| ugexe | i have a script that launches a bunch of processes in sequence, totalling 800ish. 2017.06 keeps dying at 255 though... | 00:56 | |
| does 255 correspond to something like a handle limit or something? | 00:57 | ||
| timotimo | what does strace tell you? | 00:58 | |
| is there any text to the exceptions you get? | |||
| ugexe | i dont know how to use strace. but it just stops - no message, but always at 255 | 01:00 | |
| well, 254 | 01:01 | ||
| timotimo | you can strace -f textfile.txt perl6 foo bar | 01:02 | |
| then you can let me have a quick look | |||
| so if it "just stops", does it hang, or terminate? | 01:04 | ||
| geekosaur | os x by any chance? | 01:05 | |
| oh, changed in 2017.06, n/m | 01:06 | ||
| ugexe | strace gives me permissions errors when i try to use it with this gist.github.com/ugexe/4ec8a4258819...58b192c1b4 | 01:10 | |
| it terminates with no message. | 01:11 | ||
| geekosaur | -f does not tell it a file to save tyo | 01:13 | |
| timotimo | it's trying to execve a .txt file | ||
| oh, sorry | |||
| geekosaur | -f means "follow forks" | ||
| timotimo | right, it must be -o | ||
| my bad, i'm tired | |||
| geekosaur | so your intended log file is actually the program to run | ||
| ugexe | with strace it says Segmentation Fault at least | 01:18 | |
| samcv | ok i need to adjust the Unicode collation bitfield items for the nqp::unicmp_s op. to allow you to configure reverse sorting of things | ||
| timotimo | that's interesting. maybe with perl6-gdb-m we can get more info out of it? | ||
| ugexe | i'm copying over the log, but its like 80mb | ||
| samcv | like for uppercase to come before lowercase and things like that. since lowercase comes first by default in UCA | ||
| so i'm thinking still a bitfield, but primary+ is 1 primary- is 2; secondary+ is 4, secondary- is 8 | 01:19 | ||
| added to the bitfiled values. because i need to designate *no* sorting and designate normal direction and also designate reversed sorting | |||
| timotimo, does that seem sane? | 01:20 | ||
| ugexe | Program received signal SIGSEGV, Segmentation fault. | ||
| 0x00007ffff7833b57 in gc_mark () from //home/nickl/.rakudobrew/moar-2017.06/install/lib/libmoar.so | |||
| samcv | i've seen that | ||
| ugexe | thats what perl6-gdb-m shows | 01:21 | |
| samcv | yep | ||
| maybe the same one i got yseterday | |||
| building without debugging made it go away | |||
| timotimo | oh, you probably didn't build with --debug set to anything | ||
| ugexe | mine is somehow tied to the number 255 | ||
| samcv | gist.github.com/samcv/9f7d8820cb9e...9aaf88b6e2 | ||
| ugexe | this is the last few hundred lines of the strace gist.github.com/ugexe/0d38dcd3d3e6...b1cf1f5a8a | 01:24 | |
| timotimo | i expect the multiple "killed by" is one per thread | 01:30 | |
| ugexe | its not exactly multi threaded... but it does something like `map { run ... }, @bar` | 01:32 | |
| timotimo | hmm, run is now based on proc::async, isn't it? | 01:34 | |
| ugexe | github.com/ugexe/Perl6-App--ecogen...en.pm6#L40 specifically here | ||
| right | |||
| timotimo | sorry, i'm too tired to think straight | 01:35 | |
| do i need to grab anything more to run that script? | |||
| it doesn't have a main by itself, is there a bin/ with a script in it? | |||
| ah, the readme tells it all | |||
| ugexe | yeah. no deps, will want to do `perl6 -Ilib bin/ecogen p6c cpan` | 01:36 | |
| Geth | nqp: 3ff261cb41 | (Samantha McVey)++ | tools/build/MOAR_REVISION [unicode-grant] Bump MoarVM to get unicmp_s improvements and changes The collation_mode argument has changed semantics and has added features. It now allows reverse sorting of any of the collation levels. See this MoarVM commit for full details: github.com/MoarVM/MoarVM/commit/b34c3748 |
05:57 | |
| ¦ nqp: version bump brought these changes: github.com/MoarVM/MoarVM/compare/2...-gcc27eebf | |||
| [Tux] | This is Rakudo version 2017.06-9-g6d4691fbd built on MoarVM version 2017.06 | 06:07 | |
| csv-ip5xs 2.537 | |||
| test 12.925 | |||
| test-t 4.069 - 4.192 | |||
| csv-parser 12.586 | |||
| 88.t still fails sometimes: | 06:10 | ||
| Error encoding UTF-8 string: could not encode codepoint 1309219 (0x13FA23), codepoint out of bounds. Cannot encode higher than 1114111 (0x10FFFF) | 06:11 | ||
| in block <unit> at 88.t line 21 | |||
| samcv | [Tux], is it failing any less than usual? | 06:13 | |
| [Tux] | yes, but as it is based on random data, you'll never know :) | 06:18 | |
| samcv | true :) | 06:19 | |
| Geth | rakudo/nom: 4da85879a7 | (Samantha McVey)++ | src/core/Collation.pm [unicode-grant] Collation object updated to work with new unicmp_s semantics The Collation object now allows setting reverse sort order of different collation levels. This is backwards compatible as long as the user has not manually set the collation-level themselves. This was strongly warned against relying on for this experimental API. Setting the collation levels the recommended way still yields the same result. In addition you can now set the collation levels to negative values to reverse the sorting order for just that collation level. |
06:47 | |
| rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....g3ff261cb4 2ee5fd4125 | (Samantha McVey)++ | src/core/Str.pm nqp::unicmp_s(a, b, 85, 0, 0) results in the same functionality as nqp::unicmp_s(a, b, 15, 0, 0) did before as of MoarVM commit b34c3748 |
|||
| rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....g3ff261cb4 476780772f | (Samantha McVey)++ | tools/build/NQP_REVISION |
06:48 | ||
| ¦ rakudo/nom: version bump brought these changes: github.com/perl6/nqp/compare/2017....g3ff261cb4 | |||
| pmurias | what's the utf8-c encoding? | 07:53 | |
| found that | 08:15 | ||
| jnthn | morning, #perl6 | 08:46 | |
| masak | morning, jnthn | 08:47 | |
| samcv | pmurias, did you see my recent addition to the docs? | 08:51 | |
| morning jnthn | |||
| pmurias | samcv: yes, samcv++ | 08:57 | |
| found it in the MoarVM source comments first ;) | |||
| samcv | hehe | 09:07 | |
| lizmat | Files=1204, Tests=62523, 221 wallclock secs (13.45 usr 4.99 sys + 1342.21 cusr 137.32 csys = 1497.97 CPU) | 09:15 | |
| jnthn | m: sub foo($?) { ^2 .map: &say }; 1.map({ foo }).sink | 09:16 | |
| camelia | ( no output ) | ||
| jnthn | c: 2017.05 sub foo($?) { ^2 .map: &say }; 1.map({ foo }).sink | ||
| committable6 | jnthn, ¦2017.05: «» | ||
| jnthn | c: 2015.12 sub foo($?) { ^2 .map: &say }; 1.map({ foo }).sink | ||
| committable6 | jnthn, ¦2015.12: «» | ||
| jnthn | c: all sub foo($?) { ^2 .map: &say }; 1.map({ foo }).sink | ||
| committable6 | jnthn, gist.github.com/d052c3d939925b7dee...ad5e6f6179 | 09:17 | |
| dogbert17_ | samcv++, utf-c8 | 09:36 | |
| jnthn | *sigh* So I got a fix for RT #131593, but it regresses another test, and I'm confused now about however we could have not had that bug and passed the test in question | 09:46 | |
| dogbert17_ gets the feeling that there's a bot missing | 09:47 | ||
| lizmat | m: dd (1,2,3).WHICH # /me is confused as to why List aren't value types | ||
| camelia | ObjAt.new("List|80229648") | ||
| jnthn | m: sub foo($?) { ^2 .map: &say }; foo() for 1 | 09:48 | |
| camelia | ( no output ) | ||
| jnthn | c: 2017.05 sub foo($?) { ^2 .map: &say }; foo() for 1 | ||
| committable6 | jnthn, ¦2017.05: «» | ||
| jnthn | Wait, what was the "this worked since summer of 2016" in rt.perl.org/Ticket/Display.html?id=131593 based on? | 09:49 | |
| I don't see any example in the ticket that shows that it ever worked | 09:50 | ||
| lizmat | m: dd (1,2,3).WHICH # jnthn do you agree that this .WHICH is faulty ? | 09:51 | |
| camelia | ObjAt.new("List|69772560") | ||
| jnthn | lizmat: List is a reference type, ain't it? | ||
| lizmat | is it? | 09:52 | |
| A list is supposed to be immutable, no ? | |||
| granted, it could contain mutable containers | |||
| jnthn | Yes, but that doesn't mean "it's a sensible value type" | ||
| Right, exactly | |||
| Also it's lazy | |||
| pmurias | should I remove a test which tests 'need STD; STD.parse("say <oh hai">)'? | 09:53 | |
| jnthn | Oh, it's the .&foo case that it's claimed changed | 09:55 | |
| Geth | roast: 336e3c069d | pmurias++ | S05-grammar/std.t Remove a test that doesn't seem to make sense anymore |
09:58 | |
| jnthn | Ah, and phew, I think that one is covered by the fix I already made | 09:59 | |
| Geth | rakudo/nom: 6b634a3695 | (Jonathan Worthington)++ | src/Perl6/Actions.nqp Restore sinking semantics of `for`. The optimization performed in 9b0b9effe5 caused some previously sunk cases to not be sunk. This restores that behavior, thus addressing RT #131593. |
10:06 | |
| roast: 3dc96c76dd | (Jonathan Worthington)++ | 2 files Tests for RT #131593. |
10:07 | ||
| samcv | what causes files to be put in ~/.perl6 ? is it zef or something in rakudo | 10:33 | |
| geekosaur | the CompUnitRepo stuff in rakudo. happens automatically when you run a perl 6 script, it gets compiled into the CUR | 10:48 | |
| I think zef/panda just force compilation with perl6 -c or equivalent | |||
| lizmat wonders what the rationale is for subst-mutate to return its matches | 10:49 | ||
| current behaviour prohibits fastpathing Str.subst-mutate(Str,Str) | 10:52 | ||
| and it is tested for :-( | |||
| moritz: shouldn't perl6book.com be updated ? | 11:09 | ||
| :-) | |||
| moritz | lizmat: it should | 11:12 | |
| lizmat: I've asked jj merelo for a pull request, and I'll add my own stuff as well soon :-) | |||
| tadzik wonders how could one integrate a "I have no idea what this graph does" into the graph and make it point to "Think Perl6" :P | 11:13 | ||
| moritz | tadzik: I tried to not add a bias for one particular book | 11:24 | |
| tadzik: if there's community-wide consensus that that is the best book for beginners, by all means, send a pull request :-) | 11:25 | ||
| tadzik | moritz: sure; I just half-jokingly imagine someone looking at a flowchart not understanding how flowcharts work :P | ||
| lizmat | moritz: perhaps make the start point: do you know how to program? No -> Think Perl 6 | 11:28 | |
| Yes -> current start point | |||
| moritz | lizmat: but there are two multiple books targeting folks who don't know how to program | 11:30 | |
| lizmat | which ones? | 11:31 | |
| moritz | lizmat: Think Perl 6, the new one by JJ merelo, and "Learning Perl 6" | 11:32 | |
| lizmat | ah, yes, good point | ||
| hmmm... suggest both ? | 11:33 | ||
| moritz | the first question should be: who do you trust in the Perl 6 community? | 11:36 | |
| and then you get a different flowchart, depending on what you answer :-) | |||
| lizmat | hehe :-) | 11:43 | |
| Geth | rakudo/encoding-objects: 6959211f59 | (Jonathan Worthington)++ | 4 files Extract Encoding::Decoder role. The API defined in VMBackedDecoder is now used for all of the I/O decoding use cases in Rakudo; additionally, it has been demonstrated that good performance can be attained using it. Therefore, it now becomes the tentative API that userspace decoders will implement also. |
12:41 | |
| rakudo/encoding-objects: 39efcd9fb7 | (Jonathan Worthington)++ | 4 files Add Encoding and Encoding::Encoder roles. Where Encoding represents an encoding, and can be used to obtain both a streaming decoder and an encoder. The Encoding::Encoder role represents an encoder (which is, thankfully, a far simpler thing than a decoder). It's tempting to call the method just `encode`, but given it will have a different API to `Str.encode` it may be better to have it different (a `Encoding::Encoder` will pre-commit to things like replacement characters and newline translation). |
12:57 | ||
| rakudo/encoding-objects: 203737f652 | (Jonathan Worthington)++ | 4 files Add an Encoding::Registry class. This will hold encodings that have been registered. We'll register the various built-in ones in here, and userspace encodings will also be registered this way too. This will come to be used to look up encodings by name. |
13:51 | ||
| roast: 4b979dd7fc | (Jonathan Worthington)++ | S32-encoding/registry.t Add tests for Encoding::Registry. |
13:54 | ||
| travis-ci | NQP build canceled. Samantha McVey '[unicode-grant] Bump MoarVM to get unicmp_s improvements and changes | 14:01 | |
| travis-ci.org/perl6/nqp/builds/244394686 github.com/perl6/nqp/compare/bc4ff...f261cb41f0 | |||
| Geth | rakudo/encoding-objects: 4e6fa79869 | (Jonathan Worthington)++ | t/spectest.data Run S32-encoding/registry.t. |
14:29 | |
| rakudo/encoding-objects: 2c80d5e6aa | (Jonathan Worthington)++ | 7 files Move Rakudo::Internals::VMBackedDecoder. It's now `Encoding::Decoder::Builtin`. We'll not spec or document this type, but rather say that for built-in encodings the exact types you get back from `Encoding::Registry.find`, and from `.decoder` and `.encoder` on the retunred object, are all implementation defined and so subject to change; the only thing to be relied on is them doing the `Encoding`/`Encoding::Encoder`/`Encoding::Decoder` roles. Also generalize various places to the `Encoding::Decoder` role. |
|||
| rakudo/encoding-objects: da1b37a79a | (Jonathan Worthington)++ | 3 files Add Encoding::Builtin; register built-ins. An implementation of `Encoding` for the built-in encodings. Register all of the built-in encodings using it. |
14:51 | ||
| rakudo/encoding-objects: 798ce1fbf8 | (Jonathan Worthington)++ | src/core/Encoding/Builtin.pm Register utf8-c8 encoding. |
15:16 | ||
| rakudo/encoding-objects: c4e9a8708a | (Jonathan Worthington)++ | 6 files Use Encoding::Registry to obtain a decoder. This also means that all I/O just got support for userspace decoders. |
15:18 | ||
| timotimo | the explosion ugexe found in ecogen is the same place that exploded with dogberts program; the gc_mark of MVMCallCapture | 15:23 | |
| Geth | rakudo/encoding-objects: e62ab72c1c | (Jonathan Worthington)++ | src/core/Encoding/Builtin.pm Use typed exception for illegal new. MasterDuke++ and Zoffix++ for code review. |
15:25 | |
| jnthn | Started the encoding side of it, will get that in place tomorrow, along with what should be a nice speedup for output. :) | 15:46 | |
| ugexe | will we be able to do multiple encodings, e.g. encode to latin-1 and then base64 encode? | 16:01 | |
| jnthn | ugexe: Should work out, would just need to write an implementation of the roles that could pass stuff from one to the other or some such | 16:04 | |
| I think we should probably allow all the places you can pass an encoding to let you pass either a string (resolved by the registry) or an Encoding object | 16:06 | ||
| So you don't need to register a chain anywhere | |||
| Just build it and pass it in when opening the handle | 16:07 | ||
| [Coke] waves. | 17:07 | ||
| [Coke] wonders if eveo's fix for the proc in/out stuff should be in a fresh-built nom. | 17:20 | ||
| eveo | [Coke]: it is. It's in the release too. But it's a fix for a race for lost input. There's another problem that involves segfaults | 17:22 | |
| [Coke]: rt.perl.org/Ticket/Display.html?id=131592 | 17:23 | ||
| Which will likely be fixed when the piping of procs is reworked some time later this week. | |||
| (by jnthn++) | |||
| [Coke] | I'm having (i think) the lost input in the docs xt/ tests that pipe input. | ||
| eveo | Reworked to fix rt.perl.org/Ticket/Display.html?id...et-history | 17:24 | |
| [Coke] | seems to run fine standalone, but as part of 'make xtest' get a few failures. (2017.06-13-g6b634a369) | ||
| [Coke] does a bit more testing | |||
| eveo | [Coke]: my fix makes this program not fail the test on multiple runs: gist.github.com/zoffixznet/440cf8f...dbb3a625b3 | 17:25 | |
| [Coke] | ah. this may be the segfault then. I just ran one by hand it just stopped after 72 out of 305 tests. | 17:27 | |
| Thanks. | |||
| eveo | cooll | 17:28 | |
| ugexe | added file locking to zef because of your blog post :x | 17:50 | |
| i probably missed something in the fetching to tmp though... | 17:51 | ||
| AlexDaniel | dogbert17: re “gets the feeling that there's a bot missing”: yeah, you can write a bot too :) | 19:18 | |
| dogbert17 | AlexDaniel: missing as in crashed | 19:29 | |
| AlexDaniel | dogbert17: we need a bot that will monitor other bots and graph their crashes! | 19:30 | |
| dogbert17 | its the RT bot, is it called synapsebot? | ||
| RT #131384 | |||
| geekosaur | synopsebot | 19:33 | |
| dogbert17 | thx, who controls it? | ||
| geekosaur | actually synopsebot6. checking logs to see if I can spot who normally keeps track of it | 19:34 | |
| dogbert17 | seems to have gone missing around a month ago | 19:35 | |
| Geth | roast: 0f743ddbb0 | (Brad Gilbert)++ (committed using GitHub Web editor) | S04-statements/for.t remove copy pasta-ed 'modifier' from description |
19:37 | |
| geekosaur | looks like you want timotimo | ||
| dogbert17 | .seen timotimo | ||
| yoleaux | I saw timotimo 18:01Z in #perl6: <timotimo> you forgot a "sub" in front and { ... } after the signature | ||
| geekosaur | also, last quit was 8 days ago | ||
| dogbert17 | geekosaur, thx | 19:38 | |
| geekosaur | #perl6.log:2017 Jun 11 05:38:26 *synopsebot6 has quit (Read error: Connection reset by peer) | ||
| dogbert17 | .tell timotimo synopsebot6 seems to have died on June 11th, can you restart it? | ||
| yoleaux | dogbert17: I'll pass your message to timotimo. | ||
| AlexDaniel | should we write that timotimo manages it? github.com/perl6/doc/issues/711#is...-235414744 | 19:41 | |
| timotimo | hey | 19:44 | |
| yoleaux | 19:38Z <dogbert17> timotimo: synopsebot6 seems to have died on June 11th, can you restart it? | ||
| timotimo | oops | ||
| dogbert17 | yo :) | ||
| RT #131384 | 19:45 | ||
| synopsebot6 | Link: rt.perl.org/rt3/Public/Bug/Display...?id=131384 | ||
| dogbert17 | yay | ||
| timotimo: how would you describe synopsebot with a single sentence? | 19:46 | ||
| AlexDaniel | .tell dalek It's still alive!!! | ||
| yoleaux | AlexDaniel: I'll pass your message to dalek. | ||
| timotimo | creates links to the synopses and turns mentions of RT ticket numbers into clickable RT links | ||
| dogbert17 | thx, I'll add that to the doc issue pointed out by AlexDaniel above | 19:47 | |
| list edited | 19:48 | ||
| timotimo++ | |||
| AlexDaniel, didn't you create a new bot the other day? | 19:49 | ||
| AlexDaniel | ooh… should list 9 bots right now | 19:50 | |
| dogbert17: full list here: github.com/perl6/whateverable/wiki | |||
| yeah, greppable6 is missing | |||
| dogbert17 | fixed | 19:52 | |
| AlexDaniel | do we have anything better than an editable comment for this? | 20:12 | |
| github.com/perl6/doc/wiki ? | 20:13 | ||
| lizmat | does anybody know where I can find an updated graph: twitter.com/zoffix/status/865690079217561600 | 20:42 | |
| ? | |||
| eveo | benchable6: help | 20:44 | |
| benchable6 | eveo, Like this: benchable6: f583f22,HEAD my $a = ‘a’ x 2¹⁶; for ^1000 {my $b = $a.chop($_)} # See wiki for more examples: github.com/perl6/whateverable/wiki/Benchable | ||
| eveo | benchable6: 2017.05,2017.06 for ^10000000 {} | 20:45 | |
| benchable6 | eveo, starting to benchmark the 2 given commits | ||
| eveo, benchmarked the given commits, now zooming in on performance differences | |||
| eveo, ¦2017.05: «4.4396» ¦2017.06: «1.6663» | |||
| eveo | hm.... I thought it was the bot that spit out that chart in svg | 20:46 | |
| on gist | |||
| lizmat | benchable6: 2017.01,2017.02,2017.03,2017.04,2017.05,2017.06 for ^10000000 {} | ||
| benchable6 | lizmat, starting to benchmark the 6 given commits | ||
| eveo | ah needs more than 5 commits to graph | 20:47 | |
| lizmat | yup, /me just read the documentation | ||
| benchable6 | lizmat, benchmarked the given commits, now zooming in on performance differences | 20:48 | |
| lizmat, gist.github.com/601f08c103e3a1a05d...664cec570b | 20:49 | ||
| eveo | heh, doean't display as an image on my phone :/ | 20:50 | |
| lizmat | works fine here :-) | 20:52 | |
| and another Perl 6 Weekly hits the Net: p6weekly.wordpress.com/2017/06/19/...l-6-books/ | 21:26 | ||
| afk for a bit& | 21:28 | ||
| AlexDaniel | .tell eveo so is RT #131592 resolved now? | 22:17 | |
| yoleaux | AlexDaniel: I'll pass your message to eveo. | ||
| synopsebot6 | Link: rt.perl.org/rt3/Public/Bug/Display...?id=131592 | ||
| lizmat | good night, #perl6-dev! | 22:54 |