| lizmat | looks like test-t just got 15% faster again | 09:46 | |
| (at least for me) | |||
| dogbert17 | lizmat: what time did you get? | 12:22 | |
| and where is [Tux]? | |||
| Zoffix: can you test the example in RT #131375 on your Win 64? | 14:06 | ||
| synopsebot6 | Link: rt.perl.org/rt3/Public/Bug/Display...?id=131375 | ||
| Zoffix | dogbert17: "Unable to allocate an array of 20000000 elements" this is on 2017.07 | 14:34 | |
| dogbert17 | Zoffix: thx, but you need MoarVM 2017.08.1-19-g151a2563 | 14:55 | |
| :) | |||
| Nine++ made the fix and it works on my 32bit Linux system, it would be nice if we can test it on W64 as well | 14:56 | ||
| Zoffix | Ain't got one | ||
| dogbert17 | what system did you try it on, 32 bit lInux? | ||
| Zoffix | Win64. I mean, I don't have bleed Rakudo; only Star. Can't compile on this box; maybe one the weekend will test on my home box | 14:57 | |
| .in 3d check on Win10: rt.perl.org/rt3/Public/Bug/Display...?id=131375 | 14:58 | ||
| yoleaux | Zoffix: I'll remind you on 27 Aug 2017 14:58Z | ||
| dogbert17 | thanks a lot | 15:04 | |
| ugexe | the biggest array i could initialize on win10 using that RT example was 16_777_216 | 15:10 | |
| 1 more element and it blows up | 15:11 | ||
| stackoverflow.com/questions/125966...77216-in-c | |||
| dogbert17 | ugexe: what version of MoarVM do you have? | 15:14 | |
| the fix was made ~two days ago | 15:15 | ||
| ugexe | .08 | ||
| the last release | |||
| dogbert17 | ok, the fix was made after the release | 15:16 | |
| github.com/MoarVM/MoarVM/commit/f9...1351ca6058 | |||
| ugexe | ah i see. any idea why that wasn't causing issues like the RT in linux? | 15:17 | |
| dogbert17 | it did on 32 bit | 15:27 | |
| dunno why it worked on 64 bit Linux but not on W 64 | 15:28 | ||
| ugexe | definitely fixed on windows too | 15:31 | |
| I need a zef command like -to in `zef install -to=...` that acts just like -to but installs all dependencies into the -to location (even if the dependencies are already installed in other visible repos) | 15:55 | ||
| looking for ideas on the -option=... to use for this | 15:56 | ||
| remember that it could be mixed with -to, e.g. zef install foo -to=home -option=local | 15:58 | ||
| TimToady back from eclipse roadtrip | 16:10 | ||
| moritz | TimToady: was it worth it? | 16:12 | |
| moritz saw one in Germany some years ago; it was partially cloudly, so couldn't see the Corona; but still very impressive | 16:13 | ||
| TimToady | it was very worth it; we were in western Oregon near Corvallis where it was clear from clouds and wildfire smoke | 16:17 | |
| had about 2 minutes of totality, and we could see the corona | |||
| and Venus, and maybe Mercury | 16:18 | ||
| ah, no, it was Mars I saw | 16:19 | ||
| anyway, one can make all sorts of interesting observations, but as people who've been to an eclipse frequently point out, it's primarily a gut-level emotional experience | 16:22 | ||
| there's something primal about it | |||
| but also one just invests so much emotional energy in getting to the right place at the right time, with extra gasoline, all sorts of emergency supplies, etc. | 16:23 | ||
| when we got up at 3am in Medford, we still didn't know whether we'd be in eastern or western Oregon | |||
| western was closer, and the weather was supposed to be good both places, so we started up toward Salem, planning to go east into the mountains from there | 16:24 | ||
| but as the sun started coming up, we realized that there were tremendous amounts of smoke from forest fires where we were planning to go, so we headed west instead | 16:25 | ||
| and got out from under the smoke in time to find a little village called Pedee | |||
| where some kind folks had opened up their church so we could have a bathroom | |||
| one of the cooler things was watching the jet planes curling in to follow the path of the eclipse | 16:27 | ||
| Zoffix | :D | 16:29 | |
| TimToady | anyway, I can now check that off my bucket list, or could if I had a bucket list :) | 16:34 | |
| Zoffix adds it to the bucket list :) | 16:46 | ||
| The sky barelly darkened here due to it, but it had this surreal quality to it. | 16:47 | ||
| TimToady | now imagine that surrealness just cranking up over an hour, and getting faster and faster, such that you can see it getting darker every second | 16:48 | |
| [Coke] | I can compile win64 if someone needs something. | ||
| TimToady | and you can see the shadows moving on the ground of convection around the shadow | 16:49 | |
| and the wind changes as the umbra passes over | 16:51 | ||
| Zoffix | :D | ||
| TimToady | but there's nothing really to compare to just seeing a hole in the sky where there shouldn't be one | ||
| I can see why people get addicted to it | 16:53 | ||
| [Coke] | btw, getting this warning building on MSVS17: | 16:59 | |
| LINK : warning LNK4044: unrecognized option '/LC:\Users\User\rakudo\install\bin'; ignored | |||
| LINK : warning LNK4044: unrecognized option '/lmoar'; ignored | |||
| (the build is suprisingly fast, given that I'm doing it on windows inside a VM) | 17:00 | ||
| dogbert17: outputs "starting" then 'array initialized.' then exits. | 17:02 | ||
| This is Rakudo version 2017.08-5-g593fa5f87 built on MoarVM version 2017.08.1-19-g151a2563 | |||
| implementing Perl 6.c. | |||
| dogbert17: 1 billion is too many, though. | 17:05 | ||
| dogbert17: 100 million : Starting, followed by a moarvm panic | 17:06 | ||
| Geth | roast: cfa23079e3 | (Zoffix Znet)++ (committed using GitHub Web editor) | S05-grammar/methods.t Fix typo in test description |
17:16 | |
| dogbert11 | [Coke]: thx for testing, 100 million takes quite a bit of memory, how much do you have? | 17:52 | |
| samcv | TimToady, i have a question for you about regex and hoping you know something about this | ||
| m: my $a = 'aaaaaab'; 'a' x 999999 ~ 'b' ~~ /$a/; say now - INIT now | |||
| camelia | 7.31652197? | ||
| samcv | m: 'a' x 999999 ~ 'b' ~~ /aaaaaab/; say now - INIT now | 17:53 | |
| camelia | 0.00505117? | ||
| samcv | it's 150x different the amount of time it takes from literally typing in the value or if it's in a variable. and that's kind of unacceptable. so i'm wondering how we can fix this | ||
| Geth | roast: 1e4fd21486 | (Samantha McVey)++ | S03-operators/repeat.t Add tests for `x` operator to ensure normalization retained |
17:56 | |
| 6.d-prep: 8236870838 | (Zoffix Znet)++ (committed using GitHub Web editor) | TODO/FEATURES.md Add to TODO: Consistify Naming of `$*INITTIME` - github.com/perl6/roast/issues/296 - github.com/perl6/doc/issues/1462 - github.com/perl6/doc/issues/510 |
18:24 | ||
| perlpilot | m: my $a = rx/aaaaaab/; 'a' x 999999 ~ 'b' ~~ /$a/; say now - INIT now | 18:36 | |
| camelia | 4.2885446? | ||
| perlpilot | m: my $a = rx/aaaaaab/; 'a' x 999999 ~ 'b' ~~ $a; say now - INIT now | ||
| camelia | 0.0068104? | ||
| perlpilot | samcv: So ... I agree with you, but the user does have other options :) | 18:38 | |
| Geth | 6.d-prep: 594b10032a | (Zoffix Znet)++ (committed using GitHub Web editor) | TODO/FEATURES.md Better new name for $*INITTIME per github.com/perl6/roast/issues/296#...-324719710 |
18:40 | |
| samcv | perlpilot, well there's something that is taking such a long time when it's in a variable, and it shouldn't take that long | 18:54 | |
| when it completes 150x faster if you put the literal word there or use `index` or `nqp::index` | |||
| i mean the regex does use nqp::index but before that it does ??? idk what it does but it takes 149x longer than just running nqp::index | 18:55 | ||
| what good is my KMP algorithm if it spends 149x of its time doing something else? | |||
| also i wonder if i can implement the Knuth-morris-pratt algorithm backwards for MVM_string_index_from_end it must be possible. just process the needle backwards into the jump table, and then start at the end... | 18:58 | ||
| though i'm not sure how much that is used | |||
| yeah i don't see it used in the nqp source other that having the op added | 18:59 | ||
| Geth | roast: eaec1961ef | (Samantha McVey)++ | S03-operators/repeat.t Fix todo count for JVM in repeat file |
19:00 | |
| samcv | wow the unicode conference is expensive $995 for the conference only, or 1495 for conference and tutorial. 100 dollar discount for unicode members | 19:05 | |
| it's located right by where i live right now though. conference is October 16-18 | 19:06 | ||
| AlexDaniel | and another 100 dollar discount if you pay by September 1st :) | 20:23 | |
| b2gills | Rakudo is currently broken for me, “Missing or wrong version of dependency '/opt/rakudo/bin/../share/nqp/lib/MAST/Nodes.nqp' (from 'src/Perl6/Pod.nqp” | 21:08 | |
| I use this to build Rakudo: gist.github.com/b2gills/34bc96f5e8...dc6e703148 | 21:15 | ||
| Geth | specs: Xliff++ created pull request #122: Update S27-perl-culture-draft.pod6 |
21:20 | |
| specs: baec197a05 | Xliff++ (committed using GitHub Web editor) | S27-perl-culture-draft.pod6 Update S27-perl-culture-draft.pod6 Minor typo fix. |
21:24 | ||
| specs: cdca1ad1a9 | (Brad Gilbert)++ (committed using GitHub Web editor) | S27-perl-culture-draft.pod6 Merge pull request #122 from Xliff/patch-1 Typo in S27-draft Lost -> Lots |
|||
| samcv | b2gills, have you tried recompiling nqp then recompiling rakudo | 22:44 | |
| b2gills | What managed to fix it is basically “cd ~/Program/Perl6/rakudo/nqp; git checkout master; git pull; perl Configure.pl --prefix=/opt/rakudo --gen-moar; make install; cd ..; git pull; Configure.pl --prefix=/opt/rakudo; make install | 23:40 | |
| I did do a lot of things before that, but I don't think they changed anything | 23:41 |