Parrot 3.8.0 "Magrathea" | parrot.org | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 29 September 2011.
dukeleto blarg, my ssh key isn't setup properly on parrot@ftp-osl.osuosl.org 00:50
whiteknight they never are 00:51
it's amazing we have any docs up there
dukeleto whiteknight: leto.net/tmp/parrot/ 00:52
benabik dukeleto: Did you try my fixf ro postgres?
dukeleto whiteknight: can you copy those files over to the ftp server?
whiteknight I don't think my key is set up there either 00:53
dukeleto benabik: yep, already in master and part of the release. it got rid of the coredump, but test 13 was still failing for me, during some compilation envs
whiteknight in either case, I don't have time tonight. I can do it tomorrow morning
dukeleto whiteknight: ok, no worries
cotto: ping
soh_cah_toa: ping
benabik dukeleto: Ah. Awesome. Saw the TODO and had worried I hadn't fixed it.
dukeleto benabik: that test passes or fails depending on my compile flags. Haven't narrowed it down yet 00:54
benabik dukeleto: Ah. Heisenbug. Fair enough.
dukeleto in any case, i have pushed a tag and people are free to break master 00:55
dukeleto breaks to eat dinner 00:58
whiteknight dukeleto++ 00:59
cotto dukeleto: pong
just too late 01:00
whiteknight he wanted somebody to upload the docs 01:04
benabik Oh hey. My doc fixes will be online! 01:05
whiteknight benabik++ 01:08
soh_cah_toa dukeleto: pong 01:16
cotto dukeleto: the tarballs and checksums should be propagated rsn 01:35
dukeleto cotto++ 01:47
soh_cah_toa: i think cotto++ took care of it, was looking for people to send tarballs to our ftp site 01:48
soh_cah_toa dukeleto: ok. are you still having the error w/ auto_release.pl? it's really strange :\\
dukeleto soh_cah_toa: the error is caused my "make reconfig" needing to be run, i think 01:49
soh_cah_toa ah, ok
dukeleto soh_cah_toa: it is mentioned in the release manager guide
soh_cah_toa: auto_release.pl is not blocking me, but feel free to improve it :)
soh_cah_toa: it helps automate the first half of the release, so that was very nice 01:50
soh_cah_toa yup
dukeleto soh_cah_toa++ on writing that
soh_cah_toa yeah, i know people had been asking for something like that for a *long* time
dukeleto github.com/leto/writing/commit/c77...3a1f181575 01:51
cotto: wanna review my release announcement? ^^^ anybody else is welcome to as well
soh_cah_toa: take a look at the stuff I added to the bottom of the release announcement 01:52
soh_cah_toa: i would like to automate that stuff. It is easy.
soh_cah_toa: also, this is a simple task to make the release easier: github.com/parrot/parrot/issues/169
cotto dukeleto: looks good after a quick reading 01:53
soh_cah_toa dukeleto: the bottom the release announcement? you mean automating the authors since the last release? 01:54
dukeleto soh_cah_toa: yeah 01:55
dalek rrot: 79c26d4 | dukeleto++ | docs/project/release_manager_guide.pod:
[doc] Fix a typo in the release manager guide about SHA1 that should be SHA256
soh_cah_toa yeah, that would be kinda cool actually
dukeleto soh_cah_toa: i used git summary RELEASE_3_8_0...RELEASE_3_9_0
soh_cah_toa simple
dukeleto soh_cah_toa: git summary is from github.com/visionmedia/git-extras
soh_cah_toa: but it is just candy on top of plain git 01:56
soh_cah_toa oh cool
ooohhh...changelog popluation. very cool 01:57
yeah, this is totally awesome :)
dukeleto loves git extras 01:58
i use git ignore all the time
like: git ignore *.sw?
adds the appropriate regex to .gitignore
soh_cah_toa neat
dukeleto actually added the git summary x..y syntax :) 01:59
i really hate our website
02:11 nbrown joined
dukeleto www.parrot.org/news/2011/3.9.0 02:19
dalek website: dukeleto++ | Parrot 3.9.0 "Archaeopteryx" Released
website: www.parrot.org/news/2011/3.9.0
moderator Parrot 3.9.0 "Archaeopteryx" | parrot.org | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC 02:19
soh_cah_toa yay! \\o/ 02:22
dukeleto feel free to test a tarball in ftp.parrot.org/releases/supported/3.9.0/ 02:24
soh_cah_toa archaeopteryx: that's like a pterodactyl, right? or the same family or something
dukeleto soh_cah_toa: very similar. archaeopteryx is considered the transition point between dinosaurs and birds 02:25
soh_cah_toa ah...very appropriate :)
are you suggesting that parrot is in a transitional phase? :)
i'd agree w/ that. i love the symbolism 02:26
dalek rrot: a20b5c8 | dukeleto++ | docs/project/release_manager_guide.pod:
Update wiki location in release manager guide
 02:29
soh_cah_toa double yay \\o/
new wiki. i love it 02:30
very pretty
benabik soh_cah_toa: I think you're just biased. :-D
soh_cah_toa aw, come on. it just doesn't compare :) 02:31
it's much much smaller but most the stuff on the old wiki is just garbage
dukeleto en.wikipedia.org/wiki/Parrot_virtual_machine is updated 02:32
i have done everything in the release guide, except update the docs 02:50
because the instructions don't say where they actually are
cotto ~~ 02:57
parrotvm.osuosl.org iirc
dukeleto cotto: yes, which directory on that machine? 02:58
cotto /var/www/something/somethign
dukeleto cotto: github.com/parrot/parrot/issues/170
lulz
cotto for once I'm glad that parrot will soon have more issues 02:59
gah 03:00
pre--
that tag needs to be banned
benabik pre?
dukeleto cotto: lulz. are you complaining about the release blog post, or the github issue?
cotto I've seen ~2 sites that it doesn't break, and it's still annoying there.
dukeleto benabik: <pre>
cotto dukeleto, github issue
benabik What's wrong with it? 03:01
dukeleto cotto: yeah, i wrapped stuff in ``` in the issue, which <pre>-ifies. Feel free to remove if it offends
cotto benabik, try to read github.com/parrot/parrot/issues/170 without scrolling
benabik Ahhhh....
cotto dukeleto, it's a persistent pet peeve of mine, but probably not enough to change the issue 03:02
benabik I can preview comments, but not the issue itself? Wierd.
cotto: How's that? 03:03
Although it changed the number. Bleh.
cotto scolling violates laziness
benabik It now reads 1. instead of 6., but you can read it without scrolling.
cotto benabik, ah. didn't catch that you updated the issue 03:04
benabik++
I'm excited to see people interested in moving to issues 03:06
dukeleto indeed
the day i don't have to log into trac anymore will be a good day
benabik Hm. Made it actually say 6, but it looks worse. Bleh. 03:07
dukeleto doesn't really matter
the vagueness of the directions are what matter :)
cotto yes 03:08
dalek umage: 12e617a | soh_cah_toa++ | docs/ (7 files):
Changed documentation to use POD5 instead of POD6 since `perl6 --doc` just doesn't cut it right now.
 04:51
umage/soh-cah-toa/feather-format: 9c82ad0 | soh_cah_toa++ | docs/feathers/ (2 files):
Added first documents describing what a feather is and the featherspec specification.
 04:52
moritz benabik: then we'll fix it on the nqp side too
dalek rrot: 625b5bb | petdance++ | config/auto/warnings.pm:
Removed the -Wunused-local-typedefs, which is apparently still buggy.
 04:54
rrot: fd1a800 | petdance++ | src/pmc/scheduler.pmc:
removed unused var task_id_str. Consted a pointer, and marked an interp UNUSED
umage/soh-cah-toa/feather-format: f6991ef | soh_cah_toa++ | docs/feathers/intro.pod:
Fixed pathname used in docs/feathers/intro.pod.
 04:55
04:57 zpmorgan joined 05:47 contingencyplan joined 06:37 mudkip joined 06:39 mudkip joined 06:42 SHODAN joined
dalek kudo/nom: d7608c8 | moritz++ | src/Perl6/Metamodel/MROBasedMethodDispatch.pm:
speed up method cache creation. Patch by mls++
 06:49
kudo/nom: 36ebf15 | moritz++ | tools/build/NQP_REVISION:
bump NQP_REVISION
 07:29
07:42 mj41 joined
dalek p: 87a9ffb | moritz++ | src/6model/sixmodelobject.h:
typo
 07:57
p: c13964d | moritz++ | tools/build/PARROT_REVISION:
bump PARROT_REVISION
07:58 perlite joined
dalek kudo/nom: 9c3d3d1 | moritz++ | tools/build/NQP_REVISION:
fix NQP_REVISION, moritz--
 07:58
08:04 lucian joined 08:58 mj41 joined
dalek p: 7baf769 | mls++ | src/HLL/Compiler.pm:
Simplify control exception handling by using rethrow, fixes pir::exit (what was I thinking?). Tene++
 09:20
moritz where is auto_attrs documented? 09:40
ah, docs/pmc.pod 09:41
why do so many PMCs have both auto_attrs and a custom mark vtable? are they orthogonal? 10:15
10:44 jsut joined
Coke was about to update the directions for the doc update, at least, when lack of ssh access bit him. 11:10
11:22 marcel_r joined 11:34 whiteknight joined 11:38 Psyche^ joined 12:11 bluescreen joined
dalek rrot: 2206ed8 | mls++ | / (2 files):
add missing mark method to select.pmc, switch hash to interger keyed
 12:11
12:16 rfw joined 12:31 ambs joined
whiteknight good morning, #parrot 12:57
moritz good morning whiteknight 12:58
whiteknight hello moritz, how are you doing today? 12:59
moritz whiteknight: a bit tired after a not-so-quiet night, otherwise quite fine
whiteknight oh, I'm sorry to hear that. 13:00
moritz one of the prices of becoming a parent
whiteknight yeah, tell me about it. Mine has a stomach bug, and is needing diaper changes every hour or more 13:04
moritz ouch, that sounds much more painful than what I experience
(getting up 3..4 times per night, but only for short times; sleeping up to 6:30 AM) 13:05
whiteknight last week was a "bad sleep week" for some reason, but he's gotten better 13:14
moritz has the stomach bug been identified? 13:16
whiteknight no, he won't sit still long enough for analysis. I think it's just a transient thing 13:18
13:31 benabik joined
benabik o/ 13:31
13:52 dmalcolm joined 14:16 mj41 joined 14:25 bluescreen joined
benabik As a note for the long term future of Parrot… We need a storage format that doesn't become invalidated every month. Currently that's PIR, which seems LTA. 14:29
Coke this channel has a high concentration of former git-svn users, so I'll ask here: 14:33
benabik: That's /supposed/ to be PBC.
PIR was a stopgap. 14:34
NotFound Coke: PBC become inalidated more frequently than pir.
benabik Coke: The version changes on core_ops makes that impossible.
Coke (git-svn) I have master/trunk, and branch/branch ; I want to merge some changes from master into the branch. I'm happy to have them all show up as the individual commits replayed against the branch. is "git checkout branch; git merge master" all the magic I need? (I tried a git svn dcommit --dry-run after that, and it didn't look quite right.) 14:35
benabik, NotFound: I'm referring to the pre-1.0 plan, of course.
benabik I suppose the issue is actually that we change the version number of core_ops on a very very regular basis. If oplibs had a "compatibility" version, then it might last longer.
Coke (problem was, we declared 1.0 before the bytecode format was solidified to folk's content.)
benabik We might want to change which information is kept in the opmap to make it a bit more flexible. 14:36
Coke: I _think_ git-svn should be able to handle the merge. Not 100% sure. 14:37
Coke (we should be able to ship "binary" versions of parrot libs. (bytecode should be more stable than the language targetting it.)
benabik The only thing I'm aware of that invalidates PBC is the core_ops version.
This may be as simple as thinking more soundly about how we version that oplib. 14:38
Coke benabik: and is that something we need to do, or something that was easy to do because bytecode was already "unstable" ?
benabik: right.
benabik++ NotFound++
benabik Coke: There is a test in git.git that is supposed to ensure git-svn handles merges properly. 14:39
Coke: I can't tell you from experience that it works, but it looks like it's _supposed_ to.
NotFound Changes in the freezing of PMCs stored in the pbc are also a source of incompatibilities. 14:40
benabik Ah. Yes.
We don't version PMCs very well, do we?
Coke: Not sure if this helps, but here's the tests: github.com/git/git/blob/master/t/t...t-merge.sh 14:41
Are we actually changing the PMC freeze formats?
Oog. Anyway, I have places to go... 14:49
14:50 jsut_ joined
Coke benabik: aye, after looking more closely at the dry-run, seems ok. pushed, all ok. (just wish it automagically listed the author as the previous committer.) 14:53
14:55 jlaire joined, bacek_at_work joined, dukeleto joined, Khisanth joined, autark joined, alvis_ joined, athomason joined, japhb joined, pmichaud joined, mls joined, Util joined, szbalint joined 14:57 bluescreen joined, zpmorgan joined, janus joined, aloha joined, eternaleye_ joined, nopaste joined, awwaiid joined, jjore joined, TiMBuS joined, tcurtis joined, Hunger joined, Infinoid joined, nine joined, moritz joined, tadzik joined, ascent_ joined, sorear joined 14:59 jsut_ joined, mj41 joined, dmalcolm joined, nbrown joined, wagle joined, AzureStone joined, p6eval joined, rblackwe joined, dngor joined, tewk joined, betterworld joined, cosimo joined, ingy joined, slavorg joined, allison joined
dukeleto ~~ 15:09
dalek rrot: 44266eb | dukeleto++ | / (3 files):
Update changelog and fix some small nits in parrothist.pod and the release manager guide
 15:20
15:31 smash joined
smash hello everyone 15:31
dukeleto++ # release 15:32
dukeleto smash: thanks! 15:34
whiteknight hmm... plumage doesn't build for me 15:40
dukeleto whiteknight: oh noes! 15:45
whiteknight hmm....somehow I have an old version of Parrot installed in /usr/bin, not where I normally install it 15:48
Okay, that fixes it 15:55
I want to get the green_threads branch updated and merged to master soonish, unless there are objections
I love watching tests pass. Rosella and Jaesop pass all tests with 3.9.0 16:00
dukeleto whiteknight++ # merging green_threads 16:03
whiteknight: great to hear about the passing tests 16:04
whiteknight msg nine I want to talk about a green_threads merge to master soon. Let me know what you think 16:17
aloha OK. I'll deliver the message.
dalek TT #2199 created by Andy Dougherty <doughera@…>++: [PATCH] 3.9.0 Fails timer tests on NetBSD 5.1 16:32
TT #2199: trac.parrot.org/parrot/ticket/2199
rrot: 5b0b2aa | dukeleto++ | src/scheduler.c:
[TT #2199] Attempt to fix timer tests on NetBSD 5.1, doughera++
 16:39
ligne the test plan for t/dynpmc/select.t is off by one.
dukeleto ligne: it is probably off by one on your configuration. Sometimes test counts check for libraries and conditionally skip tests 16:41
ligne: can you gist the output that you see? If you can easy fix it and send a pull request, extra points :)
ligne dukeleto> just checking that :-) 16:42
i think it was introduced by 2206ed8111600cc99373e13173b96c82bbcfccf0. 16:43
16:44 fperrad joined
ligne dukeleto> done. the test all passes for me now. 16:51
dalek rrot: 6c60de1 | ligne++ | t/dynpmc/select.t:
fix the test count that wasn't incremented in 2206ed8
 16:53
rrot: b07f969 | mls++ | t/dynpmc/select.t:
Merge pull request #172 from ligne/select_test_count

fix the test count that wasn't incremented in 2206ed8
mls (sorry, didn't know there was a test count ;) ) 16:54
ligne no harm done, if you don't count the overlength cronspam that just dropped into my inbox :-D 16:55
dukeleto mls: did you run the tests? If you just run them with parrot, you wouldn't notice the issue. But prove -v t/dympmc/select.t would have caught it
mls I just used parrot, so I didn't notice it. will use prove in the future.
I patched hash.t yesterday, which doesn't have a test count, so I thought all tests don't have a count. 16:56
dukeleto mls: yeah, some do and some don't. it is an easy thing to miss 17:06
mls: there is a holy war about whether to have numbered test plans or not, and our test suite is one of the battlegrounds 17:07
numbered test plans provide a little more insurance that you are running exactly the number of tests you expect, but they are extra work to maintain 17:08
17:16 whiteknight joined
dukeleto goes on a TT-killing spree with a flamethrower 17:18
Coke -1 to plans with explicit #'s
dalek TT #954 closed by dukeleto++: Inf/NaN from ieee754-2008 17:19
TT #954: trac.parrot.org/parrot/ticket/954
TT #491 closed by dukeleto++: Use longopts in parrot debugger
TT #491: trac.parrot.org/parrot/ticket/491
NotFound The problem is that without a plan there is no way to diferentiate between a full run of the test and an unexpected premature exit. 17:22
dukeleto NotFound: indeed. And that has bit me many times. 17:23
Coke NotFound: that's what "done_testing" is for.
if you abort before done_testing, you didn't done. 17:24
dukeleto whiteknight: we need to carve up trac.parrot.org/parrot/ticket/619 17:25
Coke I will be very amused if we end up with a github ticket that points to trac that points to RT. ;)
(just amused. not sarcastically or shaedenfreudically. ;) 17:26
whiteknight :)
NotFound ... and none ot hem provide enough info.
A Parrot Archaeopteryx will be a lovely character in an anime show. 17:28
Maybe we should trademark it.
17:29 schmooster joined
dalek rrot/whiteknight/pcc_ops: ae90ef5 | Whiteknight++ | src/ops/experimental.ops:
First draft prototypes of most of the new ops we are going to need for a PCC revamp. For some reason, mk bootstrap-ops fails with a parse error, though I can't figure out why.
 17:30
whiteknight dukeleto: I think TT #619 has outlived it's usefulness. There's no reason to throw out our config system when it does a reasonably good job in a portable way 17:31
I say we close it, and if we need to address individual subtasks, we open new tickets
Coke +1 on rejecting 619. 17:32
NotFound whiteknight: probably parrot_hash_iterate is not parseable as a function call.
ack parrot_hash_iterate src/ops/ -> Nothing 17:33
Macros are evil ;) 17:34
17:34 Colin joined
sorear I find grepping compiler *output* to be very useful in cases like this. 17:34
dalek TT #610 closed by dukeleto++: cross-compile configuration 17:35
TT #610: trac.parrot.org/parrot/ticket/610
TT #2099 closed by dukeleto++: Automate more of the release manager process
TT #2099: trac.parrot.org/parrot/ticket/2099
TT #911 closed by dukeleto++: Tests for macro_local
TT #911: trac.parrot.org/parrot/ticket/911
TT #2120 closed by dukeleto++: We need sorting benchmarks
TT #2120: trac.parrot.org/parrot/ticket/2120
17:35 Colin left
NotFound sorear: I was just checking that parrot_hash_iterate was not being already used in ops. 17:35
dukeleto asks politely for people to look at their TT's and close any that are old, irrelevant, stupid or all of the above 17:37
the fewer TT's we migrate, the better. Let's get rid of the cruft
NotFound Morale of the history: parsing C with any other thing than a C compiler and preprocesor is hard... but we already know that, 17:38
whiteknight NotFound: Yeah, I'm thinking that too. I'll pull it out into a separate function soon
sorear NotFound: find . -name *.o | grep -l ... :>
xargs
NotFound sorear: you'll never find traces of a macro in *.o 17:39
sorear I thought you were looking for uses of a function that was hidden inside macros 17:40
NotFound sorear: I was checking that the macro was not used, to test the hypothesis that its first use will be the cause of whiteknight's problem. 17:41
(not used in src/ops/
Coke I am fairly certain that nqp-rx tickets assigned to pmichaud are probably not going to get worked on, given that (new) nqp is the in thing now. 17:44
17:49 linge joined
dalek TT #619 closed by dukeleto++: eliminate perl 5 dependency 17:51
TT #619: trac.parrot.org/parrot/ticket/619
TT #2184 closed by coke++: Remove Changelog
TT #2184: trac.parrot.org/parrot/ticket/2184
cotto ~~
17:54 cotto_work joined 18:01 mj41 joined, PacoLinux joined
dalek p: d2d759f | pmichaud++ | src/QRegex/Cursor.nqp:
Restore old meaning of $!regexsub to support intra-regex usage of $/.
 18:08
benabik Coke: We can migrate them to issues on our nqp-rx repo. Then if they're unimportant, we can kill them. 18:23
Oh. There isn't an parrot/nqp-rx. 18:24
Hm.
dalek kudo/nom: feb9401 | pmichaud++ | tools/build/NQP_REVISION:
Bump NQP_REVISION to get CAPSHASH/$!regexsub fix from nqp.
 18:25
benabik Hah. TT#366 is the same bug I was complaining about yesterday. 18:29
19:10 contingencyplan joined
dalek rrot: 7428bb5 | benabik++ | compilers/pct/src/PCT/HLLCompiler.pir:
HLLCompiler: Allow -e to play well with --target

  chrisdolan++ for giving us this patch in TT 366. However, it appears
NQP rewrites enough of PCT::HLLCompiler that this fix doesn't affect NQP or Rakudo nom.		 19:23
19:26 zby_home joined 19:27 whiteknight joined 19:32 bluescreen joined
dalek TT #1648 closed by petdance++: Look at using restrict keyword on pointers 19:41
TT #1648: trac.parrot.org/parrot/ticket/1648
TT #893 closed by petdance++: Rebuild dependencies in makefile
TT #893: trac.parrot.org/parrot/ticket/893
19:43 Coke joined 19:45 alester joined 19:48 mj41 joined
dalek p: 763b9c5 | benabik++ | src/HLL/Compiler.pm:
Use Hash.update for command line args
 20:06
p: 05b7873 | benabik++ | src/HLL/Compiler.pm:
HLL::Compiler: Allow --target with -e

Prior to this, using --target with -e would just result in the
  compiler ending early with no output. Of course, this would be more
useful if we taught _dumper how to deal with 6model objects.
Based on a patch from chrisdolan in Parrot TT366
benabik Github just added hooks for more events (like issues and pull requests): github.com/blog/964-all-of-the-hooks 20:11
whiteknight nice 20:12
dalek TT #366 closed by benabik++: [PATCH] --target=parse/past/post does not dump in -e mode 20:13
TT #366: trac.parrot.org/parrot/ticket/366
20:14 jsut joined
Coke benabik+ 20:23
benabik++
benabik Wow. A test file that writes 161 PIR files… Each of which tests a single line from the language. 20:29
whiteknight fun 20:30
benabik "What is this, I don't even."
Don't we have some kind of parrot config hash in the interp? How do I get at it? 20:31
Coke benabik: I can show in partcl. momentito. 20:32
github.com/partcl/partcl/blob/mast...b.pir#L130
benabik Coke++
NotFound++ # winxed 20:34
NotFound benabik: Some particular reason at this moment? 20:37
benabik NotFound: Just being generally fast and easy to use. :-D
Things like $include_const
NotFound Yeah, being easy is a main goal. 20:39
benabik Does using extern Test.More automagically load_bytecode 'Test/More'? 20:40
Apparently. O.o 20:41
NotFound benabik: yes
benabik winxed++
dalek kudo/nom: 973ebc9 | jnthn++ | src/Perl6/Actions.pm:
Fix return type check issue reported in RT#101364.
 20:53
whiteknight is really starting to hate ops2c 20:58
it sure takes a long time considering how fragile and unintelligent it is 20:59
cotto_work I blame the parents.
also, nqp
dalek rrot/whiteknight/pcc_ops: 569fd16 | Whiteknight++ | / (7 files):
Fix some parse/build problems with the new ops. Make headerizer and bootstrap-ops
 21:00
TT #1734 closed by pmichaud++: PCT::Compiler should null temporaries when they are no longer needed
TT #1734: trac.parrot.org/parrot/ticket/1734
TT #1013 closed by pmichaud++: 'undef' bareword shouldn't parse, but does
TT #1013: trac.parrot.org/parrot/ticket/1013
TT #1464 closed by pmichaud++: NQP-rx doesn't handle bare "return" from nested block correctly.
TT #1464: trac.parrot.org/parrot/ticket/1464
TT #1503 closed by pmichaud++: quote_delimited action misbehaves with single non-string quote_atom
TT #1503: trac.parrot.org/parrot/ticket/1503
whiteknight cotto_work: when you get a chance, could you take a looksie over those new ops? It's not a complete set of what we need for PCC, but it's most of it 21:01
cotto_work whiteknight: sounds like more fun than what I'm doing now.
whiteknight: no docs on Parrot_hash_flatten_hash_into? 21:02
whiteknight no docs on any of them yet
still prototyping
benabik Do we have regexen outside of nqp-rx?
whiteknight we have pcre bindings 21:03
Jaesop uses them
benabik Do we always have that, or are the optional?
whiteknight optional. Only if you have libpcre installed
benabik Bah.
21:06 soh_cah_toa joined
Coke PGE 21:06
benabik Coke: Is PGE painful to access outside of nqp-rx? 21:07
cotto_work whiteknight: how are those ops getting added? All I see is the generated ops code, not the original.
whiteknight: found them 21:08
Coke nqp-rx doesn't use PGE. 21:09
benabik Ah.
Coke Is using nqp-rx a problem? or, you want to use them from winxed? 21:10
benabik Trying to convert P5 split / *\\t\\s*|\\s{3,}/ to Winxed
Or NQP, I suppose. 21:11
NotFound I've not implemented regex in winxed because of PGE supposed to be a dead end. 21:13
benabik We should probably get a standard regex library in. It's not a very difficult thing 21:14
whiteknight libpcre is the standard
NotFound whiteknight: is Jaesop support finished? I can borrow from it. 21:15
whiteknight NotFound: no, but it's a start 21:16
dalek TT #1514 closed by pmichaud++: NQP-rx doesn't check storage mode for contextuals
TT #1514: trac.parrot.org/parrot/ticket/1514
whiteknight github.com/Whiteknight/jaesop/blob...inxed#L442
the only method I really have now is RegExp.test(). I don't have RegExp.exec() working yet 21:17
that's the hard one
benabik whiteknight: Then we should either say 'libpcre is required' or write a simple one in PIR.
This "I might have regexes available" kinda blows. 21:18
NotFound benabik: In PIR? 21:19
benabik NotFound: Compiles regexen to PIR.
NotFound benabik: ah, yes, but better write the compiler in winxed 21:20
benabik adds it to his list of projects.
21:21 soh_cah_toa joined
benabik Does winxed have nested functions? 21:23
And multiple returns? 21:24
NotFound If someone is really bored and want to test winxedxx, now it works with parrot provided winxed
benabik: anonymous functions yes 21:25
benabik: multiple returns, yes
benabik NotFound++
NotFound Nested functions with name, no. 21:26
benabik Nested named functions are var name = function(){}; :-D
NotFound benabik: barely named. You don't get a name in a backtrace 21:27
In case someone doesn't know, winxedxx is a winxed backend that compiles a subset of winxed to C++ 21:29
dalek TT #1518 closed by pmichaud++: NQP-rx generates wrong twigil for contextual lookups 21:32
TT #1518: trac.parrot.org/parrot/ticket/1518
TT #1968 closed by pmichaud++: Loading of nqp-setting breaks rakudo.
TT #1968: trac.parrot.org/parrot/ticket/1968
TT #1429 closed by pmichaud++: Add -L support to parrot-nqp
TT #1429: trac.parrot.org/parrot/ticket/1429
TT #161 closed by pmichaud++: abc should not evaluate to 'last' in the abc language
TT #161: trac.parrot.org/parrot/ticket/161
benabik abc is still in the repo. 21:33
21:33 mj41 joined
dalek kudo/nom: 07b9b00 | jnthn++ | src/Perl6/Grammar.pm:
Fix for RT#89208 (oddness when declaring a sub in a for loop that needs to get an implicit $_).
 21:33
21:43 nbrown joined 21:45 Coke joined
benabik How do you get multiple return values out of winxed? 21:47
Oh. I was doing it wrong. Nevermind... 21:48
NotFound benabik: $(a, b) = foo();
dalek TT #1447 closed by pmichaud++: [NQP-rx] failure to capture two duplicate subrules in the same rule
TT #1447: trac.parrot.org/parrot/ticket/1447
whiteknight :(a, b) = foo()
NotFound Ups.
whiteknight or can you also use $?
NotFound I'm just a bit sleepy 21:49
whiteknight it's okay.
benabik Can you int :(a, b, c) = foo() ? 21:51
nope 21:52
Shoot. Hm. How to capture something outputted by PIR code? Blah. 21:56
This is too much work anyway. The test harness isn't broken, I don't know why I'm trying to fix it. :-D 21:57
Tene benabik: "output" == print?
or return?
benabik print
Trying to test the printed output of a bit of code. 21:58
Tene you'll need to set something else to stdout or stderr, unfortunately. It's a bit awkward.
there's a setstdout method on the itnerpreter, I think?
benabik Fun, fun. Maybe I'll look into it. Maybe I'll leave it.
Tene: There is. Interesting.
And a string handle…
Tene yeah, exactly 21:59
benabik Looks about as painful as it should be. :-D
It's a somewhat perverse thing to do. ;-)
Tene yeah, it's not so bad, as I recall, just a bit awkward.
benabik I'm trying to make the abc test harness not create temp files. I'm not sure why other than "it irritated me". :-D 22:01
But I should have left work 30min ago. 22:02
dukeleto github.com/blog/964-all-of-the-hooks 22:09
soh_cah_toa msg fperrad i stumbled upon a bug in parrot-lua where the scope of local variables declared inside a loop doesn't include the loop condition; which it should in lua 5.1. i opened a gh issue at github.com/fperrad/lua/issues/4 22:11
aloha OK. I'll deliver the message.
dalek umage: b4d9ca1 | soh_cah_toa++ | TODO:
Removed item about version numbers from TODO list since it has been completed.
 22:15
umage/soh-cah-toa/feather-format: 74651d3 | soh_cah_toa++ | TODO:
Added items regarding the featherspec to TODO list.
 22:25
soh_cah_toa i would like to add some tests to plumage but it uses parrot-prove and i need to use p5's Test::More. how do i get around this? 22:27
dukeleto soh_cah_toa: what do you need from Test::More? 22:33
soh_cah_toa dukeleto: well it needs to be in p5 b/c i need to use IPC::Open3. i want to test the cli
dalek kudo/nom: 433839e | jnthn++ | src/Perl6/Actions.pm:
Ensure import, use and need return something (Nil for now), so we don't get IMCC errors if you try and use the result of the use statement for something.
 22:35
dukeleto soh_cah_toa: parrot-prove is only the test harness, iirc. It should hopefully obey a shebang line 22:39
soh_cah_toa nah, i don't think so but i'll double check 22:40
or maybe it does ;) 22:44
aw, `parrot setup.pir test` and `parrot-prove` do not traverse subdirectories :( 22:45
dukeleto soh_cah_toa: that is the -r option to prove, which i don't think got implemented 22:46
soh_cah_toa nope
dalek kudo/nom: aa32037 | jnthn++ | src/Perl6/Actions.pm:
Fix (and greatly simplify) handling of 'use fatal' (doubt we actually look at it anywhere yet, but this at least gets us beyond 'use fatal' giving a weird error).
soh_cah_toa damn. while parrot-prove respects the shebang line, `parrot setup.pir test` does not 22:48
gah, so annoying!
dalek kudo/nom: a0c370d | jnthn++ | src/Perl6/Actions.pm:
Get require to work again, though no import lists yet.
 22:59
dukeleto soh_cah_toa: parrot is not a tap harness 23:00
soh_cah_toa: what is parrot setup.pir test using? not parrot-prove ?
soh_cah_toa whatever you set 'prove_exec' to i think 23:02
which for plumage is the value of get_nqp_rx() which i assumed parrot-nqp used parrot-prove 23:03
23:05 Coke joined
dukeleto oy 23:10
23:11 Coke joined
soh_cah_toa i suppose i could just convert all tests to p5. the current one's are kinda crappy anyway and *really* need to be extended; something that'll be easier in p5 23:13
whiteknight tests for what? 23:15
fuuuu 23:17
updating green_threads to master creates conflicts and heartaches
soh_cah_toa whiteknight: plumage 23:18
whiteknight oh yes, plumage does need more tests
soh_cah_toa yeah, *a lot* 23:19
23:19 Coke joined
whiteknight coughs something about rosella 23:19
soh_cah_toa well, i need to use IPC::Open3 for some other tests so i kinds need to be p5 23:20
plus, rosella is for losers :P 23:21
whiteknight bleh 23:22
I really need to add 3-way pipes to Parrot 23:23
dukeleto i was the last person to write tests for plumage
whiteknight Infinoid was like 95% of the way there, but we lost it
dukeleto i did all that i could with what was available, but ran into similar issues as soh_cah_toa is running into
soh_cah_toa yeah, nqp isn't really great for...well..anything besides rakudo but especially testing 23:24
23:25 Coke joined
whiteknight Most of Rosella's own test suite is written in NQP, and it's always worked very well 23:43
in fact, similar tests are typically less verbose in NQP than they would be in winxed 23:44
...too much email... 23:50
soh_cah_toa eh, i really don't like nqp. especially parrot-nqp : 23:51
dalek Heuristic branch merge: pushed 45 commits to parrot/green_threads by Whiteknight 23:56
whiteknight msg cotto: I want to push forward with a green_threads merge. I'm waiting on nine to make sure he doesn't have any outstanding todos. Any objections?
aloha OK. I'll deliver the message.
dalek parrot: a2422a7 | gerd++ | plannodes.h:
add dirty fix for struct "Hash" name conflict
 23:59
parrot: 49723ae | gerd++ | plparrot.c:
add dirty fix for struct "Hash" name conflict

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!