00:00 <headius> ahh

00:00 <headius> ok master is already on a recent jcodings and this started failing there so I excluded it

00:00 <lopex> ah

00:01 <lopex> only one excluded ?

00:02 <headius> there's one other

00:02 <headius> look at TestTranscode.rb on master

00:02 <headius> it looks like the change happened going from 1.0.25 to 1.0.26

00:03 <headius> wait maybe not

00:04 <headius> https://github.com/jruby/jruby/issues/4731

00:05 <lopex> yeah, I dint want to comment since I didnt know what was there

00:06 <lopex> but now I think we're good

00:06 <lopex> headius: but is this related?

00:06 <headius> this is the bug I associated these tagged specs with

00:06 <headius> but I don't know if it's related

00:06 <headius> I'm going to just tag that one on 9.1 and you can look into it

00:06 <headius> everything else looks ok

00:07 <lopex> headius: which specifically

00:07 <lopex> maybe stale mri tests ?

00:07 <headius> it's possible

00:07 <headius> master should have more recent 2.4 tests but not updated since summer

00:08 <lopex> cms is another thing which I'm working on now

00:08 <lopex> *cmp

00:08 <headius> test_transcode.rb, test_windows_1255

00:08 <headius> I'm only excluding that one because it's also excluded on master

00:08 <lopex> yeah, I'll try to reduce that

00:08 <lopex> Fsure

00:09 <headius> there are two Integer#chr failures too, same sort of thing where it expects an error

00:09 <lopex> headius: hmm

00:10 <headius> I'm not super worried about either of these but we should fix them

00:10 <lopex> headius: there's one change you might be interested in https://github.com/jruby/jcodings/commit/fbb7a04fbd7e0e625b1a73e69809f6ddfd466c5f

00:11 <lopex> and it's correct

00:11 <headius> ahh yeah that one

00:11 <lopex> I mean it;'s fixed wrt mri

00:12 <lopex> headius: https://github.com/jruby/jcodings/commit/9a0a9e4f96b38b5441e7ef52d0c190e6c154761e

00:12 <headius> Encoding.default_internal = 'US-ASCII'; 0x100.chr

00:12 <headius> that should error but does not

00:13 <lopex> no idea about hat one

00:14 <lopex> ascii should error on 0x80 right ?

00:14 <lopex> ah, us ascii

00:15 <headius> I['ll look into it

00:15 <lopex> headius: who asked for that

00:15 <headius> that's in Integer#chr specs

00:15 <lopex> oh cool

00:16 <headius> looks like missing chr logic

00:18 <lopex> headius: that if (0xff < i) { ?

00:18 <lopex> er

00:19 <lopex> headius: what's the case ?

00:20 <headius> there's at least one explicit check for US-ASCII I don't see in our logic

00:21 <lopex> but it might translate to encoding specific logic

00:22 <lopex> rb_enc_uint_chr ?

00:22 <headius> there's some uint juggling

00:22 <headius> I have to call it a night too though

00:23 <headius> I'm just going to leave things as is until we can look at them

00:23 <lopex> but why are you focused on it ?

00:24 <lopex> ginme the case

00:27 <headius> lopex: https://gist.github.com/headius/ac6c14803af12033391a05ea5593b9d0

00:27 <headius> those are the two specs failing on 9.1

00:28 <lopex> headius: cool

00:28 <lopex> it's something

00:29 <lopex> if its that someting

00:30 <GitHub182> [jruby] headius pushed 1 new commit to jruby-9.1: https://git.io/vNPPo

00:30 <GitHub182> jruby/jruby-9.1 566405d Charles Oliver Nutter: Bump up complete jar size to accommodate joni and jcodings update.

00:35 <headius> ttfn!

06:26 <GitHub17> [jruby] headius opened issue #5012: Consider switching JSR223 engine to use "persistent" local variables https://git.io/vNPxP

08:31 <StevenV> Hi guys, I was wondering why a lot of the download links to older jruby versions don't work?

09:26 <olle> I want to send praise for nomadium's work on improving tests and gardening the code around that! https://github.com/jruby/jruby/pull/5011

19:16 <enebo> lopex: https://travis-ci.org/jruby/jruby/jobs/333553098

19:16 <enebo> lopex: are you examing the what I suspect are jcodings errors?

19:19 <GitHub63> [jruby] enebo closed pull request #5010: Fix `convertCommon` (jruby-9.1...rational_convert) https://git.io/vN60j

19:19 <GitHub84> [jruby] enebo pushed 4 new commits to jruby-9.1: https://git.io/vN1Rg

19:19 <GitHub84> jruby/jruby-9.1 ca984cb yui-knk: Update function names...

19:19 <GitHub84> jruby/jruby-9.1 436527f yui-knk: Fix spec:ruby:fast

19:19 <GitHub84> jruby/jruby-9.1 ca0f447 yui-knk: Fix `convertCommon`...

19:19 <lopex> enebo: will look tomorrow

19:20 <lopex> enebo: that chr most likely is jruby issue

19:20 <enebo> lopex: ok. I think it is 3 things

19:20 <enebo> lopex: yeah it definitely would not surprise me

19:20 <lopex> enebo: fix in jcodings exposed it

19:20 <enebo> lopex: but you nearly have killed all test_regexp.rb failures

19:20 <lopex> enebo: mostly worried about that transcoder thing

19:21 <enebo> lopex: yeah it is on windows though and we do weird shit for windows

19:21 <enebo> lopex: that one also makes me suspect JRuby itself

19:21 <enebo> but who knows

19:21 <enebo> you will :)

19:21 <lopex> enebo: yeah, I havent touched transcoding code

19:21 <lopex> enebo: there's been one bug in transcoding data thoigh

19:22 <lopex> *though

19:22 <lopex> enebo: where do you see it in that output ?

19:22 <lopex> those failures

19:23 <lopex> enebo: that ungetc also was jruby issue

19:23 <enebo> lopex: I think it is travis test:mri. Obviously it is running on linux so I am confused but it is a windows encoding

19:23 <enebo> or codepage or whatnot

19:24 <enebo> https://api.travis-ci.org/v3/job/333553088/log.txt

19:24 <enebo> lopex: ^

19:24 <enebo> TestTranscode#test_windows_1255

19:24 <enebo> and it is an error check as well and we seem to pass :)

19:27 <lopex> I tent to run test:mri:fullint locally

19:27 <lopex> *tend

19:28 <lopex> enebo: there's two low hanging ones in test_regexp

19:28 <lopex> enebo: one is message name

19:29 <lopex> enebo: the other is RangeError where we raise RegexpError

19:29 <lopex> but it's from the parser

19:30 <enebo> lopex: ah well you can commit to jruby too if you want :)

19:31 <lopex> enebo: I did for at least three failires

19:31 <lopex> *failures

19:34 <lopex> enebo: also String#casecmp is coming too

19:34 <lopex> and those up/down casing

19:34 <lopex> since jcodngs has all the bits now

20:54 <GitHub173> [jruby] arjanvandervelde opened pull request #5013: fix issue where expr fails on freebsd. (master...master) https://git.io/vN1Kb

21:39 <GitHub121> [jruby] nomadium opened pull request #5014: Remove TestArray#test_uniq_bang_with_freeze exclusion test case (master...fix-array-test-uniq-bang-with-freeze) https://git.io/vN1MZ

22:19 <pcarlisle> I've been profiling and have noticed that when I have a stored regex it gets recompiled every time I match against it. It seems to be because the regex is in US-ASCII and the string I'm matching is UTF-8. Is this expected? From reading the java it seems like it should at least be caching the recompiled regex but the profiling suggests it isn't (or isn't using the cached version). It's surprising behavior since

22:19 <pcarlisle> it seems if you write a literal regex and a literal string in currently jruby the regex is ascii and the string is utf-8.