Camelia, the Perl 6 bug

IRC log for #parrot, 2010-10-04

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:07 cotto whiteknight, if you know your way around distutils all of the nqp files in parrot-instrument need to be compiled down to pbc and merged into InstrumentLib.pbc with pbc_merge.
00:29 whiteknight left #parrot
00:47 whiteknight joined #parrot
01:27 whiteknight left #parrot
01:29 plobsing cotto: ping
01:31 theory joined #parrot
01:57 theory left #parrot
02:25 Andy joined #parrot
02:35 janus left #parrot
02:42 janus joined #parrot
02:56 dukelet0 'ello
02:56 GeJ Heya.
02:57 cotto plobsing, pong
02:57 dukelet0 GeJ: how goes it? what are you hacking on with your new commit bit?
02:59 plobsing cotto: I've been looking at the profiling runcore failure some more. I now suspect my initial assessment is dead wrong.
02:59 plobsing notably, reverting the commit I blamed doesn't solve the problem
02:59 cotto do tell
03:00 plobsing what I do notice though is that the code makes use of a lot of preop_x  variables *after* the op has been executed.
03:01 plobsing somewhere one of these may be pointing to state that is no longer valid, or at least no longer valid in the context it is being used.
03:03 cotto Are you thinking that I should try to copy the data I'll need into someplace where I know it won't be changed
03:04 plobsing I think we should calculate all the preop_x stuff needed for accounting *before* we actually execute the op.
03:05 cotto That sounds like a good idea in any event.
03:06 plobsing also, I'm having trouble wrapping my head around a function that big with so much branching. I might take my refactor hammer to it later.
03:06 cotto You're welcome to.
03:07 jsut joined #parrot
03:07 cotto My last attempt was to factor out common code, but there's not a whole lot.
03:08 cotto Breaking it into functions, even if they're only called once, could do a lot for readability.
03:09 cotto How soon is "later" for you?  It's starting to sound like a good thing for me to tackle.
03:10 plobsing if you're eager to fix it, be my guest
03:11 * cotto starts hammering
03:11 plobsing just be sure to move the preop_line calculation stuff to before DO_OP
03:12 cotto I'll try breaking it into functions first, but that's on the todo list.
03:12 jsut_ left #parrot
03:12 GeJ dukeleto: Right now, a real estate app in Perl 5 for $work.
03:13 cotto plobsing, thanks for the feedback.  It'd a valuable data point that my code is hard to follow.
03:13 cotto s/'d/'s/
03:13 GeJ dukeleto: but at home, in my secret lair, I'm working on taking over Data::Dumper.
03:14 plobsing cotto: at least you're smart enough to be able to do something about it. all I can do is stay away from bus routes.
03:16 * cotto recalls that naming things is one of the two hard problems in programming.
03:17 plobsing the other being implementing said things?
03:17 cotto the other being cache invalidation
03:21 dalek parrot: r49433 | plobsing++ | trunk (7 files):
03:21 dalek parrot: add --ar and --arflags options to Configure.pl
03:21 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49433/
03:33 Andy left #parrot
03:38 cotto --arflags, in the middle of --arstreet
03:51 dalek parrot: r49434 | cotto++ | trunk/src/runcore/profiling.c:
03:51 dalek parrot: break some profiling code off into separate functions
03:51 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49434/
04:07 dalek parrot: r49435 | plobsing++ | trunk/src/sub.c:
04:07 dalek parrot: add some sanity and sanity checking to Parrot_Sub_get_line_from_pc
04:07 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49435/
04:37 dalek parrot: r49436 | cotto++ | trunk/src/runcore/profiling.c:
04:37 dalek parrot: break some more code into separate functions
04:37 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49436/
04:56 cotto plobsing, I may be done for now.  Any further refactoring you care to contribute is welcome.
05:02 plobsing cotto++ # I think I understand it better now
05:02 plobsing emphasis on the think
05:10 cotto It feels good to break that code up.
05:30 plobsing cotto: profiling runcore should be fixed in r49437.
05:32 bacek joined #parrot
05:33 cotto plobsing++
05:33 cotto now let's see if I can run nqp under the profiling runcore
05:37 plobsing can you confirm this is working for you? then I'll update the ticket, but not close it pending testing
05:37 cotto I'm just about to test it.
05:38 cotto I think you may have nailed it.
05:38 cotto It's taking a long time to run, but it used to fail immediately.
05:38 cotto nm
05:39 cotto It explodes with a gc assert failure
05:39 dalek parrot: r49437 | plobsing++ | trunk/src/runcore/profiling.c:
05:39 dalek parrot: lookup op name *before* executing the op
05:39 dalek parrot: interp->code may not point to where we expect after the op
05:39 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49437/
05:39 cotto That's definitely a more different failure though.
05:41 plobsing damn. I hate it when I slow things down. Parrot used to fail so quickly.
05:42 * cotto tries a differenter gc
05:46 cotto I get an assert failure in src/gc/gc_ms2.c whether I use --gc=ms or --gc=ms2
05:46 cotto Ah.  --gc ms is the correct spelling
05:46 cotto that fails much more quickly
05:48 plobsing parrot's longopts parser isn't very flexible. not that this is necessarily a bad thing...
05:49 cotto Attempting to use inf is as bad an idea as I'd expect it to be with nqp.
05:50 plobsing what? you don't have an infinite amount of memory?
05:51 cotto I settled for infinity minus one.
05:51 * plobsing wonders what the swap behaviour would be for gc-inf. if the generational heuristic is accurate, it may not be that bad.
05:52 sorear plobsing: there's been quite a bit of work on that (see: log-structured file systems)
06:04 cotto get_ns_cstr shows up in the backtrace
06:19 bluescreen left #parrot
06:26 jsut_ joined #parrot
06:30 jsut left #parrot
06:33 ash_ left #parrot
07:02 dukeleto left #parrot
07:07 bacek left #parrot
07:17 eternaleye left #parrot
07:36 fperrad joined #parrot
07:57 dalek parrot: r49438 | mikehh++ | trunk/src/runcore/profiling.c:
07:57 dalek parrot: fix codetest failures - remove space before closing parenthesis,
07:57 dalek parrot: remove trailing space and add function documention
07:57 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49438/
07:57 dalek parrot: r49439 | mikehh++ | trunk/src (2 files):
07:57 dalek parrot: fix codetest failures - remove space after opening parenthesis and
07:57 dalek parrot: from before closing parenthesis, and forgot to update copyright
07:57 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49439/
08:03 cosimo_ left #parrot
08:12 luben_work joined #parrot
08:15 cosimo joined #parrot
08:22 eternaleye joined #parrot
08:25 dukelet0 is now known as dukeleto
08:32 perlite left #parrot
08:33 perlite joined #parrot
08:41 esskar left #parrot
08:43 mikehh plobsing: ping
09:08 mikehh aloha msg kid51 t/steps/progs-01 thru 04.t failing, I think due to r49433 - I tried adding ar and arflags to the list(s) but that didn't seem to work
09:08 aloha mikehh: OK. I'll deliver the message.
09:33 mikehh aloha msg plobsing t/steps/progs-01 thru 04.t failing, after r49433 (post-config tests)
09:33 aloha mikehh: OK. I'll deliver the message.
09:53 cogno joined #parrot
09:57 contingencyplan left #parrot
10:08 cogno left #parrot
10:27 cogno joined #parrot
10:42 cogno left #parrot
10:47 cogno joined #parrot
11:01 cogno left #parrot
11:02 kurahaupo joined #parrot
11:13 bacek joined #parrot
11:21 kurahaupo left #parrot
11:24 cogno joined #parrot
11:27 sjn left #parrot
11:30 sjn joined #parrot
11:36 cogno left #parrot
11:50 whiteknight joined #parrot
11:50 tadzik joined #parrot
12:10 whiteknight good morning, #parrot
12:18 ascent left #parrot
12:19 ascent joined #parrot
12:29 cogno joined #parrot
12:35 ruoso joined #parrot
12:54 plobsing mikehh: tests should be fixed as of r49440.
12:55 plobsing also, do we care that t/steps/gen/crypto-01.t fails?
13:04 dalek parrot: r49440 | plobsing++ | trunk (5 files):
13:04 dalek parrot: fix failing config tests
13:04 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49440/
13:10 tadzik left #parrot
13:14 mikehh plobsing: ok tests PASS - it doesn't seem to run t/steps/gen/crypto-01.t for me
13:14 plobsing yeah, I stumbled over it by running prove t/steps/*
13:14 mikehh plobsing: if I prove it fails after first test
13:15 Patterner left #parrot
13:16 mikehh plobsing: I usually run configure with --test
13:17 mikehh ie - date && time perl Configure.pl --test --maintainer --configure_trace --optimize
13:17 mikehh or even date && time perl Configure.pl --test --cc=g++-4.5 --cxx=g++-4.5 --link=g++-4.5 --ld=g++-4.5 --maintainer --configure_trace --optimize
13:19 mikehh I drop the --optimize if I want to test the ASSERTs
13:21 mikehh Ubuntu/Kubuntu 10.10 defaults to gcc-4.4 but I built perl 5.12.2 with gcc-4.5 so I try and use that
13:29 plobsing I can't seem to find when gen::crypto was removed
13:29 PerlJam joined #parrot
13:29 Psyche^ joined #parrot
13:30 Psyche^ is now known as Patterner
13:33 mikehh plobsing: looking for it
13:38 cogno left #parrot
13:45 Coke crypto stuff was moved out of core, neh?
13:49 plobsing AFAICT, yes. this appears to be a fossil
14:04 Coke perhaps we need a test to check for invalid configure tests. ;)
14:06 dalek parrot: r49441 | plobsing++ | trunk (2 files):
14:06 dalek parrot: remove useless tests for now-removed gen::crypto
14:06 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49441/
14:20 patspam joined #parrot
14:22 PacoLinux left #parrot
14:28 bluescreen joined #parrot
14:52 PacoLinux joined #parrot
14:55 Andy joined #parrot
15:01 patspam1 joined #parrot
15:01 ash_ joined #parrot
15:04 patspam left #parrot
15:18 PerlPilot joined #parrot
15:19 PerlPilot left #parrot
15:19 mikehh left #parrot
15:19 Tanami left #parrot
15:23 PerlPilot joined #parrot
15:34 contingencyplan joined #parrot
16:14 davidfetter joined #parrot
16:15 cogno joined #parrot
16:22 cogno left #parrot
16:23 patspam1 left #parrot
16:31 cotto ~~
16:38 tadzik joined #parrot
17:00 theory joined #parrot
17:02 PerlPilot left #parrot
17:04 dukeleto 'ello
17:10 mikehh joined #parrot
17:12 whiteknight hello duke
17:14 davidfetter aaaa...what's up, duke?
17:14 davidfetter </bugs>
17:16 dukeleto davidfetter: drinking my morning coffee thinking about writing this in perl 6 : http://rosettacode.org/wiki/Th​iele%27s_interpolation_formula
17:17 davidfetter h0ly cr4p!
17:17 Andy left #parrot
17:17 davidfetter i presume that this is some numerically efficient way to compute approximations, or a one that converges quickly, or something
17:18 * davidfetter reads
17:20 bluescreen left #parrot
17:20 bluescreen joined #parrot
17:20 davidfetter dukeleto, um, i'm missing some context. what makes thiele worth doing?
17:22 dukeleto davidfetter: to show that perl6 can do it in a few lines.
17:23 dukeleto davidfetter: which isn't a very good reason, i know :)
17:24 luben_work left #parrot
17:28 Coke NotFound++
17:29 Coke dukeleto: use it as an excuse to profile parrot.
17:30 dukeleto Coke: that is a good point. I have been thinking of using generated PIR as benchmarks, since that is the real use case
17:31 dukeleto Coke: most of our benchmarks are pathological or not realistic
17:31 dukeleto davidfetter: what are *you* up to?
17:31 Coke Since I will likely not be able to make parrotsketch again (bad time for my $DAYJOB), I'd appreciate it if someone can bring up my recent tcl-related bug.)
17:32 whiteknight I wonder if we could move it to a different time, again
17:33 whiteknight we always have people who can't make certain times, moving the meeting around would help to get more people involved, at least temporarily
17:33 Coke don't move it on my account, though.
17:34 patspam joined #parrot
17:34 patspam left #parrot
17:34 Coke as my tuits for actual parrot work are not going to improve any time soon.
17:35 Coke (ugh, my home macbox is low on memory)
17:35 dukeleto Coke: do you have a link for the thing you want brought up?
17:37 Coke https://trac.parrot.org/parrot/ticket/1811
17:37 mikehh left #parrot
17:37 davidfetter dukeleto, it's a fine reason. are there things that go to elegant and useful in there, or is that "just" a way to show off the power?
17:38 PerlJam left #parrot
17:38 fperrad_ joined #parrot
17:38 PerlJam joined #parrot
17:39 dukeleto davidfetter: there are some really interesting rosetta code problems, and reading all the solutions in different languages really makes you see the strengths and weaknesses of different languages
17:39 Coke I have another failure in partcl-new that I should be able to bisect sometime today.
17:39 dukeleto no parent SL1.00sc00001.g1.t1 (SL1.00sc00001.g1.t1);
17:39 dukeleto oops, accidental paste.
17:40 davidfetter what have you run across in rosetta that shows off parrot's weaknesses so far?
17:41 dukeleto davidfetter: i don't know if there are any PIR solutions yet, but there a quite a few perl 6 solutions
17:41 fperrad left #parrot
17:41 davidfetter :)
17:42 fperrad_ is now known as fperrad
17:42 dukeleto Coke: perhaps mentioning it on parrot-dev will get some eyes on it
17:43 dukeleto Coke: telling chromatic that he broke something usually gets it fixed pretty soon :)
17:44 davidfetter what's the latest re: the Great Git Migration™?
17:44 Coke I cc'd him on the ticket.
17:45 cotto davidfetter, the plan is to start it shortly after the next release
17:45 davidfetter sweet
17:45 davidfetter cotto++ :)
17:46 davidfetter i don't know whether i really love git, but i know it's hard to live without once you've had some
17:47 cotto there are still a few pieces that need doing
17:47 dukeleto yeah, like mk_language_shell and create_language
17:48 cotto like those
17:54 cotto dukeleto, are you to the point of reconsidering whether those need to be hard dependencies?
18:01 M_o_C joined #parrot
18:05 dukeleto cotto: what do you mean?
18:09 cotto Are the scripts enough of a mess that we should figure out the problem of dealing with minimum version requirements before updating the language creation scripts?
18:10 kurahaupo joined #parrot
18:12 cotto Maybe the best approach would be to use another hll (Rakudo or Partcl) as a guinea pig.
18:20 dukeleto One big problem: mk_language_shell and create_language have no tests.
18:20 cotto sigh
18:20 cotto I guess running them manually isn't too hard.
18:21 dukeleto the fact that we have two scrips that do *almost* the same thing, that duplicate each others code, and that have no tests => sucks
18:21 dukeleto that isn't a huge deal, but it factors in
18:21 whiteknight add tests
18:22 whiteknight then pick the tool that you want, and add new features to it
18:26 fperrad_ joined #parrot
18:26 fperrad left #parrot
18:27 fperrad_ is now known as fperrad
18:28 whiteknight better yet, consolidate the two tools, and use a command-line switch to select behaviors
18:28 whiteknight And then, create a thin front-end for them that emulates the current behavior but calls back into a single, well-tested core
18:28 whiteknight viola
18:30 plobsing why not leave it up to the parties whose political differences are responsible for there being two tools in the first place?
18:30 whiteknight because they are obviously doing it wrong
18:31 mikehh joined #parrot
18:31 plobsing when I said it, I meant fixing their respective tool to work with git
18:32 preflex left #parrot
18:35 preflex joined #parrot
18:36 whiteknight There are two unrelated problems: First, that we need to move tools to git, and Second that we have two tools that do almost the same thing and are poorly designed and untested
18:40 dukeleto Look at the madness here: http://trac.parrot.org/parrot/ticket/1491
18:41 davidfetter "don't do the browns. the browns are a bummer"
18:44 PerlPilot joined #parrot
18:47 whiteknight dukeleto: we can still provide the behaviors that everybody wants, and the interfaces that everybody wants, but improve the internals dramatically
18:47 fperrad left #parrot
18:49 Coke partcl doesn't use either of those tools.
18:49 Coke pmichaud may have used one when he rolled up partcl-nqp, but it's a one shot, and if you fix the script, i get no benefit from it.
18:50 Coke of all the things we could work on improving, the internals of those scripts seems like a low-impact one.
18:50 dukeleto whiteknight: that is what I just suggested on the ticket
18:51 dukeleto Coke: those two scripts are currently the blockers for moving to git
18:51 Coke I've said this before: do a braindead translation of them and be done with it. don't let git block on that.
18:52 Coke and there's no point in worrying about people that are currently using svn, because they won't get the new scripts by default.
18:52 Coke (that is, in terms of fixing /those scripts/)
18:52 whiteknight in terms of speeding along the git translation, a "brain dead translation" is a fine short-term solution
18:52 dukeleto Coke: what do you mean a "braindead translation" ?
18:52 whiteknight in the longer term, if we want people to use and rely on these things, we need to put more tought into them
18:52 Coke dukeleto: what do you mean they're blocking you?
18:53 Coke why are they so much more effort to translate over than the other scripts that have already been done?
18:53 dukeleto Coke: the way they are written now is incompatible with git
18:53 Coke ok. what does that have to do with the fact that there's 2 of them?
18:53 dukeleto Coke: there is no monotonically increasing integer revision number to compare in git
18:53 Coke the whole (political) issue of which one is the right one should not be blocking the git migration.
18:54 dukeleto Coke: i have to convert two almost identical things to git, and they are untested
18:54 Coke ok. that's not the impression I get of the blocker when this issue is discussed.
18:54 Coke (almost identical) that's the part Ig et as the major blocker.
18:54 dukeleto Coke: and I have to change their behavior when converting them to git
18:54 Coke and it's just a fact at the moment. you're not going to solve that problem pre-git, IMO.
18:54 dukeleto Coke: the "almost identical" part just doubles my work, is all
18:55 cotto and is an excellent demotivator
18:55 dukeleto Coke: i didn't quite parse what you just said. Can you repeat?
18:55 Coke I don't know how to make my position any clearer. nevermind.
18:56 Coke Wake me up when git is here. ;)
19:00 dukeleto Coke: "(almost identical) that's the part Ig et as the major blocker"
19:01 dukeleto Coke: i don't understand what you are tying to say there
19:01 dukeleto Coke: and if you can exactly describe what you mean by "braindead translation", that would be useful.
19:02 dukeleto Coke: it is not like translating a shell script to perl. Fundamental changes to the algorithm need to happen.
19:02 dukeleto Coke: and doing those twice, in an untested manner, irks me, the person doing all the work.
19:02 dukeleto "all the work" to translate them, I mean.
19:03 whiteknight until we know what we want, maybe the best way forward is to rip out all source control logic
19:03 dukeleto we could go the route of "we will fix those scripts after we migrate", i guess
19:04 whiteknight Rip out all the svn code and don't replace it with anything until we know what we want
19:04 dukeleto whiteknight: that is a possibility
19:06 whiteknight these scripts probably need a complete rewrite anyway, so we can't turn that into a migration blocker too
19:06 cotto What about making a language like Rakudo or Partcl work with git?  Getting the git logic working in one of those would help us figure out how the language creation scripts will need to change.
19:07 cotto it'd also let us work on one thing at a time and get it working before moving on to the next
19:07 dukeleto cotto: yeah. i have a branch of rakudo for doing that, but it doesn't have anything useful yet
19:07 cotto rather than figuring out a git strategy *and* migrating a bunch of duplicated code
19:07 cotto Commit something stupid and ask people if they have any suggestions. ;)
19:08 dukeleto cotto: the issue is that those two scripts are a nexus of problems. deciding how many to fix before the git migration is the issue
19:08 atrodo cotto> I love doing that
19:08 dukeleto cotto: perhaps I will just make both scripts do a "rm -rf /" and say "patches welcome"
19:08 cotto At this point, +1
19:09 cotto but only if you add sudo
19:09 dukeleto cotto: it seems at this point, tests for either or both scripts are a big win, and are mostly independent of svn/git
19:10 dukeleto cotto: basic tests like "create language X, then make sure that the directory X was created"
19:11 cotto or "create a language and make sure its tests pass"
19:11 cotto Mmmm.  Test tests.
19:12 dukeleto cotto: the fact that parrot version numbers are hard to parse and compare is the actual hardest part of the migration, but that is wrapped up inside all the other issues with those scripts
19:13 cotto Right.  I'm suggesting that we solve that problem first for an existing language (where the scripts don't come in to play), the use that information to tackle the other issues.
19:14 dukeleto cotto: ok, so what would the actual steps of that look like?
19:15 dukeleto cotto: create a rakudo branch that depends on parrot.git ?
19:16 cotto yes, and focus on that first
19:17 dukeleto cotto: one issue is that a lang can't depend on a specific git commit without having the parrot.git repo
19:19 cotto dukeleto, we can modify the parrot executable to spit out the information needed so that a language can tell if the installed parrot is recent enough.
19:19 dukeleto cotto: that is what parrot_config is for, but the situation is hairy
19:20 cotto right.  I just thought of that after typing my response
19:22 dukeleto cotto: git describe can help us
19:22 dukeleto cotto: but there are still some edge cases that make things fun, like x.2.1 coming out after x.3.0
19:23 dukeleto cotto: to get around that, i think appending a date to the output of "git describe" may work
19:26 dukeleto "perl tools/dev/create_language.pl --help" creates a language called "--help". LTA.
19:26 cotto We could make sure hlls specify a date and a minimum version, i.e. something later than 2.8 and after 10/20/2010.
19:26 cotto ...
19:28 cotto I don't even know how to respond to that.
19:29 Andy joined #parrot
19:30 cotto Our scripts probably shouldn't make fun of people who ask for help.
19:35 kurahaupo left #parrot
19:41 atrodo That's just generally a bad design decision
19:42 GeJ Bonjour everyone.
19:42 davidfetter salut, GeJ
19:44 dukeleto GeJ: welcome to the madness
19:45 davidfetter iaaa! iaaa!
19:46 GeJ Just to get used to it... #ps is in roughly 24h, correct?
19:47 cotto GeJ, closer to 25
19:48 cotto 24:42
19:49 GeJ Thanks cotto.
19:50 cotto np
19:51 GeJ can someone with tuits help decrypt some PIR code?
19:51 dukeleto create_language also massively fails if a directory already exists with the name of the language you want to create
19:51 dukeleto GeJ: sure, what are you up to?
19:51 Coke the scripts in question are only run once, generate a bunch of other files, and then /those/ become the language directory, eys?
19:52 GeJ runtime/parrot/library/dumper.pir  lines 166~176
19:52 dukeleto Coke: yes
19:52 Coke ok. so are you asking for tests for those scripts themselves, or for the generated language directory?
19:52 dukeleto GeJ: looks like some funky code going on there
19:53 dukeleto Coke: i am writing tests for create_language right now, and the answer to your question is "both"
19:53 Coke ok. you might want to be clear about that when asking for tests.
19:54 GeJ dukeleto: if I grok it correctly, get me an instance of Data::Dumper if I already stashed one in Data::Dumper::global, make a test to se if I get something back. and in any case, create a new one.
19:54 Coke I find testing the script itself about as helpful as testing a configure step.
19:54 GeJ ... and store it in Data::Dumper::global.
19:56 cotto Coke, we don't expect people to run it often, but the times when they do are pretty important.
19:57 Coke are we talking about configure steps? i'd be happy to hear a use case where that makes sense.
19:58 dukeleto Coke: the first thing a new HLL dev uses is usually one of those scripts. It would be nice if we could be sure they work correctly.
19:58 Coke yes. by running them and testing the result. sure.
19:59 Coke That seems a perfectly reasonable test for the script itself.
19:59 Coke (I can see a few cases where things like 'does it respect this command line option') are helpful. I just would hate to see us head down the same path we did with configure.
20:01 dukeleto Coke: what path is that?
20:02 Coke where we test that each configure step has test coverage, rather than relying on parrot functionality tests.
20:03 Coke in very broad strokes, t/configure is good. t/steps is overkill.
20:06 cotto whiteknight++ for the #1491 reply
20:13 cotto karma git
20:13 cotto aloha, karma git
20:14 GeJ clock?
20:14 aloha GeJ: LAX: Mon, 10:07 PDT / CHI: Mon, 12:07 CDT / NYC: Mon, 13:07 EDT / UTC: Mon, 17:07 UTC / LON: Mon, 18:07 BST / BER: Mon, 19:07 CEST / TOK: Tue, 02:07 JST / SYD: Tue, 04:07 EST
20:15 dalek parrot: r49442 | dukeleto++ | trunk (2 files):
20:15 GeJ aloha: you're drifting sweety.
20:15 dalek parrot: [t] Add some basic tests for create_language.pl
20:15 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49442/
20:15 dalek parrot: r49443 | dukeleto++ | trunk (2 files):
20:15 dalek parrot: [t] Add some basic tests for mk_language_shell.pl
20:15 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49443/
20:15 dukeleto why isn't aloha keeping track of karma?
20:16 mikehh left #parrot
20:17 cotto explain git
20:17 cotto aloha, explain git
20:18 cotto aloha, msg bacek aloha isn't displaying karma
20:18 aloha cotto: OK. I'll deliver the message.
20:19 cotto seen bacek
20:19 aloha bacek was last seen in #parrot 8 hours 46 mins ago joining the channel.
20:20 whiteknight karma dukeleto
20:20 whiteknight karma dukeleto?
20:20 whiteknight aloha, karma dukeleto?
20:20 aloha whiteknight: Search me, bub.
20:21 cotto According to the github code, the karma module is loaded.
20:21 cotto http://github.com/bacek/aloha/blob/master//bot.pl
20:21 cotto clock?
20:21 aloha cotto: LAX: Mon, 10:15 PDT / CHI: Mon, 12:15 CDT / NYC: Mon, 13:15 EDT / UTC: Mon, 17:15 UTC / LON: Mon, 18:15 BST / BER: Mon, 19:15 CEST / TOK: Tue, 02:15 JST / SYD: Tue, 04:15 EST
20:22 whiteknight karma?
20:22 atrodo That clock is a bit slow
20:23 cotto must be due to to the clock on bacek's dev box
20:32 M_o_C left #parrot
20:50 perlite left #parrot
20:50 perlite joined #parrot
20:57 whiteknight left #parrot
21:04 mikehh joined #parrot
21:21 bluescreen left #parrot
21:23 tadzik left #parrot
21:26 cotto dukeleto++
21:38 cotto dukeleto, do those tests work for you?  They look broken on my box:
21:38 dukeleto cotto: they passed for me when I committed them
21:39 dukeleto cotto: you have a smoke report?
21:39 nopaste "cotto" at 192.168.1.3 pasted "new language tools tests broken" (15 lines) at http://nopaste.snit.ch/23986
21:39 dukeleto no good deed goes unpunished
21:39 dukeleto cotto: what version of File::Path do you have?
21:39 dukeleto cotto: you must have an old one
21:40 dukeleto cotto: i can probably just switch to using the older rmtree function
21:40 cotto 2.04 apparently
21:40 cotto must have come with ubuntu
21:40 cotto yes, it has rmtree
21:47 dalek parrot: r49444 | cotto++ | trunk/src/runcore/profiling.c:
21:47 dalek parrot: add a const and cast to make the C++ build work
21:47 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49444/
21:51 dukeleto cotto: fixed
21:55 cotto wfm now.  Thanks.
21:56 ruoso left #parrot
21:56 cotto The tests are a little sparse, but that'll be a good start to build on.
21:57 cotto especially people who aren't you. ;)
22:03 dalek parrot: r49445 | dukeleto++ | trunk/t/tools (2 files):
22:03 dalek parrot: [t] Use rmtree instead of remove_tree to support elderly File::Path's
22:03 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49445/
22:04 PerlPilo1 joined #parrot
22:25 bacek_at_work ~~
22:26 aloha left #parrot
22:26 aloha joined #parrot
22:27 cotto Nice.  I can run headerizer on a single file by throwing it the object file generated by the file I care about.  It's a bit baroque, but it'll save me some time.
22:28 cotto aloha, karma git
22:28 aloha cotto: git has karma of 0.
22:28 cotto aloha, karma c
22:28 aloha cotto: c has karma of 7.
22:28 cotto aloha, karma bacek
22:28 aloha cotto: bacek has karma of 38.
22:28 cotto bacek_at_work, did that get reset again recently?
22:29 bacek_at_work cotto, it was. When I moved it from my home laptop.
22:29 bacek_at_work about couple of weeks ago
22:33 cotto It's good to have that working again.
22:33 bacek_at_work cotto, next time just try to "!load Karma" in aloha's console :)
22:34 cotto Now I know.
22:34 cotto thanks
22:35 cotto Both the bot and the users take some training.
22:35 cotto Would it be possible to scrape moritz's irclogs to recover karma?
22:37 sorear bacek_at_work: btw
22:37 sorear aloha: clock
22:37 sorear clock?
22:37 aloha sorear: LAX: Mon, 15:37 PDT / CHI: Mon, 17:37 CDT / NYC: Mon, 18:37 EDT / UTC: Mon, 22:37 UTC / LON: Mon, 23:37 BST / BER: Tue, 00:37 CEST / TOK: Tue, 07:37 JST / SYD: Tue, 09:37 EST
22:37 sorear oh, you fixed that
22:37 sorear bacek++
22:38 bacek_at_work sorear, yeah... ntpd on virtual machine doesn't work properly...
22:39 sorear any chance of getting aloha to respond only to ^aloha, and not spuriously trigger when people in #perl6 start a sentence with "explain" or "clock"?
22:40 bacek_at_work sorear, I probably can. But code is same for #parrot and #perl6. So behaviour will be same.
22:42 bacek_at_work clock?
22:43 bacek_at_work aloha, clock?
22:43 aloha bacek_at_work: bacek_at_work: LAX: Mon, 15:42 PDT / CHI: Mon, 17:42 CDT / NYC: Mon, 18:42 EDT / UTC: Mon, 22:42 UTC / LON: Mon, 23:42 BST / BER: Tue, 00:42 CEST / TOK: Tue, 07:42 JST / SYD: Tue, 09:42 EST
22:43 aloha bacek_at_work: No clue. Sorry.
22:43 bacek_at_work sigh...
22:43 bacek_at_work There is a "bug" in Bot::BasicBot::Pluggable. It doesn't stop processing modules after first hit.
22:52 Andy left #parrot
23:01 * cotto almost has the main loop of the profiling runcore down to one screenful of code
23:02 cotto not that that's a goal, but it won't hurt anything either
23:04 dalek parrot: r49446 | cotto++ | trunk/src/runcore/profiling.c:
23:04 dalek parrot: break a couple more chunks of code into separate functions plus small cleanups
23:04 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49446/
23:05 plobsing cotto: any luck with that gc bug?
23:05 cotto I'm going to take a crack at factoring out more code and then tackle that one.
23:07 cotto It'd also be nice to make the code a little less branchy.
23:12 whiteknight joined #parrot
23:12 plobsing it's already significantly less deep. that update_ctx stuff was messy
23:14 ruoso joined #parrot
23:16 whiteknight good evening, #parrot
23:17 davidfetter guten abend, whiteknight
23:17 whiteknight hello davidfetter, how are you?
23:17 davidfetter i'm well, thanks. yourself?
23:18 whiteknight doing pretty damn good myself, thanks
23:18 Kulag left #parrot
23:22 GeJ G'Day bacek.
23:23 Kulag joined #parrot
23:24 GeJ bacek_at_work: lost one hour of sleep this week-end?
23:26 bacek_at_work GeJ, almost.
23:28 GeJ gained, then?
23:28 * GeJ finds this DST thingy very confusing.
23:28 * davidfetter springs forward
23:28 davidfetter ...only upside down from bacek_at_work
23:35 dalek parrot: r49447 | cotto++ | trunk/src/runcore/profiling.c:
23:35 dalek parrot: [profiling] break line number retrieval into a separate function
23:35 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49447/
23:46 hercynium joined #parrot

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

Parrot | source cross referenced