Perl 6 - the future is here, just unevenly distributed

IRC log for #perl6-dev, 2017-04-22

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:45 bisectable6 joined #perl6-dev
01:13 astj joined #perl6-dev
01:27 Zoffix m: say "abc" ~~ m:i/caaaaa/
01:27 camelia rakudo-moar 9ed89d: OUTPUT: «「c」␤»
01:28 Zoffix pretty scary we don't have any tests catching something as basic as that :o
01:30 samcv timotimo, need a bit of help fixing this bug
01:30 samcv m: say "b" ~~ m:i/ba/
01:30 camelia rakudo-moar 9ed89d: OUTPUT: «「b」␤»
01:31 samcv only on m:i and the last letter of left string batches the first of the right
01:31 samcv m: say "aaaaaaaab" ~~ m:i/ba/
01:31 camelia rakudo-moar 9ed89d: OUTPUT: «「b」␤»
01:32 samcv and it doesn't happen with variables but only if it's actually written in the source code or EVAL'd
01:32 samcv m: say "aaaaaaaaaaaaaaaaab" ~~ m:i/abaa/
01:32 camelia rakudo-moar 9ed89d: OUTPUT: «「ab」␤»
01:33 samcv m: say "aaaaaaaabcd" ~~ m:i/abcd111111/
01:33 camelia rakudo-moar 9ed89d: OUTPUT: «「abcd」␤»
01:33 samcv m: say "aaaaaaaabcd" ~~ m:i/.abcd111111/
01:33 camelia rakudo-moar 9ed89d: OUTPUT: «「aabcd」␤»
01:33 samcv m: say "aaaaaaaabcd" ~~ m:i/3abcd111111/
01:33 camelia rakudo-moar 9ed89d: OUTPUT: «False␤»
01:34 samcv hmm actually it's a problem in the moarvm function. ok
01:34 samcv i know how to fix this then
01:35 samcv probably....
01:38 Ben_Goldberg joined #perl6-dev
01:46 Ben_Goldberg joined #perl6-dev
01:49 ilbot3 joined #perl6-dev
01:49 Topic for #perl6-dev is now Perl 6 language and compiler development 2.0 | Logs at http://irclog.perlgeek.de/perl6-dev/today
02:10 Zoffix Maybe we should knock it on the head and make Rakudo Star from 2017.05 instead of .04
03:32 BenGoldberg joined #perl6-dev
03:35 sasheto joined #perl6-dev
03:38 sasheto joined #perl6-dev
04:35 astj joined #perl6-dev
04:51 Geth ¦ roast: 5401f288be | (Samantha McVey)++ | S05-modifier/ignorecase.t
04:51 Geth ¦ roast: Add tests for m:i bug
04:51 Geth ¦ roast:
04:51 Geth ¦ roast: If the beginning of the needle matches towards the end of the haystack,
04:51 Geth ¦ roast: it can return a partial match, when it hasn't traversed the needle fully.
04:51 Geth ¦ roast: Add tests for this bug.
04:51 Geth ¦ roast: review: https://github.com/perl6/roast/commit/5401f288be
04:51 samcv ok tests added. i have them passing in my build of mvm. will commit it in a bit
05:37 Geth ¦ nqp: ae6a610d84 | (Samantha McVey)++ | 2 files
05:37 Geth ¦ nqp: Add tests for MVM indexic/equatic bug
05:37 Geth ¦ nqp:
05:37 Geth ¦ nqp: Tests are todo until fix pushed to MVM.
05:37 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/ae6a610d84
06:41 [Tux] This is Rakudo version 2017.04.2-22-g9ed89d947 built on MoarVM version 2017.04
06:41 [Tux] csv-ip5xs        3.060
06:41 [Tux] test            12.986
06:41 [Tux] test-t           5.194 - 5.237
06:41 [Tux] csv-parser      13.342
06:48 Geth ¦ nqp: 856d5c51a5 | (Samantha McVey)++ | tools/lib/NQP/Configure.pm
06:48 Geth ¦ nqp: Enhance error reporting in Configure.pm when rev parse errors
06:48 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/856d5c51a5
07:03 stmuk messages?
07:03 yoleaux 19 Apr 2017 18:31Z <Zoffix> stmuk: does R* have some sort of mechanism to apply patches or use non-release commit for release or something? The b4fa6d6792 fixes a REPL bug that affects REPL users on OSX who load Readline or Linenoise.
07:03 yoleaux 19 Apr 2017 22:38Z <Zoffix> stmuk: never mind. It's a much more minor issue than I originally thought. The history file just isn't saved between REPL runs and it displays a message about mkdir '/' 0o777 at startup. I also got a workaround that involves setting up an alias the users who really want their command line history: https://gist.github.com/zoffixznet/6a0d212ada41fce7aeca15fa9efcc263
07:13 samcv ok i'm going crazy someone please help
07:14 samcv see https://travis-ci.org/MoarVM/MoarVM/jobs/224609074#L1062
07:14 samcv i have no clue where this is coming from
07:14 samcv tag 9999-please-do-not-use-these-7-gde3e5cf is bogus. but it seems to be trying to check it?
07:14 samcv but it builds fine even on a clean checkout for me
07:15 samcv hmm maybe i should clone more than depth 1 for nqp
07:15 stmuk .tell zoffix https://github.com/rakudo/star/tree/master/patches
07:15 yoleaux stmuk: I'll pass your message to zoffix.
07:16 samcv oh wait. nvm. i just pulled tags on mvm
07:16 samcv and someone added a tag 9999-please-do-not-usethese
07:16 samcv GREAT! and tags are not deletable!
07:20 nine WTF?
07:20 samcv i think we can probably ignore it though
07:20 samcv hopefuly
07:21 samcv rev =~ /(\d+)$sep(\d+)(?:$sep(\d+))?(?:-(\d+)-g[a-f0-9]*)?$/
07:21 samcv or die "Unrecognized revision specifier '$rev' $what\n";
07:21 samcv is what we have in Configure.pm now
07:23 samcv ok i'm getting the failure too now. i wasn't getting it on a clean install, but installed mvm to a branch new prefix and i hit the bug
07:26 stmuk .tell zoffix BTW I'm on holiday in Greece with poor connectivity so R* is likely to be delayed. I had problems with Grammar::Debugger but that's now fixed
07:26 yoleaux stmuk: I'll pass your message to zoffix.
07:26 samcv working on it right now...
07:29 nine samcv: we could also change MoarVM's Configure.pl to use git describe instead of git describe --tags
07:30 samcv oh
07:30 samcv actually you're right
07:30 samcv i'm looking in the wrong spot :)
07:30 samcv i thought it was checking the moarvm repo tags or something
07:30 nine Yeah, because nqp's Configure.pl uses moarvm --version
07:30 nine err... moar --version
07:32 nine pushed
07:33 samcv yay
07:34 samcv weird that that tag appeared
07:35 samcv maybe timo accidently merged it in with telemeh
07:35 samcv seems to work now
07:36 samcv thanks nine
07:36 nine my pleasure :)
07:37 samcv now i can stop tearing my hair out
07:38 nine Yeah, first Travis builds passing
07:39 samcv i pushed it to my fork so i can check appveyor too
07:50 nine 0.16s$ perl Configure.pl --prefix=$MOAR_PREFIX --backends=moar;
07:50 nine Unknown option: backends
07:50 nine WTF?
07:50 nine https://travis-ci.org/MoarVM/MoarVM/jobs/224612634
07:51 samcv weird. idk why that's happening maybe the uh
07:52 samcv readlink -f command words differently on mac?
07:52 samcv i replaced `cd nqp` with `cd $NQP_FOLDER` which is done from readlink -f nqp # while in the moarvm folder
07:54 samcv hmm https://stackoverflow.com/questions/1055671/how-can-i-get-the-behavior-of-gnus-readlink-f-on-a-mac
07:54 samcv lame-o
08:01 samcv fixing now
08:19 samcv there we go
08:20 samcv that was more annoying that i thought it'd be at first. settled on
08:20 samcv writing my own: resolve_folder      () { O="$(pwd -P)"; D="$(dirname $1)"; T="$(basename $1)"; cd -- "$D" && cd -- "$T" && pwd -P; cd -- "$O" > /dev/null; }
08:20 samcv since i don't need to resolve links or anything, and that should work for any normal folders
08:22 samcv don't think i need quotes around VAR="$( )" though. yeah. i've learned if in doubt just quote everything :)
08:33 samcv appears to be working now. good
08:53 robertle joined #perl6-dev
08:54 samcv still gotta fix the windows build though
09:49 bartolin lizmat: it looks like 3b0dec3255 does not work with the JVM backend :-/
09:49 bartolin during stage parse it blows up with: Method 'X::TypeCheck::Binding::Parameter' not found for invocant of class 'QAST::SVal'
10:01 pmurias joined #perl6-dev
10:26 NeuralAnomaly joined #perl6-dev
10:26 NeuralAnomaly_ joined #perl6-dev
10:27 lizmat bartolin: does removing the BEGIN fix it?   If so, maybe we should remove the BEGIN in case of the JVM ?
10:29 Zoffix .tell stmuk R* likely delayed you say... I'm thinking of maybe delaying it by 4 weeks and cutting it off 2017.05 or something isn't a bad idea. There are two pretty serious bugs in 2017.04.2 that we'd need to patch or point-release for
10:29 yoleaux 07:15Z <stmuk> Zoffix: https://github.com/rakudo/star/tree/master/patches
10:29 yoleaux Zoffix: I'll pass your message to stmuk.
10:29 yoleaux 07:26Z <stmuk> Zoffix: BTW I'm on holiday in Greece with poor connectivity so R* is likely to be delayed. I had problems with Grammar::Debugger but that's now fixed
10:30 bartolin lizmat: I'll test (will take a while :-)
10:31 Geth joined #perl6-dev
10:40 bartolin lizmat: looks good so far. I'm running a spectest and if that looks good as well, I'll add a special case for r-j (for the line with BEGIN only)
10:40 lizmat yeah, that's what I had in mind  :-)
10:40 lizmat bartolin++
10:41 lizmat afk again&
10:54 jnthn Aha, my TPF grant report is posted: http://news.perlfoundation.org/2017/04/perl-6-performance-and-reliabi-3.html
10:54 yoleaux 21 Apr 2017 23:27Z <AlexDaniel> jnthn: fwiw https://irclog.perlgeek.de/perl6/2017-04-21#i_14466196
10:55 jnthn For anyone reading it and thinking "huh, did jnthn do those recently" - no, I wrote the grant report 2 months ago, it just took until now to be posted. :-)
10:58 Zoffix jnthn++
10:58 Zoffix Reliability definitely improved.
11:05 jnthn Glad it's felt by others :)
12:05 MasterDuke_ joined #perl6-dev
12:29 AlexDaniel joined #perl6-dev
12:58 dogbert17 o/
12:59 dogbert17 Zoffix: my apologies for disappearing during yesterdays TAP.pm6 debug session, someone at $work suggested that some beers should be consumed at a local pub and I fell for it
13:01 dogbert17 I have now run the spectest through gdb and here's what it looks like: https://gist.github.com/dogbert17/74e00499cffeb9c1117b5ef6d9769308
13:07 dogbert17 for those who didn't follow the discussion yesterday, the problem is that, if you run 'make spectest HARNESS_TYPE=6' with the default TEST_JOBS=1, then the process will run out of file handles if your system is set up to only have 1024
13:08 dogbert17 it seems as if one fd is lost for each test file being run and when you get to file ~1000 the spectest is aborted
13:09 dogbert17 hopefully the gist can give us a clue as to why
14:33 ZoffixMobile joined #perl6-dev
14:33 ZoffixMobile dog
14:34 ZoffixMobile Ugh
14:34 ZoffixMobile dogbert17: what perl6 version is that gdb for?
14:36 dogbert17 dogbert@dogbert-VirtualBox ~/repos/rakudo $ gdb -v
14:36 dogbert17 GNU gdb (Ubuntu 7.7.1-0ubuntu5~14.04.2) 7.7.1
14:37 dogbert17 the trick here is to break into the running perl6 process with 'sudo gdb -p <pid>', then you'll be able to set breakpoints an stuff
14:38 timotimo i think he meant perl6 --version
14:38 dogbert17 This is Rakudo version 2017.04.2-22-g9ed89d947 built on MoarVM version 2017.04
14:38 timotimo there's a config option somewhere in /sys or /proc that you can change so you can attach to your own processes
14:38 timotimo so you don't need to run gdb as root
14:38 dogbert17 aha
14:40 timotimo echo 0 > /proc/sys/kernel/yama/ptrace_scope
14:42 dogbert17 cool
14:45 dogbert17 https://github.com/rakudo/rakudo/blob/nom/lib/TAP.pm6#L885
14:47 timotimo nobody ever awaits the closing of the process?
14:47 timotimo like, $proc.out.lines(:close) or whatsit?
14:47 * jnthn has a headache today, and figured he's share it by writing a blog post... https://6guts.wordpress.com/2017/04/22/massively-reducing-moarvm-fixed-size-allocator-contention/
14:47 jnthn *he'd
14:48 lizmat jnthn++
14:49 lizmat jnthn: also: s/dontate/donate/
14:49 * dogbert17 finds the code in TAP.pm6 quite hard to follow even without a headache
14:49 timotimo dotonate?
14:49 dogbert17 jnthn++ (blogging)
14:49 * lizmat goes off and will read more when back
14:50 timotimo "mallocand friends"
14:56 Geth ¦ rakudo/nom: 624fe8fa2b | usev6++ | src/core/Exception.pm
14:56 Geth ¦ rakudo/nom: [JVM] Don't initialize hash at compile time
14:56 Geth ¦ rakudo/nom:
14:56 Geth ¦ rakudo/nom: The BEGIN phaser was added with 3b0dec3255, but doesn't work on JVM. It lead to
14:56 Geth ¦ rakudo/nom: an error during stage past: Method 'X::TypeCheck::Binding::Parameter' not found
14:56 Geth ¦ rakudo/nom: for invocant of class 'QAST::SVal'.
14:56 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/624fe8fa2b
15:00 timotimo good post
15:01 jnthn lizmat, timotimo: fixed those two, thanks
15:10 MasterDuke_ jnthn++, but "VM for a dyanmic language"
15:13 * dogbert17 has trouble understanding "A given bin consists of:"
15:13 dogbert17 1 or more pages, each with space for a certain number of allocations of the bin size
15:14 MasterDuke_ also, "So, I belive"
15:18 timotimo dogbert17: well, there's one FSA, in it are multiple "size bins", where each bin holds objects of an upper and lower size limit
15:18 timotimo each of these "size bins" allocates pages full of slots of these sizes, i.e. it'll malloc 128 slots * 16 bytes
15:21 dogbert17 timotimo: thx
15:23 astj joined #perl6-dev
15:40 dogbert17 "When a thread termiantes,"
15:52 astj joined #perl6-dev
15:54 Geth ¦ nqp: MasterDuke17++ created pull request #354: Make profile sql output better
15:54 Geth ¦ nqp: review: https://github.com/perl6/nqp/pull/354
15:56 MasterDuke_ timotimo, nine: if you care to check it out, https://github.com/perl6/nqp/pull/354 should enable the functionality you were asking for
15:57 timotimo sorry, i'm a half AFK
16:02 MasterDuke_ timotimo: i'm sorry, but this PR is going to expire in 10min unless you devote yourself to reviewing it. all the blame will rest on you
16:03 MasterDuke_ that's a reasonable expectation, right?
16:04 timotimo :)
16:05 timotimo yeah, i'll merge that
16:06 timotimo not quite a thorough code review, but since you've tested it and things give proper results, it's an improvement either way :)
16:06 MasterDuke_ well, might be good to have someone else test too
16:07 Geth ¦ nqp/master: 5 commits pushed by MasterDuke17++, timo++
16:07 Geth ¦ nqp/master: 9ab704a7c0 | Use 0 instead of NULL in the profiler SQL output
16:07 Geth ¦ nqp/master: d8949b4716 | Rename 'allocators' to 'types' in sql profile
16:07 Geth ¦ nqp/master: ec195684d2 | Make 0s strings since they're going to get stringified anyway
16:07 Geth ¦ nqp/master: 95b8926e6e | Allow tracking of allocations by routine
16:07 Geth ¦ nqp/master: f2cfdb617f | Merge pull request #354 from MasterDuke17/make_profile_sql_output_better
16:07 Geth ¦ nqp/master: review: https://github.com/perl6/nqp/compare/856d5c51a5...f2cfdb617f
16:07 timotimo pff, testing
16:07 MasterDuke_ heh, but then again, i suspect i'm the only one who's ever used the sql profile output
16:07 timotimo i've just re-learned today that just committing stuff to master is totally fine
16:07 MasterDuke_ so even if it broke...
16:08 timotimo i've used it as if it were csv :P
16:08 timotimo i.e. not taken advantage of the structure and inter-table relations at all
16:08 timotimo or even joining a table with itself, as would be required with the call graph table
16:08 MasterDuke_ ah, still nice to know someone has used it
16:09 MasterDuke_ need to remember to correct the example in the docs when that hits rakudo
16:09 timotimo and now we should be in good shape to build a good tool for this
16:09 timotimo i didn't know this was docced :P
16:10 MasterDuke_ barely, https://docs.perl6.org/language/performance#Profile_locally
16:11 timotimo i didn't even know!
16:11 timotimo is sqlite still in browsers for free?
16:11 MasterDuke_ but now you do, and that's half the battle
16:12 MasterDuke_ i think ff still uses it for some internal stuff, don't know otherwise
16:13 timotimo it might be time to remove "use of gather/take" from the performance page as a "performance weakness"
16:13 timotimo hmm.
16:14 timotimo if sqlite were still in browsers for free, we could host a static .html file on moarvm.org or rakudo.org or whatever and people  would just pick their .sql file from disk and get all the data without it traveling over the network
16:14 MasterDuke_ that would be pretty cool
16:14 timotimo of course there's stuff like https://github.com/kripken/sql.js/
16:15 timotimo i wonder why its build is failing?
16:16 timotimo there is nothing in the travis build logs for the latest failed build
16:17 timotimo could it be it fell prey to the travis failure a few weeks back?
16:17 MasterDuke_ builds fine locally and passes `make m-test`
16:17 timotimo m-test? how come it has a moarvm backend?
16:18 MasterDuke_ i just specify it because sometimes i have a stray old jvm build lying around
16:18 timotimo i was refering to sql.js' build
16:19 MasterDuke_ ah, nm then
16:20 timotimo though you could just use the sql.js it offers and at least the demo page works
16:24 timotimo Clone done, sent: 1430  received: 25072049  ip: 67.18.92.124
16:24 timotimo server returned an error - clone aborted
16:24 timotimo ?!? :(
16:27 timotimo could be because my version of fossil is ancient :P
16:28 MasterDuke_ fossil?
16:28 timotimo the SCM made and used by the sqlite folks
16:29 MasterDuke_ huh, never used it
16:30 timotimo it's slightly niche, but i hear great things about it
16:35 * timotimo grabs an emscripten sdk
16:48 timotimo ugh, the makefile for sql.js is bogus
16:52 timotimo huh
16:52 timotimo how in the heck does this work
16:53 timotimo how does it know to put emcc in front ...
16:54 timotimo and now i have an up-to-date sql.js
16:58 timotimo now, do i even want to write this in html and js at all?
16:58 timotimo i didn't even consider that question :)
16:59 timotimo i have experience with python and qt, maybe i'll give that a shot, though maybe through Inline::Python rather than python directly
17:00 MasterDuke_ could the existing HTML be adapted to optionally the sql output?
17:00 timotimo probably
17:07 timotimo s: {a => 1}, "gist", \()
17:07 SourceBaby timotimo, Ehhh... I'm too scared to run that code.
17:08 timotimo s: Hash.new(), "gist", \()
17:08 SourceBaby timotimo, Sauce is at https://github.com/rakudo/rakudo/blob/624fe8f/src/core/Hash.pm#L213
17:22 * samcv yawns-
17:22 samcv too early
17:23 samcv timotimo, if telemeh is not messing up the windows builds it's okay to bump nqp to get the m:i fix
17:24 timotimo well, i'm not saying it doesn't mess up the windows builds, but it does pass on appveyor
17:25 samcv ..
17:25 samcv timotimo, mess up how?
17:25 timotimo i have no idea
17:26 samcv then how do you know it messes it up
17:26 timotimo it's windows, you know
17:26 timotimo and if it's windows ... you *know*.
17:26 samcv spectest passing
17:26 samcv ok so you don't *know* but guess?
17:26 timotimo well, i know that being too confident is a bad idea when windows' involved :D
17:27 samcv yeah i guess
17:27 Zoffix \o/
17:27 Zoffix Just got re-certified as Emergency Level C First Aid Responder :)
17:27 samcv well i'l be back in a few hours, going back to sleep for a bit. bump moar+nqp whenever you feel comfortable
17:27 timotimo cool
17:27 timotimo have a good sleep, samcv
17:28 samcv thank you :)
17:28 Zoffix Basically keeping you alive in any emergency long enough for ambulance to arrive... So... if you feel you're about to die. Ping me :P
17:28 samcv i feel i'm about to lose consiousness
17:28 timotimo oh
17:28 samcv zzzzzzzzzz
17:28 * samcv &
17:28 timotimo i thought you meant C Programming Language First Aid Responder
17:28 Zoffix night
17:28 Zoffix heh
17:28 samcv XD lol timotimo
17:29 samcv we could use one of those too
17:29 timotimo yeah, i seem to find myself in need of that multiple times a month
17:51 timotimo YES! i have a terminal with sixel support \o/
18:10 BenGoldberg joined #perl6-dev
18:46 timotimo i think i'm going to bump nqp and rakudo now
18:47 timotimo huh
18:50 timotimo i have no idea by what mechanism my 9999-please-dont-use-these release causes Configure.pl to break in nqp
18:52 timotimo samcv: i'm going to remove what looks like left-over debug output, "long end 0 should-fail 0"
18:58 TimToady Zoffix: the problem with putting off R* even longer is that, the longer the cycle time on releases, the more pressure there is to sneak things in at the last moment because you know you won't get another chance for N months
18:58 TimToady so we should probably figure out how to move to a more continuous release process
18:59 * TimToady is basically channeling obra++ here...
19:02 b2gills That was a big problem for Perl 5 until they went with a yearly release cycle
19:07 Geth ¦ nqp: f72fe4d04d | (Timo Paulssen)++ | 2 files
19:07 Geth ¦ nqp: remove left-over debug output, bump moarvm for eqatic fix
19:07 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/f72fe4d04d
19:07 Geth ¦ rakudo/nom: 7a3a6a8469 | (Timo Paulssen)++ | tools/build/NQP_REVISION
19:07 Geth ¦ rakudo/nom: bump nqp for eqatic fix
19:07 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/7a3a6a8469
19:07 Geth ¦ rakudo/nom: version bump brought these changes: https://github.com/perl6/nqp/compare/2017.04-12-ga5337ff...2017.04-20-gf72fe4d
19:09 Zoffix TimToady: seems apples and oranges to me. All I proposed is to base the next R* off 2017.05 Rakudo instead of making a 2017.04.3 point release or adding rakudo patches to R*. This doesn't affect the length of between future R* releases.
19:11 Zoffix As for continuous releases... what is it? Releases more often than 1 month? Perl 6 VIP will be an end user release made monthyl, same day as Rakudo
19:11 Geth ¦ rakudo/nom: 2bae61d69b | (Timo Paulssen)++ | lib/TAP.pm6
19:11 Geth ¦ rakudo/nom: sync version of process-launching tap harness must close stdout.
19:11 Geth ¦ rakudo/nom:
19:11 Geth ¦ rakudo/nom: otherwise FDs will keep piling up and exhaust the system's ulimit
19:11 Geth ¦ rakudo/nom: of file descriptors. dogbert17++
19:11 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/2bae61d69b
19:14 TimToady no biggie, and I'm not a big fan of slippery slope arguments, but "Just This Once" tickles that neuron of mine :)
19:15 Zoffix I can always cut a new point release, if needed. Let me know
19:16 * Zoffix &
19:29 dogbert17 oh, timotimo++ (bug crusher)
19:29 timotimo well, you pointed me at the exact location
19:31 dogbert17 I wouldn't have known what to do though :)
19:32 samcv yay got the merge
19:32 samcv err bump
19:32 samcv good *
19:33 timotimo was the debug output removal a good call?
19:33 samcv yeah that's fine
19:33 timotimo 'k
19:33 samcv from the .t file right
19:33 timotimo we should then un-todo some of these tests
19:33 samcv very yes
19:33 timotimo hm, is that in the roast or in nqp's test suite?
19:33 samcv both
19:34 timotimo t/moar/08-indexic.t                  (Wstat: 0 Tests: 117 Failed: 0)
19:34 timotimo TODO passed:   1, 92-105
19:34 timotimo yeah
19:34 samcv plus in roast too
19:34 timotimo i should have untodod that before bumping the version from rakudo's side
19:35 samcv yea
19:35 samcv m: say 'abcd' ~~ /cdfg/
19:35 camelia rakudo-moar 2bae61: OUTPUT: «Nil␤»
19:35 samcv yay
19:35 timotimo there's a terminal emulator called "cancer" with the tagline "it's terminal" m)
19:36 samcv oh my
19:39 geekosaur oy
19:40 timotimo oy geekosaur
19:41 Geth ¦ nqp: cb60c0b32c | (Samantha McVey)++ | tools/lib/NQP/Configure.pm
19:41 Geth ¦ nqp: Nicer debug output when git tags are confused
19:41 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/cb60c0b32c
19:41 Geth ¦ nqp: 33aa810320 | (Samantha McVey)++ | t/moar/08-indexic.t
19:41 Geth ¦ nqp: Untodo the now passing indexic tests
19:41 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/33aa810320
19:42 samcv looks like equatic still has the bug though
19:42 samcv hmm
19:42 timotimo oh crap
19:43 samcv idk if we use equatic anywhere?
19:43 timotimo oh
19:43 timotimo i think my commit messages were all talking about eqatic
19:44 timotimo but only indexatic was fixed?
19:44 samcv yeah
19:44 timotimo dang
19:44 samcv yeah we don't use equatic anywhere
19:44 timotimo d'oh
19:44 samcv gonna look at it in mvm though. i though what i did would have fixed it but. lemme see
19:45 samcv still not a release blocker though since it's not used anywhere
19:45 samcv ok we use it in one place
19:46 samcv m: say 'a' ~~ m:i/ab/
19:46 camelia rakudo-moar 2bae61: OUTPUT: «「a」␤»
19:46 samcv ok yeha need to fix this then
19:47 samcv cause it does equatic first before doing an index i believe
19:47 samcv at the very start of the haystack
19:47 samcv m: say 'ba' ~~ m:i/ab/
19:47 camelia rakudo-moar 2bae61: OUTPUT: «False␤»
19:49 samcv ok this is EZ fix
19:56 samcv yay i fixed it. will commit after i grab my tea
20:00 samcv m: use nqp; say nqp::eqatic('st', 'sta', 0)
20:00 camelia rakudo-moar 2bae61: OUTPUT: «1␤»
20:01 samcv m: use nqp; say nqp::eqatic('st', 'stassssssss', 0)
20:01 camelia rakudo-moar 2bae61: OUTPUT: «1␤»
20:02 samcv m: use nqp; say nqp::eqatic('st', 'stassssssss', 0)
20:02 camelia rakudo-moar 2bae61: OUTPUT: «1␤»
20:02 samcv gonna add a few more eqatic tests to nqp
20:02 samcv to cover this bug better
20:15 pmurias (adding tests to nqp)++
20:34 samcv ok added a few more tests, got some failures and further improved moar
20:34 samcv after sitting down and drawing a ton of strings
20:42 astj joined #perl6-dev
20:43 astj joined #perl6-dev
20:44 samcv running roast now
20:51 Geth ¦ nqp: 6a482c6ac9 | (Samantha McVey)++ | tools/build/MOAR_REVISION
20:51 Geth ¦ nqp: Bump MoarVM for indexic and equatic fixes
20:51 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/6a482c6ac9
20:51 Geth ¦ nqp: version bump brought these changes: https://github.com/MoarVM/MoarVM/compare/2017.04-43-g43beb32...2017.04-44-gf0db8822
20:51 Geth ¦ nqp: 87501f7b08 | (Samantha McVey)++ | 2 files
20:51 Geth ¦ nqp: Untodo now passing equatic and indexic tests
20:51 Geth ¦ nqp: review: https://github.com/perl6/nqp/commit/87501f7b08
20:51 Geth ¦ nqp: version bump brought these changes: https://github.com/MoarVM/MoarVM/compare/2017.04-43-g43beb32...2017.04-44-gf0db8822
20:51 samcv we are a gooooo
20:55 robertle gooey?
20:56 Geth ¦ rakudo/nom: ce41e46602 | (Samantha McVey)++ | tools/build/NQP_REVISION
20:56 Geth ¦ rakudo/nom: Bump MoarVM/NQP for case insensitive regex fixes
20:56 Geth ¦ rakudo/nom: review: https://github.com/rakudo/rakudo/commit/ce41e46602
20:56 Geth ¦ rakudo/nom: version bump brought these changes: https://github.com/perl6/nqp/compare/2017.04-20-gf72fe4d...2017.04-24-g87501f7b
20:56 samcv we are a GO
20:58 Geth ¦ roast: 7c72cadfcf | (Samantha McVey)++ | S05-modifier/ignorecase.t
20:58 Geth ¦ roast: Untodo now passing m:i tests
20:58 Geth ¦ roast: review: https://github.com/perl6/roast/commit/7c72cadfcf
21:01 samcv yay!
21:01 samcv 100% pass for equatic indexic and the m:i regex tests as well
21:19 ilmari joined #perl6-dev
22:13 samcv Zoffix, let me know when you're around
22:13 samcv and we can discuss another point release
22:21 AlexDaniel point release? 2017.04.3? \o/
22:22 samcv idk.
22:22 samcv yeah
22:22 samcv anything
22:23 lizmat I got the impression we're going to not do any more point releases and have a Rakudo Star on 2017.05 ?
22:23 samcv hmm
22:23 samcv idk the regex bug is bad
22:23 samcv idk why were the only releases made
22:23 samcv <samcv> m: say 'a' ~~ m:i/ab/
22:23 samcv <camelia> rakudo-moar 2bae61: OUTPUT: «「a」␤»
22:23 samcv ^
22:23 samcv fixed now though
22:24 AlexDaniel c: releases say 'a' ~~ m:i/ab/
22:24 committable6 AlexDaniel, ¦2015.12,2016.01.1,2016.02,2016.03,2016.04,2016.05,2016.06,2016.07.1,2016.08.1,2016.09,2016.10,2016.11,2016.12,2017.01,2017.02,HEAD(ce41e46): «False» ¦2017.03,2017.04.2: «「a」»
22:24 ggoebel joined #perl6-dev
22:24 samcv oh nice AlexDaniel
22:25 AlexDaniel I'd say if it was there for more than two releases, it can wait for the next one
22:25 samcv hm
22:25 samcv what were the other point releases for?
22:25 AlexDaniel not only we had the first *.2 release this time, you want it to be the first *.3 release also? :)
22:26 samcv i think the bug is severe enough to make another point release. suprised it wasn't found but
22:26 samcv it could cause... issues
22:27 AlexDaniel yeah, yeah. Like any other bug actually
22:33 jnthn At this rate we'll be doing point releases up until the next pointless release. :P
22:35 samcv well. i mean not releasing a ponit release to fix a serious bug. seems not a good idea either
22:35 samcv just because "we've had too many point releases"
22:36 timotimo every body wants a poni releaset
22:36 AlexDaniel bisect: o=2017.03 say 'a' ~~ m:i/ab/
22:36 bisectable6 AlexDaniel, On both starting points (old=2015.12 new=ce41e46) the exit code is 1 and the output is identical as well
22:36 bisectable6 AlexDaniel, https://gist.github.com/d362cb6c0d6bdecbabe09bdccb378c40
22:36 AlexDaniel meh…
22:37 AlexDaniel not even sure how to use this bug
22:37 samcv how to use it?
22:37 AlexDaniel samcv: yea, bisectable is using :i in this case
22:37 samcv that code does trigger the bug
22:37 AlexDaniel I was trying to trigger it with o=2017.03
22:37 AlexDaniel (should be old=2017.03)
22:37 AlexDaniel but apparently that's not it
22:38 samcv oh it's new in 04 i think
22:38 samcv bisect: o=2017.04 say 'a' ~~ m:i/ab/
22:38 bisectable6 samcv, On both starting points (old=2015.12 new=ce41e46) the exit code is 1 and the output is identical as well
22:38 bisectable6 samcv, https://gist.github.com/09c67862f09501b15f51068b57805734
22:39 jnthn I guess the counter-argument is that we fix things you could describe as serious in every release.
22:39 AlexDaniel samcv: no-no. Bisectable *itself* is running on a slightly older version
22:39 samcv oh
22:39 samcv jnthn, well i think regressions are different than just normal fixes
22:39 jnthn I mean, most releases I've fixed some SEGV somewhere, even if by now they're getting harder and harder to hit upon
22:39 timotimo jnthn: i could benefit from you trying to compile a moarvm on your msvc; i've done a lot to please appveyor, but i can't know if it's enough until a Real Windows is let loose upon it
22:39 samcv point releases are useful for fixing regressions
22:40 AlexDaniel samcv: sometimes I hunt for regressinos using whateverable. There were times I submitted over 20 reports I think
22:40 AlexDaniel and some of them were serious enough
22:40 samcv as serious as this?
22:41 MasterDuke_ what's the syntax for creating variable names with spaces in them? `my ::('a b') = 4` isn't it...
22:41 samcv NeuralAnomaly, status
22:41 NeuralAnomaly samcv, [✘] Next release will be in 3 weeks and 6 days. Since last release, there are 7 new still-open tickets (5 unreviewed and 0 blockers) and 43 unreviewed commits. See http://perl6.fail/release/stats for details
22:41 samcv 3 weeks is too long...
22:41 samcv that's basically a month
22:41 timotimo are you sure a space is allowed with the special syntax?
22:42 timotimo m: my ::('$a b') = 4
22:42 camelia rakudo-moar ce41e4: OUTPUT: «5===SORRY!5=== Error while compiling <tmp>␤Malformed my␤at <tmp>:1␤------> 3my ::('$a b')7⏏5 = 4␤»
22:42 timotimo m: my ::('isthisit') = 4
22:42 camelia rakudo-moar ce41e4: OUTPUT: «5===SORRY!5=== Error while compiling <tmp>␤Malformed my␤at <tmp>:1␤------> 3my ::('isthisit')7⏏5 = 4␤»
22:42 AlexDaniel c: 2016.04,HEAD say (-∞^..^∞).in-range: 0/0
22:42 committable6 AlexDaniel, ¦2016.04: «True» ¦HEAD(ce41e46): «Value out of range. Is: <0/0>, should be in -Inf^..^Inf␤  in block <unit> at /tmp/f0scnurpLz line 1␤ «exit code = 1»»
22:42 MasterDuke_ pretty sure i saw TimToady do it
22:42 timotimo mhm
22:44 AlexDaniel c: 2016.07~50,HEAD say ().rotor(1)
22:44 committable6 AlexDaniel, ¦2016.07~50: «<anon|122856224>.new» ¦HEAD(ce41e46): «()»
22:44 timotimo whateverable helped fix a bunchton of json::fast stuff
22:44 AlexDaniel samcv: I dunno, some of these things don't look very good at all
22:44 samcv hm?
22:45 samcv oh you saying bugs in releases?
22:45 samcv were they regressions though?
22:45 AlexDaniel samcv: (re “as serious as this?”)
22:45 AlexDaniel samcv: yes
22:45 samcv i think more people use regeex than rotor though
22:46 samcv that's why i think it's more serious becuase perl is known for regex...
22:46 AlexDaniel but it is only with :i, and only relevant in some particular cases
22:46 samcv well.
22:46 samcv there's tons of cases it can happen.
22:47 samcv aaaaabc ~~ m:i/cd/ # also will fail
22:47 samcv if the first character and last character match
22:47 AlexDaniel c: 2017.03,HEAD say ‘aaaaabc’ ~~ m:i/cd/
22:47 committable6 AlexDaniel, ¦2017.03,HEAD(ce41e46): «False»
22:47 samcv first char of the needle and last of the haystack
22:48 samcv c:  2017.04,HEAD say ‘aaaaabc’ ~~ m:i/cd/
22:48 committable6 samcv, ¦2017.04: «「c」» ¦HEAD(ce41e46): «False»
22:48 samcv soooo
22:48 samcv that's not that specific
22:48 samcv i mean somewhat specific. but uh. still too common.
22:50 timotimo we need to figure out fuzzing.
22:50 timotimo at a perl6 level, that is
22:51 AlexDaniel timotimo: well, I kinda have it figured out, in a way
22:52 AlexDaniel timotimo: back in December/January I took all m: queries from the log and executed them on 2015.12, back then, HEAD
22:53 AlexDaniel and this gave thousands of results that pointed out that everything has changed…
22:53 AlexDaniel but that's how I found all these regressions
22:53 samcv nice
22:53 timotimo oooh
22:54 AlexDaniel that's a lot of manual work, but I have some plans on optimizing it
22:57 AlexDaniel (maybe you can follow https://github.com/perl6/whateverable/issues/123 for updates on this, but this month I don't have much time to work on stuff anyway)
23:02 AlexDaniel Zoffix: by the way, is it possible we don't remove [REGRESSION] tag from the title of bug reports even if they are resolved?
23:02 AlexDaniel I'm not sure I understand the reason behind this. All I know is that it makes it harder for me to navigate through my reports
23:10 timotimo i wonder if we should try to fix the "but" + native parameter = invalid build_least_derived plan bug
23:11 timotimo it's probably just that the executor from the buildplan needs to have parts copied over to the build_least_derived one, but i think they've been different before and i have no idea why or in what way
23:13 Ven joined #perl6-dev

| Channels | #perl6-dev index | Today | | Search | Google Search | Plain-Text | summary