 'sources/51E302443A2C8FF185ABC10CA1E5520EFEE885A1' while trying to do '.modified'  https://gist.github.com/zoffixznet/b5c85c9c6c36196e416fa3a7a38f104e
00:46 Zoffix What the ... This is what's found in the DESTDIR after the error. Notice how it creates paths inside that seem to be made up out of bits of the path where I ran make install :S https://gist.github.com/zoffixznet/5504f760b0a99fb78d060280ea9e7b26
00:48 Zoffix Is 'DESTDIR' even a thing? Shouldn't that be PREFIX?
00:51 Zoffix nm
01:27 FROGGS_ joined #perl6-dev
01:48 ilbot3 joined #perl6-dev
01:48 Topic for #perl6-dev is now Perl 6 language and compiler development 2.0 | Logs at http://irclog.perlgeek.de/perl6-dev/today
02:07 b2gills m: Nil for (1..10⁴⁰).race.map({$/=[.polymod(10 xx *)];$_==sum @$/ X** +$/??$_!!|()}) # I would love to know why this has such a weird error (more than just race being buggy)
02:07 camelia rakudo-moar f19241: OUTPUT«(signal XCPU)Method 'fully-reified' not found for invocant of class 'Mu'␤  in block  at <tmp> line 1␤␤Method 'fully-reified' not found for invocant of class 'Mu'␤  in block  at <tmp> line 1␤␤Method 'fully-reified' not found for invocant of clas…»
07:53 lizmat joined #perl6-dev
07:58 [Tux] This is Rakudo version 2016.05-179-gf192415 built on MoarVM version 2016.05-37-ga126e0f
07:58 [Tux] test            16.516
07:58 [Tux] test-t           9.365
07:58 [Tux] csv-parser      23.410
08:01 [Tux] http://tux.nl/Talks/CSV6/images/test-t-22.png I think we can be reasonable happy
08:05 [TuxCM] joined #perl6-dev
08:20 [Tux] joined #perl6-dev
08:31 nine Good morning!
08:31 yoleaux2 18 Jun 2016 23:33Z <Zoffix> nine: would you have any idea how to fix this issue? It happens when DESTDIR is specified for make install and it's blocking the release: http://irclog.perlgeek.de/perl6-dev/2016-06-18#i_12690874
08:32 nine Of course I cannot reproduce that here :/
09:25 RabidGravy joined #perl6-dev
10:00 sno joined #perl6-dev
10:35 FROGGS o/
10:35 Zoffix nine, even when you're using DESTDIR?
10:35 nine Ok, I can reproduce it now when trying to create packages for openSUSE. The difference may already give a hint to what's going wrong.
10:36 Zoffix make install works for me just fine, but make DESTDIR='wblahg' install fails. Where dest dir is absolute version for ../perl
10:37 nine Zoffix: I've tried make DESTDIR/tmp/try1 install from my development directory and it went just fine. I even reconfigured with --prefix and it succeeded. I can only reproduce it with rpmbuild
10:38 Zoffix damn
10:38 nine 27  9359 RMD: dependency: 51E302443A2C8FF185ABC10CA1E5520EFEE885A1 sources/51E302443A2C8FF185ABC10CA1E5520EFEE885A1 08B249F191B4AD191965BCA6D327CC1EEA6D4DD5 NativeCall::Types
10:39 nine The relative sources path is a bit strange. It should either be prefixed by a repo name (like perl#/sources/...) or be an absolute path.
11:04 nine Good news is that I may have a patch that fixes this. Bad news is that NativeCall tests fail when I build the rakudo package
11:07 dalek rakudo/nom: 53af022 | niner++ | src/core/CompUnit/Repository/Installation.pm:
11:07 dalek rakudo/nom: Fix rakudo not finding the souce of a dependency in a non-standard repo
11:07 dalek rakudo/nom:
11:07 dalek rakudo/nom: Paths to souce files must either be relative to a named repository as
11:07 dalek rakudo/nom: in: perl6#sources/123456789ABCDEF or absolute as in /tmp/foo/1234567...
11:07 dalek rakudo/nom:
11:07 dalek rakudo/nom: Should fix issues with make DESTDIR=/foo/bar install
11:07 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/53af022577
11:35 AlexDaniel joined #perl6-dev
11:36 AlexDaniel [Tux]: perhaps it's time to plot this data on a logarithmic scale :)
11:58 brrt joined #perl6-dev
13:09 lizmat joined #perl6-dev
13:15 * lizmat waves from http://www.yapcna.org/yn2016/event/1858
13:22 lizmat m: for ^10 { say $^_ }   # wonder if this should be an error, or just silently assume $_
13:22 camelia rakudo-moar 53af02: OUTPUT«===SORRY!=== Error while compiling <tmp>␤Redeclaration of symbol $^_ as a placeholder parameter␤at <tmp>:1␤------> for ^10 { say $^_⏏ }   # wonder if this should be an error␤»
13:34 MasterDuke joined #perl6-dev
13:35 MasterDuke re http://irclog.perlgeek.de/perl6/2016-06-19#i_12692803, the link points to http://rakudo.org/downloads/star/rakudo-star-2016.04.dmg, but the actual filename is rakudo-star-2016-04.dmg
13:36 MasterDuke the filename seems to be wrong, all the other filenames (older versions, different types) in that directory have <year>.<month>
13:52 MasterDuke switching topics entirely, why is 'my @l = <path/to/file>.IO.lines' in perl 6 so much slower than 'open(my $F, "<", <path/to/file>); my @l = <$F>; chomp @l' in perl 5?
13:53 lizmat MasterDuke: we ensure it's valid UTF-8 for one
13:54 stmuk MasterDuke: I'm just fixed that
13:54 stmuk I need new glasses :(
13:54 mst MasterDuke: the general answer to such questions will tend to be
13:55 mst MasterDuke: (1) because perl6 is doing something more clever with a good reason, and there'll be a low level thing you can call to get something more like the perl5 behaviour
13:55 mst MasterDuke: (2) because in the run up to freezing for christmas, perl6 focused on optimizability over optimizations, since the former need to be baked into the feature set and then the latter can be added over time
13:58 MasterDuke looking at a profile, all the time seems to be in IO::Handle.get, which is pretty simple
13:58 MasterDuke and would lead me to believe any improvement in speed would have to come at the MoarVM level?
14:00 mst when I say perl6 I mean rakudo/nqp/MoarVM
14:00 mst somebody with more clue can give you a more detailed answer
14:02 Zoffix It's weird make DESTDIR='blahg' doesn
14:02 MasterDuke fyi, i'm not intending to criticize (i realize perl 6 hasn't had the same # of years to optimize), just find out if it's something i could attemp to work on
14:02 Zoffix It's weird make DESTDIR='blahg' doesn't install moar too.
14:03 MasterDuke and further down the rakudo->nqp->moarvm chain the less i'm able to do so
14:13 mst MasterDuke: yar, I was more trying to set expectations than anything
14:13 mst also it's entirely possible that digging down with profilers even if you can't make the optimisation would help somebody else
14:16 lizmat joined #perl6-dev
14:16 MasterDuke i must say, i absolutely love that a profiler is built in
14:16 Zoffix All this stresstesting makes me wish I had a meatier box :(
14:16 MasterDuke i find myself profiling everything
14:16 * Zoffix drolls over http://www.tigerdirect.ca/applications/category/guidedSearch.asp?CatId=22&amp;sel=Detail%3B112_1258_125488_125488
14:17 Zoffix drools too
14:18 MasterDuke same here. i'm starting to find 8gb ram a bit limiting
14:19 Zoffix I have 16GB and I find it limiting ATM (trying to run two VMs at the same time)
14:19 MasterDuke i use some 40 core, 378gb ram machines at $dayjob, would love to have one at home
14:19 Zoffix :o
14:21 Zoffix For CAD$1,625.90 I can build a 6-core (12 threads) + 128GB RAM box. Which isn't too bad
14:24 MasterDuke i'm torn. i want to upgrade my cpu/ram/mb, i want a high dpi monitor, i want an htc vive, i want a new snowboard
14:24 AlexDaniel 128 GB RAM hmmm
14:24 MasterDuke oh, also video card
14:24 AlexDaniel sounds great
14:24 MasterDuke i can do one, maybe two of those
14:24 MasterDuke but which!?
14:25 AlexDaniel you don't really need a drive with such setup :)
14:25 mst at the point where I'm planning to do lots of CPU heavy stuff I think I'll spin up temporary cloud instances
14:25 mst I'd only build a bigass home box if I needed the I/O
14:25 Zoffix m: say 1.64 * 48
14:25 camelia rakudo-moar 53af02: OUTPUT«78.72␤»
14:26 Zoffix m: say 1.44 * 24
14:26 camelia rakudo-moar 53af02: OUTPUT«34.56␤»
14:26 Zoffix 20-core Linod box for a day
14:27 AlexDaniel Zoffix: some weeks ago I've switched from 3GB to 6GB RAM :) it's kinda nice that I don't have to unswap things when stuff goes wrong
14:28 AlexDaniel the reason is that I usually run CPU/memory intensive stuff on my server (where bisectable is running)
14:28 Zoffix t/spec/S02-lists/indexing.t ................................... ok
14:28 Zoffix ===(     185;4  10/23   6/50  0/?  0/?  0/? )===========================*** Error in `/home/zoffix/CPANPRC/rakudo/install/bin/moar': double free or corruption (fasttop): 0x00007f744801ec90 ***
14:28 Zoffix :o
14:29 Zoffix oh, never mind. I can't read... it was actually from t/spec/S17-lowlevel/lock.t .................................... Failed 14/23 subtests
14:29 mst Zoffix: GCE 32 vCPU, 120Gb RAM seems to be $5.50 per day
14:29 Zoffix Where?
14:29 Zoffix What's GCE
14:30 mst google compute engine
14:30 Zoffix Cool.. Will look into that if I'm ever releasing rakudo again :)
14:31 mst at some point I'm going to rig something up that lets me spin one of those up for build work
14:31 AlexDaniel not a 12-core but if you want to shell out some money perhaps it makes sense to get a better dedicated server, hm
14:58 b2gills There is also Digital Ocean for virtual servers
14:58 mst yeah, DO are most cool too
15:17 CQ joined #perl6-dev
15:29 Woodi how vCPUs compare to CPUs ? :)
15:33 moritz they are more elastic :-)
15:38 mst depends on whether you have noisy neighbours or not, generally
15:40 moritz I guess we are the noisy neighbours around here :-)
15:42 Woodi btw. do you compile things on disk or in ram ?
15:43 moritz on disk
15:43 moritz but the buffer cache means it stays in ram at first
15:44 moritz well, it's compiled in ram, then written to disk, but the OS caches the disk write in RAM
15:44 moritz full circle
15:49 nine Zoffix: did my patch fix the DESTDIR issue for you?
15:50 Zoffix nine, yup
15:50 Zoffix nine++
15:51 Zoffix Do you mean an rpm package? <nine> [...] Bad news is that NativeCall tests fail when I build the rakudo package
15:53 nine Zoffix: yes
15:54 Zoffix I'll mention that possibility in the announcement, but I think we can still release.
15:54 * Zoffix is running the last stresstest on unpacked tar
15:56 nine Zoffix: note that while make test failed during rpmbuild, I can use the installed rpm to install and run panda and Inline::Perl5
15:57 Zoffix Sweet
16:02 lizmat lunch&
16:07 skids joined #perl6-dev
16:52 Zoffix I've retested and rebuilt the release package with nine++'s fix. Release PR: https://github.com/rakudo/rakudo/pull/793 tars: http://temp.perl6.party/2016.06-release/
16:52 Zoffix After the PR is merged, someone with rakudo repo access needs to tag the release
16:53 Zoffix and the tars need to be uploaded to rakudo.org
17:19 dalek Heuristic branch merge: pushed 18 commits to rakudo/nom by lizmat
17:19 lizmat Zoffix: done
17:19 Zoffix Sweet. lizmat++
17:20 lizmat Zoffix: fwiw, I can't upload tarballs either, jnthn moritz FROGGS can afaik
17:20 Zoffix Someone needs to run this too:     git tag -u <your-email> -s -a -m "tag release #100" 2016.16; git push --tags
17:20 Zoffix errr
17:20 Zoffix Someone needs to run this too:     git tag -u <your-email> -s -a -m "tag release #100" 2016.06; git push --tags
17:20 Zoffix :)
17:21 lizmat my email, or yours ?
17:21 japhb Personally, I think 16 months would be great, if all 4 of the new ones were vacation ....
17:21 Zoffix lizmat, yours :)
17:22 lizmat well, then I don't need to specify that, do I  ?
17:22 Zoffix Oh, I guess not
17:22 * Zoffix just copy-pasted from the release guide
17:22 lizmat argh, I don't have gpg
17:22 lizmat so someone else needs to do it...
17:22 lizmat I'm in Damian's course  :-)
17:23 lizmat [Coke] : you around ?
17:23 Zoffix I will see him in 8 days :)
17:28 FROGGS I'm here
17:29 lizmat FROGGS: could you tag the release ?
17:29 FROGGS just nom HEAD?
17:29 Zoffix Sweet. Are you able to tag the release and upload the tarballs+keys? http://temp.perl6.party/2016.06-release/
17:29 Zoffix Yeah
17:31 FROGGS tagged
17:31 lizmat FROGGS++
17:31 lizmat Zoffix: just FYI, spectest clean for me
17:33 Zoffix Thanks
17:34 FROGGS nqp uploaded
17:36 FROGGS rakudo uploaded
17:36 Zoffix FROGGS++ thanks a lot!
17:37 FROGGS pleasure :o)
17:45 FROGGS Zoffix++
17:47 dalek rakudo/nom: 1d93adf | lizmat++ | src/core/control.pm:
17:47 dalek rakudo/nom: Make samewith() also work on none-multi's
17:47 dalek rakudo/nom:
17:47 dalek rakudo/nom: Inspired by TheDamian's functional bogosort example from the Perl 6
17:47 dalek rakudo/nom: Workshop at The Perl Conference
17:47 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/1d93adf2f7
17:47 Zoffix And that's it. Finished last step of the release
17:47 * Zoffix celebrates with the appropriate amount of fun by murdering a bunch of people in Far Cry 4
17:48 FROGGS hehe
17:56 lizmat Zoffix++
17:58 lizmat fwiw, the bogosort example:
17:59 lizmat sub bogosort(*@list) { [!after] @list ?? @list !! samewith @list.pick(*) }
17:59 lizmat the example needed a multi before
18:04 Zoffix What's [!after] ?
18:04 lizmat m: say 42 !after 666
18:04 camelia rakudo-moar 1d93ad: OUTPUT«True␤»
18:05 lizmat m: say [!after] 42,666
18:05 camelia rakudo-moar 1d93ad: OUTPUT«True␤»
18:05 Zoffix Ah, now I get it
18:05 Zoffix Thanks
18:30 timotimo will samewith give us many callframes in a stack?
18:31 lizmat yeah, for now, it just adds callframes
18:31 lizmat in the future, hopefully not  :-)
18:31 timotimo yay
18:41 vendethiel- joined #perl6-dev
18:44 ilmari_ joined #perl6-dev
18:44 jdv79_ joined #perl6-dev
18:44 DrForr_ joined #perl6-dev
18:44 jnthn With the caveat that someone's gotta design and write the logic to check it's not needed :-)
18:45 lizmat "in the future"  :-)
18:46 moritz Zoffix: if you /msg me a ssh pubkey, I can add it to rakudo.org for you
18:48 Zoffix moritz, last one in the list: https://github.com/zoffixznet.keys
18:50 moritz Zoffix: you should now be able to ssh to rakudo@rakudo.org
18:50 Zoffix Works. Thanks, moritz++
18:52 jnthn lizmat: btw, getting the return changes ported to JVM took more time than expected, so didn't get to look into nameds and multi-caching just yet.
18:52 jnthn Plan to look in the next few days, as well as start looking into hyper/race
18:55 lizmat jnthn: cool!  :-)
18:55 [Tux] joined #perl6-dev
19:17 Zoffix m: sub postfix:<⁰º> { say $^a  }; say 2⁰º¹⁰
19:17 camelia rakudo-moar 1d93ad: OUTPUT«2␤1␤»
19:22 Zoffix oh, two says... I thought there were a bug
19:25 Zoffix m: sub infix:<+> { [$^a, $^b]  }; say 2+⁹¹⁰
19:25 camelia rakudo-moar 1d93ad: OUTPUT«[2 3486784401]␤»
19:25 Zoffix m: sub infix:<+> { [$^a, $^b]  }; say 2+¹⁰
19:25 camelia rakudo-moar 1d93ad: OUTPUT«[2 1]␤»
19:25 Zoffix m: sub infix:<+> { [$^a, $^b]  }; say 2+⁰
19:25 camelia rakudo-moar 1d93ad: OUTPUT«[2 0]␤»
19:25 Zoffix :S
19:25 Zoffix m: sub infix:<+> { [$^a, $^b]  }; say 2+⁹⁹¹⁰
19:25 camelia rakudo-moar 1d93ad: OUTPUT«[2 2294476371992180765796336392059464475488556595672096606348636582774505268492131019873809154452049089072384430715158203166898748858912919529376258202366553002451202043854328895833407639020439342007502942577141701582582003616314512684556225471683197867920…»
19:26 Zoffix m: say ⁹⁹¹⁰
19:26 camelia rakudo-moar 1d93ad: OUTPUT«2294476371992180765796336392059464475488556595672096606348636582774505268492131019873809154452049089072384430715158203166898748858912919529376258202366553002451202043854328895833407639020439342007502942577141701582582003616314512684556225471683197867920589…»
19:26 Zoffix m: say 9 ** 910
19:26 camelia rakudo-moar 1d93ad: OUTPUT«2294476371992180765796336392059464475488556595672096606348636582774505268492131019873809154452049089072384430715158203166898748858912919529376258202366553002451202043854328895833407639020439342007502942577141701582582003616314512684556225471683197867920589…»
19:26 Zoffix interesting
19:58 ilmari_ m: say ² # root of the problem
19:58 camelia rakudo-moar 1d93ad: OUTPUT«2␤»
20:20 dalek star: 02fd148 | (Steve Mynott)++ | ports/darwin_dmg/package_darwin_dmg.pl:
20:20 dalek star: correct final DMG name
20:20 dalek star: review: https://github.com/rakudo/star/commit/02fd148beb
20:24 Zoffix IMO it'd make more sense for ⁹⁹¹⁰ == 9910 than 9 ** 910
20:25 lizmat m: say ²⁹
20:25 camelia rakudo-moar 1d93ad: OUTPUT«512␤»
20:25 lizmat yeah, not making sense
20:27 Zoffix or die
20:27 lizmat yeah, should probably be a compile time fail
20:28 AlexDaniel that's an old issue
20:29 AlexDaniel #126732
20:29 synopsebot6 Link:  https://rt.perl.org/rt3//Public/Bug/Display.html?id=126732
20:29 AlexDaniel was rakudobugged the same day superscripts were implemented :)
20:53 b2gills Dangit ... of course I had to run the bogosort example that lizmat posted, and then forget about it until after my computer started to get reeeaally slow
20:57 lizmat hehe...
21:17 lizmat joined #perl6-dev
21:21 cognominal joined #perl6-dev
21:22 cog-nominal joined #perl6-dev
22:04 cog-nominal joined #perl6-dev
23:04 skids joined #perl6-dev

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary