Perl 6 - the future is here, just unevenly distributed

IRC log for #moarvm, 2017-09-21

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:55 ilbot3 joined #moarvm
01:55 Topic for #moarvm is now https://github.com/moarvm/moarvm | IRC logs at  http://irclog.perlgeek.de/moarvm/today
02:00 mtj_ joined #moarvm
02:22 geekosaur joined #moarvm
02:34 geekosaur joined #moarvm
03:08 geekosaur joined #moarvm
03:51 zakharyas joined #moarvm
04:41 samcv just got my new 3k screen in my laptop. experimenting with the joys of hidpi display on linux. luckily kde seems to have the best support for it. allowing you to set fractional scale's, while gnome can do 1x, 2x or 3x only
04:53 Ven`` joined #moarvm
05:32 brrt joined #moarvm
05:49 domidumont joined #moarvm
06:11 domidumont joined #moarvm
06:46 brrt joined #moarvm
07:15 nine samcv: are you on X or Wayland?
07:17 moritz .oO( Xayland! )
08:05 domidumont joined #moarvm
08:10 leont joined #moarvm
08:21 robertle joined #moarvm
08:56 jnthn morning o/
09:21 brrt morning jnthn
09:21 jnthn o/ brrt
09:23 brrt i'm working on new documentation for the expr JIT by the way
09:24 jnthn Cool
09:24 brrt i think the tiler is still fantastically complex, even though it is actually composed of simple elements
09:24 jnthn I'm continuing my re-working of supply internals
09:24 brrt uhuh. that's mostly in perl6 land, isn't it?
09:25 jnthn Entirely, yeah
09:25 jnthn Still tricky though :)
09:25 moritz oh I can imagine
09:25 brrt oh, i'm sure it is :-)
09:26 moritz jnthn: btw a proof-reader of my current Perl 6 regexes book wrote: ```My wish is that one of you guys with the big brains will write a thick book: "Perl 6 Concurrency and Asynchronous Programming" full of elegant examples of useful constructs.
09:27 moritz '''
09:27 moritz :-)
09:27 jnthn moritz: Yeah, don't tell anyone but I'm already working on it. :)
09:27 jnthn Unfortunately, I'm also working on many other things.
09:27 jnthn So it's taking a while and probably will take a while longer yet
09:27 moritz jnthn: I wrote that you are toying with the idea of writing such a thing :-)
09:28 jnthn Yeah
09:28 jnthn I mean, there's actually text slowly coming together for pieces of it, but nothing I'm ready to share just yet
09:28 moritz if you happen to want a private beta (or alpha) reader, I'm all yours :-)
09:28 jnthn I should perhaps resume my writing sessions in the pub that has no wifi :)
09:29 moritz and if you happen to want a contact with a publisher at some point, that might be arranged too :-)
09:29 jnthn Ah, cool. I've started out with LeanPub, fwiw.
09:30 jnthn Happy with it so far
09:30 jnthn Like writing in Markdown
09:31 jnthn Then, I like writing in Markdown a bit too much. I wrote all the Cro docs in Markdown too instead of finally learning Pod6 :P
09:40 samcv i'm realling loving writing in asciidoc. having auto generated table of contents is nice. and it's basically markdown, except anything you can't do in markdown you can do in asciidoc. like start numbered lists from values other than 1 for example.
09:42 samcv you can even include data from a csv to create a table if you want
09:42 jnthn Oh, interesting.
09:43 samcv having a NOTE: and WARNING: macros is nice too
09:44 samcv like the one i have here: https://github.com/MoarVM/MoarVM/blob/master/docs/strings.asciidoc#abstract so i'd think it'd be nice if you're writing a book. though not sure if leanpub supports it. you can also control alignment of tables. or make tables without a header. so a plain table
09:45 samcv uh. and bold things inside single ` ` marks
09:54 samcv asciidoc at least has a book format you can output to i know. but use whatever is best for you :)
09:54 timotimo let's make a markup format called "Whatever" and base it on the syntax of whatever stars and method calls
09:59 TimToady joined #moarvm
09:59 stmuk most of the current ecosystem docs (such as they exit) for perl 6 modules uses markdown
10:00 stmuk s/exit/exist
10:12 zakharyas joined #moarvm
10:25 samcv a lot of them are in pod6 though which gets converted to markdown
10:25 samcv anyway i'm off to bed. o/
10:26 jnthn Rest well, samcv o/
10:38 jnthn Righty, time to stresstest this big refactor...
10:43 jnthn Darn, two new hangs and two new regressions. But one fix.
10:44 nwc10 jnthn: I still see a SEGV when building the setting.
10:44 nwc10 have a coredump, but can only reproduce it "live" - running under gdb makes it go away
10:45 nwc10 http://paste.scsys.co.uk/565159
10:45 nwc10 ASAN's first report is "SEGV" (nothing naughty before then)
10:46 jnthn Hm,odd
10:46 jnthn Haven't seen that at all, though I'm off in a branch and maybe not at MoarVM HEAD
10:48 nwc10 slightly more info http://paste.scsys.co.uk/565160
10:48 nwc10 and really can't say more useful than that
10:48 nwc10 (sorry)
10:48 nwc10 and don't want to distract you
10:48 nwc10 and I'm also mostly AFK
10:48 brrt jnthn, re: markdown, i think you should like org-mode, if you can stomach emacs anyway
10:51 brrt otoh, what samcv did with asciidoc looked decnetly nice as well
10:52 timotimo nwc10: would also like to see the STable's debug_name perhaps
10:52 jnthn OK, two hangs and 1 regression fixed
10:55 jnthn Hm, and the final regression looks interesting
10:57 jnthn ah, mebbe a thinko :)
10:58 jnthn Yay, indeed
10:59 jnthn Now I'm one test ahead of where I was before the refactor. Plus another problem fixed :)
11:00 jnthn And I think only at one regression vs. nom. Which is in throttle.t /o\
11:06 nwc10 (gdb) p st->debug_name
11:06 nwc10 $6 = 0x21934e0 "Perl6::Grammar"
11:25 jnthn lunch &
12:07 domidumont joined #moarvm
12:08 * jnthn back
12:08 jnthn So, this throttle test...
12:19 jnthn urgh
12:21 nine jnthn: there may be one open issue with jit_nativecall. The branch changes the nativecallbuild op to have a return value. I it ok to change ops that way? Or should I create a new op and deprecate the old one?
12:22 jnthn nine: We kinda try not to, and if the op were used in the NQP bootstrap you'd totally not get away with it
12:22 nine jnthn: its only use in nqp is in: QAST::MASTOperations.add_core_moarop_mapping('buildnativecall', 'nativecallbuild', 0);
12:22 [Coke] to brighten jonathan's day, Coke notes that he just found a 5 year old typo in some coldfusion code he wrote to deal with multi-threaded issues as: “lock type="exlusive" { ... }” ... which apparently never errored out, nor complained until we upgraded our ColdFusion version recently. whoops.
12:23 jnthn An exlusive bug... :)
12:23 moritz nice one
12:23 jnthn nine: In this case you'd get away with it, and it's not like we really yet have the situation where people upgrade their Moar version and Rakudo version independently, except in development...
12:23 moritz these are the times when I use "git blame" :-)
12:23 [Coke] s/jonathan/jnthn/:oops
12:24 jnthn nine: Though if anyone does that then they'd get explosions :)
12:24 jnthn So the throttle problem is...partly from its rather questionable API design
12:25 nine jnthn: it wouldn't be the first time that one has to upgrade nqp/rakudo to really work with a new MoarVM version... Well, sounds like it'd not be worth the hassle.
12:25 AlexDaniel joined #moarvm
12:25 jnthn nine: Indeed. In the long run, such things aren't so wise, but present reality is we get away with it, and we have limited development resources
12:26 timotimo [Coke]: colonpairs in regexes go in the front! ;)
12:26 nine So, does anyone want to review jit_nativecall some more? Or should I just merge today?
12:26 [Coke] timotimo: :)
12:26 jnthn I'd like to give it some review, but my head has been (still kinda is) so full of the supply changes... :)
12:27 jnthn throttle is doing something I think I'd call broken
12:27 timotimo [Coke]: did you see my privmsg?
12:27 jnthn It takes a Supplier for :$control
12:27 jnthn Which is very odd in that it should really take a Supply since it's using it to receive control messages
12:27 jnthn It will then `.done` that Supplier
12:28 jnthn But that
12:28 jnthn results in deadlock because it's currently in a whenever block, and trying to run another of its own whenever blocks
12:29 jnthn So effectively, it's emitting a value to itself
12:30 vendethiel- joined #moarvm
12:34 jnthn Fixing throttle to not run into this is easy, though part of me wonders what else might run into it
12:42 zakharyas joined #moarvm
15:10 brrt joined #moarvm
16:52 Geth ¦ MoarVM: jdv++ created pull request #699: Fix returned tcp port values for async sockets (ntohs).  See RT#132135.
16:52 synopsebot6 Link:  https://rt.perl.org/rt3/Public/Bug/Display.html?id=132135
16:52 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/pull/699
16:53 jdv79 joined #moarvm
16:54 jdv79 if someone could merge 699 then i could push the tests or should it be the other way round
16:58 domidumont joined #moarvm
16:59 jdv79 tests are in a rt132135 branch on roast for now.  lunch&
17:00 jnthn It looks good to me; guess I'll let the automated checks have at it
17:01 domidumont joined #moarvm
17:02 leont joined #moarvm
17:05 erratum left #moarvm
18:07 jdv79 where are these robot checks?
18:10 AlexDaniel joined #moarvm
18:18 Zoffix travis/appveyor
18:21 jdv79 i dont know where those are
18:21 jdv79 also on my phone.  is there a doc or wiki thing on this stuff?
18:26 timotimo jdv79: when you open the pull request page it'll show you the two build services
18:27 timotimo something strange is happening on travis
18:27 timotimo in only one of the builds it's complaining:
18:27 timotimo Can't locate strict.pm in @INC (@INC contains: /home/travis/perl5/perlbrew/perls/5.10/lib/5.10.1/x86_64-linux /home/travis/perl5/perlbrew/perls/5.10/lib/5.10.1 /home/travis/perl5/perlbrew/perls/5.10/lib/site_perl/5.10.1/x86_64-linux /home/travis/perl5/perlbrew/perls/5.10/lib/site_perl/5.10.1 .) at Configure.pl line 3.
18:27 timotimo how can "use strict" be missing?!?
18:28 timotimo since it may just be a fluke i can just restart that individual job and see what happens
18:29 timotimo the appveyor build fails at what i consider a very suspicious run time, so probably got killed for time-out or something?
18:29 timotimo 59 min 59 sec
18:30 timotimo though it kind of looks like the next build is already at exactly the same spot in the tests but has only been running for 35 minutes?
18:30 timotimo i don't really know how to properly debug and fix an appveyor build :|
18:31 jdv79 i dont see that stuff on my android thingee
18:31 jdv79 lame mobile
18:31 jdv79 but ok.  ill look later from my real box.
18:35 AlexDaniel timotimo: re whatever markup: This reminded me of something. I toyed with the idea of subclassing from multiple grammars (for example CommonMark and Creole), so that you get the best of all worlds. It does work, if you get it done right :) Surely you have to take care of all syntactic conflicts and stuff, but it looks doable.
18:36 timotimo OK, the travis ci build succeeded, and appveyor is just unconditional fail it looks like
18:36 AlexDaniel the biggest problem for me at the time was writing a grammar that parses *any* input
19:10 AlexDaniel joined #moarvm
19:17 geekosaur joined #moarvm
19:56 robertle joined #moarvm
20:04 AlexDaniel joined #moarvm
20:09 AlexDaniel joined #moarvm
20:39 Geth ¦ MoarVM: c71c778145 | (Justin DeVuyst)++ | src/io/asyncsocket.c
20:39 Geth ¦ MoarVM: Fix returned tcp port values for async sockets (ntohs).  See RT#132135.
20:39 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/c71c778145
20:39 Geth ¦ MoarVM: 9b2e253baa | (Jonathan Worthington)++ (committed using GitHub Web editor) | src/io/asyncsocket.c
20:39 Geth ¦ MoarVM: Merge pull request #699 from jdv/master
20:39 Geth ¦ MoarVM:
20:39 synopsebot6 Link:  https://rt.perl.org/rt3/Public/Bug/Display.html?id=132135
20:39 Geth ¦ MoarVM: Fix returned tcp port values for async sockets (ntohs).  See RT#132135.
20:39 synopsebot6 Link:  https://rt.perl.org/rt3/Public/Bug/Display.html?id=132135
20:39 Geth ¦ MoarVM: review: https://github.com/MoarVM/MoarVM/commit/9b2e253baa
20:41 Ven`` joined #moarvm
21:57 Ven`` joined #moarvm
23:13 lizmat joined #moarvm

| Channels | #moarvm index | Today | | Search | Google Search | Plain-Text | summary