Camelia, the Perl 6 bug

IRC log for #parrot, 2010-11-02

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:00 cotto ok
00:00 hatseflats left #parrot
00:02 mikehh cotto: for the same reason I configure with --test and --configure_trace, it adds more things to test.
00:04 mikehh cotto: been doing that for a while now, probably should add some other things
00:06 lucian_ left #parrot
00:12 dmalcolm left #parrot
00:21 kid51_at_dinner left #parrot
00:55 ash_ joined #parrot
00:56 * dukeleto is going to eat some food, then dig some heels into the git migration
00:57 kid51 joined #parrot
01:00 cotto dukeleto, how's the dry run going?
01:10 dngor_ joined #parrot
01:10 dngor left #parrot
01:15 dngor joined #parrot
01:16 dngor_ left #parrot
01:26 dukeleto cotto: i am crafting the filter-branch command to use the new author map
01:28 dalek parrot: r49767 | jkeenan++ | branches/tt1841_osvers:
01:28 dalek parrot: Creating branch to explore possibility of adding 'osvers' during
01:28 dalek parrot: configuration.
01:28 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49767/
01:28 cotto dukeleto, does that mean that the final switch won't happen tonight?
01:31 dukeleto right now, i am deciding between using git-svn clone with a local repo or git filter-branch
01:37 dukeleto cotto: i am timing a clone with the new author map now
01:38 dukeleto cotto: it just passed r500
01:38 cotto wheeee
01:39 dukeleto cotto: i am doing a new clone with the -A param, which lets you set an author map
01:39 dukeleto cotto: it might be slower than filter-branch, but i would have to write a small bit of code for filter-branch that i can't seem to find already written currently
01:40 cotto dukeleto, good for you.  Go for the lazy approach.
01:41 dukeleto and it blew up because "(no author)" was not in the author map
01:42 cotto who's the culprit?
01:44 dukeleto expected "feature". It blows up if it doesn't know what to do with an author. I acn re-run with a default author flag, methinks
01:45 cotto credit them to purl
01:48 dukeleto but nothing should have "no author", it thinks r1345, which happens to be REL_0_0_5, has no author.
01:48 * dukeleto feels the ground shake as if a herd of yaks are on the move
01:49 * cotto breaks out the electric yak shaver
01:53 * dukeleto has the clone going again, this time rewriting things to a "nobody" person if there is no author
01:55 dukeleto we are just about at r2000
01:56 * cotto still thinks it should be "purl"
01:59 dalek parrot: r49768 | jkeenan++ | branches/tt1841_osvers/config (5 files):
01:59 dalek parrot: First pass at determining 'osvers' during configuration.
01:59 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49768/
02:01 dukeleto we just passed r3000
02:02 dukeleto at this rate we are looking at 4-5 hours
02:02 dukeleto my machine is bogged down doing other things, though, so it is slower than normal
02:02 dukeleto filter-branch is probably much faster
02:03 dukeleto cotto: check this out : http://github.com/haarg/convert-git-dbic
02:04 dukeleto DBIx::Class did some serious stuff
02:04 dukeleto cotto: http://github.com/haarg/convert​-git-dbic/blob/master/10.update <-- this means we can incrementally update while fiddling with authors
02:05 dukeleto author maps, actually. We don't want to fiddle with the authors.
02:05 cotto win
02:06 dukeleto cotto: i would say we will have dry run data tonight, if things look good, we can plan for tomorrow
02:07 cotto Being the author of the trac plugin, I'd like to be around in case anything goes south.
02:07 whiteknight tomorrow? if you can forward the relevant info to me, I will email the group tomorrow morning when I get up
02:07 cotto I have an interview from 12:00 probably until 5-ish.
02:08 whiteknight so just let me know when to tell people it's happening
02:08 cotto and a thing I'd like to do in the evening, but that's not vital.
02:08 whiteknight but for now, I'm heading to bed
02:08 cotto whiteknight, I can send out the mail since I'll be up later.
02:09 cotto do you mind?  I'd like to give as much notice as possible.
02:09 whiteknight okay, that's fine too. email me with whatever you need me to do, and I'll do it when I get up
02:09 whiteknight goodnight
02:09 cotto 'night
02:09 whiteknight left #parrot
02:17 lidden left #parrot
02:29 dukeleto we are at r6000
02:33 cotto ETA seems to be "don't hold your breath"
02:47 dukeleto yep.
02:53 cotto What's the current plan?
03:15 davidfetter joined #parrot
03:17 dalek parrot: r49769 | jkeenan++ | branches/gcc_defines/config/auto (2 files):
03:17 dalek parrot: Implement attempt at probe for rtems based on suggestion by doughera++.
03:17 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49769/
03:17 dalek parrot: r49770 | jkeenan++ | branches/gcc_defines/MANIFEST:
03:17 dalek parrot: Update MANIFEST.
03:17 dalek parrot: review: http://trac.parrot.org/parrot/changeset/49770/
03:23 dukeleto we are at r11000
03:24 cotto dukeleto, what's the plan?  Once you're done can you put that repo on github and catch later changes via incremental update?
03:26 dukeleto cotto: yes. The plan is to look at the output and see if it is sane.
03:28 cotto ok.  Will there be a way to update existing clones or is the best approach to nuke and start over?
03:28 kid51 left #parrot
03:29 dukeleto cotto: easiest and most probably best is to nuke and start over
03:30 dukeleto cotto: otherwise, people will have huge disjoint history in their index, which will waste lots of space, which they then have to gc => pain in the ass
03:30 davidfetter ...it's the only way to be sure
03:30 davidfetter lag :P
03:34 cotto dukeleto, can you draft a quick note to parrot-dev notifying that people will need to nuke their existing clones once the new repo is up?
03:37 dukeleto cotto: yeps
03:38 cotto dukeleto++
03:52 patspam left #parrot
04:35 [hudnix] left #parrot
05:22 theory joined #parrot
05:31 Kulag joined #parrot
05:32 simcop2387_ joined #parrot
05:33 FullMetalHarlot joined #parrot
05:33 x3nU_ joined #parrot
05:33 szabgab_ joined #parrot
05:33 ascent_ joined #parrot
05:33 dngor_ joined #parrot
05:33 Koven joined #parrot
05:33 jnthn___ joined #parrot
05:33 plobsing_ joined #parrot
05:33 pmichaud_ joined #parrot
05:33 Tene_ joined #parrot
05:33 Maddingu1 joined #parrot
05:34 Drossel left #parrot
05:35 simcop2387 left #parrot
05:35 dngor left #parrot
05:35 confound left #parrot
05:35 Kovensky left #parrot
05:35 simcop2387_ is now known as simcop2387
05:41 plobsing left #parrot
05:41 estrabd left #parrot
05:41 dzoe_ left #parrot
05:41 x3nU left #parrot
05:41 ascent left #parrot
05:41 szabgab left #parrot
05:41 jnthn_ left #parrot
05:41 Maddingue left #parrot
05:41 pmichaud left #parrot
05:41 Tene left #parrot
05:41 KatrinaTheLamia left #parrot
05:41 moritz left #parrot
05:41 sECuRE left #parrot
05:41 confound joined #parrot
05:41 dzoe joined #parrot
05:41 moritz_ joined #parrot
05:41 estrabd joined #parrot
05:41 sECuRE joined #parrot
05:41 dukeleto we are at ~r21000, almost half way
05:44 jhelwig left #parrot
06:37 theory left #parrot
07:01 uniejo joined #parrot
07:26 fperrad joined #parrot
07:52 dngor_ is now known as dngor
08:04 GeJ dukeleto: how are things going on your side?
08:04 GeJ aloha: clock?
08:04 aloha GeJ: GeJ: LAX: Tue, 01:04 PDT / CHI: Tue, 03:04 CDT / NYC: Tue, 04:04 EDT / UTC: Tue, 08:04 UTC / LON: Tue, 08:04 GMT / BER: Tue, 09:04 CET / TOK: Tue, 17:04 JST / SYD: Tue, 19:04 EST
08:05 GeJ probably asleep now.
08:05 cotto dukeleto, I'm off to bed.  If can figure out a schedule, feel free to send out the message to parrot-dev, etc to give people as much notice as possible.
09:11 moritz_ is now known as moritz
09:16 bacek joined #parrot
09:33 contingencyplan left #parrot
09:45 Maddingu1 is now known as Maddingue
10:02 PacoLinux left #parrot
10:03 PacoLinux joined #parrot
10:17 moritz http://olabini.com/blog/2010/07/qu​estioning-the-reality-of-generics/ "I’m increasingly coming to the conclusion that multi language VM’s benefit from being as dynamic as possible."
10:27 fperrad_ joined #parrot
10:31 fperrad left #parrot
10:31 fperrad_ is now known as fperrad
10:37 sjn left #parrot
10:37 sjn joined #parrot
11:35 kid51 joined #parrot
12:05 kid51 left #parrot
12:29 whiteknight joined #parrot
12:41 whiteknight good morning, #parrot
12:41 whiteknight dukeleto++ for all his work last night
12:45 bluescreen joined #parrot
12:48 hudnix joined #parrot
12:54 hatseflats joined #parrot
12:57 hatseflats left #parrot
13:56 uniejo left #parrot
14:13 ash_ left #parrot
14:35 plobsing_ left #parrot
14:41 theory joined #parrot
14:49 dukeleto this thing is still going. oy.
14:53 davidfetter nu?
14:54 * dukeleto goes to the dentist
15:15 theory left #parrot
15:18 whiteknight what thing is still going, the git import?
15:41 jsut joined #parrot
15:46 cotto ~~
15:46 jsut_ left #parrot
15:46 cotto whiteknight, that's probably what he was talking about
15:50 dmalcolm joined #parrot
15:50 * cotto is glad that dukeleto can do an incremental update once the git import is done
15:51 cotto It's Hofstadter's Law at work.
15:55 cotto moritz, nice article
15:56 * davidfetter contemplates hofstadter's law at play
15:59 cotto Don't contemplate it.  You'll just make things take even longer.
15:59 atrodo hofstadter is overrated
16:01 * cotto goes back to sleep.  And by "sleep", I mean shower.
16:03 atrodo I wish i could go back to sleep. And by sleep, I mean sleep
16:20 patspam joined #parrot
16:30 whiteknight I suspect it's dangerous to sleep in the shower
16:33 mikehh #join #parrotsketch
16:34 mikehh I think that was bad typing, but maybe a message to youse guys
16:39 fperrad left #parrot
16:40 tadzik joined #parrot
16:41 tadzik ~~
16:42 davidfetter left #parrot
16:46 contingencyplan joined #parrot
17:17 whiteknight msg dukeleto is it okay if I list you as  a "resource" for the embedding API task force? I know you're probably too busy to get directly involved, but I want to make sure we have a list of users of the API that we can ask questions of
17:17 aloha OK. I'll deliver the message.
17:17 * whiteknight is looking for additional members for the API task force, if anybody is interested
17:20 atrodo whiteknight> I have a vested interested in it
17:20 whiteknight atrodo: you want on? I'd be happy to list you
17:20 whiteknight or, do you want to be listed as a resource?
17:20 whiteknight we need both
17:21 whiteknight people who embed parrot and can answer questions about their experiences, or who would be willing to review proposals and test changes are very very valuable
17:27 atrodo I can probably be on the team to some degree
17:27 whiteknight what project do you work on that embeds parrot?
17:28 atrodo trying to use parrot as the code generation of my compiler
17:31 fperrad joined #parrot
17:43 whiteknight so you're using the embedding API in parrot/embed.h?
17:44 whiteknight is your project public?
17:59 bluescreen left #parrot
17:59 bluescreen joined #parrot
17:59 atrodo yes and yes
17:59 atrodo http://github.com/atrodo/draak
18:01 whiteknight methinks I've seen this before
18:03 donaldh joined #parrot
18:04 atrodo i think you have.  I sent a few emails to parrot-dev about it
18:10 jhelwig_ joined #parrot
18:18 donaldh left #parrot
18:25 diakopter joined #parrot
18:27 tadzik whiteknight: how could one help with ProductManagementTeam?
18:27 tadzik left #parrot
18:27 tadzik joined #parrot
18:31 whiteknight tadzik: The big task right now is to run triage on the embedding API, and start prototyping an API that is really useful
18:44 whiteknight tadzik: you want to get involved?
18:46 tadzik whiteknight: I'm not really embedding. What would I need to do?
18:47 whiteknight well, you don't need to be embedding yourself in order to know what a good interface looks like, and to get in touch with the people who are doing the embedding
18:48 whiteknight or, to hunt for tickets involving the embed interface and evaluate them
18:48 tadzik the only embedding I ever used was embedding Lua in C
18:48 whiteknight ha. Everybody embeds Lua
18:48 tadzik and I'd love to embed Parrot Lua one day
18:48 whiteknight yes, that would be quite nice
18:49 tadzik and maybe one day run Awesome WM on Parrot Lua, so one could script it in any Parrot language
18:53 whiteknight :)
18:53 whiteknight tadzik: what kinds of projects do you work on?
18:53 whiteknight you on github or anywhere I can browse your handicrafts?
18:54 tadzik whiteknight: Perl 6 modules and module management, few Rakudo patches, also writing my xmpp client. I'm on github, as tadzik
18:55 tadzik (gtkabber sucks, beware. I'm rewriting it right now, with the experiences I got when I was writing it for the first time)
18:57 whiteknight ah nice.
19:08 whiteknight I was looking at the possibility of embedding Parrot in a chat client, like Pidgin
19:08 whiteknight I'm trying to embed it now into xchat, since I use that every day
19:08 tadzik hmm
19:09 tadzik since gtkabber is experimental as hell and no one uses it anyway, I could use Parrot under the hood, instead of Lua. In the rewrite I haven't got to the scripting part anyway. That would be some kind of research
19:10 whiteknight the important part is to get people using Parrot in an embedding context, even if only experimentally
19:11 whiteknight because the more people use it, the more feedback we get and the better product we can create
19:11 tadzik Does embedding Parrot imply embedding Parrot languages?
19:12 tadzik At what stage embedding is today?
19:15 whiteknight you can create and use a parrot interpreter from inside another program. With the interpreter loaded, you can load the bytecode for a language compiler and call that
19:15 tadzik cool
19:16 * tadzik gets more followers on GH, feels responsibility
19:16 whiteknight PL/Parrot does this, creates a Perl6 compiler inside Postgres, I think
19:19 M_o_C joined #parrot
19:20 M_o_C left #parrot
19:30 tadzik whiteknight: are there any docs on that lying around?
19:39 szabgab_ is now known as szabgab
19:41 whiteknight you'd have to talk to dukeleto
19:42 Tene_ You'd also do well to contact jhorwitz to talk about mod_parrot for apache
19:43 Tene_ I've made a couple of commits there, but I never really understood it overall.
19:43 Tene_ is now known as Tene
19:43 Tene afaik he's not really active these days, but he's always been very responsive via email.
19:43 Tene or AIM for that matter
19:46 tadzik msg dukeleto are there any docs for Parrot embedding lying around?
19:46 aloha OK. I'll deliver the message.
19:47 nwellnhof joined #parrot
19:53 whiteknight nwellnhof++
19:56 diakopter left #parrot
20:02 chromatic joined #parrot
20:10 dukeleto tadzik: yes, there are some in docs/embed.pod and docs/extend.pod
20:11 dukeleto tadzik: PL/Parrot embeds a Parrot interpreter, and optionally, a Rakudo Perl 6 interpreter as well
20:12 whiteknight those docs should be viewed with suspicion
20:12 whiteknight and a sense of sadness, since they are going to die soon
20:14 * atrodo moans
20:15 dukeleto whiteknight: why are they going to die soon?
20:15 atrodo s/moans/mourns/
20:15 dukeleto whiteknight: i read through them to figure out how to write PL/Parrot
20:15 whiteknight dukeleto: because I'm hiding in the closet, and I have an axe
20:15 dukeleto whiteknight: they are incomplete, and out-dated in a few places, but still useful
20:16 whiteknight cue scary thunder sound. cue frightening music. The killer attacks his target. Exeunt stage left
20:16 whiteknight dukeleto: fear not, they will be replaced with something better, when the time comes to kill and replace them
20:16 atrodo the cycle continues
20:17 whiteknight it is the circle of life
20:18 atrodo And that'll be in my head for the rest of the night
20:20 dukeleto There are some things i want to change about the embedding interface, such as: the requirement of giving a previously-created interp if you want to create more than one interp
20:20 dukeleto Seems like a bad design. And makes security-minded people scared, like me.
20:20 mikehh Util: congrats on joining us grandparents
20:25 whiteknight damn, I'm still working on my first regular kid. I'm far away from grand kids
20:26 atrodo whiteknight> Pretty sure it just takes time
20:26 whiteknight dukeleto: make ticket, assign to me, etc
20:27 whiteknight atrodo: ain't that the truth? I was a stupid young man, and made stupid young man decisions. I can't imagine my son will fall too far from the tree
20:28 kid51 joined #parrot
20:28 kid51 ps in 2
20:28 whiteknight and if my kid keeps smashing his face on things, I doubt he'll be too bright
20:28 whiteknight but, I have to sign off and go home now. Later
20:29 whiteknight left #parrot
20:35 Themeruta joined #parrot
20:35 NotFound left #parrot
20:42 dukeleto the dry-run of converting the parrot repo again, with a proper authormap, is still running
20:43 dukeleto it is hard to tell where it is, because it has to walk the full history of every branch that ever existed
20:43 dukeleto So i have exactly zero clue about how long it will take. I should have used filter-branch. Now I know, and knowing is half the battle...
20:46 wagle toldya toldya
20:50 wagle 8)
20:52 he_ left #parrot
20:55 dukeleto wagle: i need the invocation for git filter-branch to use an author file
20:58 * wagle has phone interview in t-minus 1 minute
20:59 GeJ Bonjour everyone.
21:02 dukeleto GeJ: howdy
21:02 dukeleto wagle: good luck
21:02 dukeleto GeJ: #ps is happening now, if you forgot
21:04 Topic for #parrot is now Parrot 2.9.1 Released | http://parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: http://nopaste.snit.ch:8001 | merge gsoc_nci |
21:09 Andy joined #parrot
21:15 lucian joined #parrot
21:17 theory joined #parrot
21:18 bluescreen left #parrot
21:26 kid51 left #parrot
21:28 dukeleto has anybody smoked gsoc_nci recently?
21:29 dukeleto mikehh: gsoc_nci is so close. Want to help me merge it today?
21:29 mikehh dukeleto: will do that after gc_ms2_sf
21:30 dukeleto mikehh++
21:30 mikehh testin' that is
21:37 donaldh joined #parrot
21:45 nwellnhof left #parrot
22:03 theory left #parrot
22:04 mikehh chromatic: gc_ms2_sf - fulltest PASS - benchmakk test 90 wallclock secs ( 0.08 usr  0.01 sys + 86.02 cusr  3.66 csys = 89.77 CPU) vs 97 wallclock secs ( 0.07 usr  0.01 sys + 92.87 cusr  4.02 csys = 96.97 CPU) in trunk
22:05 chromatic What did you use for a benchmark?
22:06 mikehh benchmark test in fulltest
22:07 chromatic Thanks.
22:07 mikehh both run with g++ on Ubuntu 10.10 i386
22:08 theory joined #parrot
22:10 bacek left #parrot
22:11 mikehh actually g++-4.5 (the default is g++/gcc 4.4 but I built perl 5.12.2 with gcc-4.5)
22:13 he left #parrot
22:25 donaldh dukeleto: a question about gsoc_nci
22:26 dukeleto donaldh: shoot.
22:26 donaldh dukeleto: is there anything still failing?
22:27 donaldh dukeleto: (I finally find some time for it this week)
22:27 donaldh dukeleto: I've installed libffi and built/tested parrot. all pass.
22:27 mikehh donaldh: she don't build with g++ and a bunch of codetest failures
22:29 chromatic Is the g++ problem missing casts?
22:30 mikehh looks like
22:30 donaldh k
22:31 dukeleto donaldh: we need help fixing codetests on the gsoc_nci branch, and getting it to build on g++
22:31 donaldh what's the configure option for g++ ?
22:32 dukeleto donaldh: i think you need --cc, --ld and perhaps another. mikehh would know
22:32 mikehh donaldh: --cc=g++ --cxx=g++ --link=g++ --ld=g++
22:33 dukeleto mikehh++
22:34 mikehh I am actually using g++-4.5 on Ubuntu 10.10 because g++ defaults to 4.4 but that is not important
22:35 * donaldh is compiling with g++ on Cygwin and Ubuntu in a VM.
22:36 dukeleto are smoke reports from gcc on gsoc_nci still useful?
22:36 * dukeleto boots his darwinpcc smoke machine
22:37 wagle fail!  waugh, etc etc
22:38 kid51 joined #parrot
22:38 mikehh wagle: oh dear, what were you trying for?
22:39 dukeleto wagle: want to help with git migration stuff?
22:40 dukeleto wagle: are you a filter-branch wizard?
22:41 wagle phone interview with google..  i wasnt addressing the edge cases he was interested in because (1) i kept changing my main cases (2) i try to design to not have edge cases, so i dont cleanly separate
22:42 wagle dukeleto: i play one on tv
22:42 wagle dukeleto: you are "just" using a translation table to change the commit authors?
22:42 kid51 with respect to 'make codetest' on gsoc_nci branch:
22:43 kid51 These are trickier than one might think.
22:43 kid51 Not just a case of trailing whitespace or tabs instead of spaces.
22:43 kid51 I discussed this a bit with plobsing over the weekend.
22:44 kid51 There's a pmc file he is creating, I believe in something called auto::libffi, which works differently from all other .pmc files.
22:44 kid51 He probes for libffi, and then copies one of two templates into position.
22:45 kid51 So this is unusual:  All our other .pmc files are source code, listed in MANIFEST, which gets them subjected to all the codingstd tests.
22:45 kid51 But the .pmc file he ends up with is *not* source code, it's generated source code.
22:45 kid51 Hence, it should not be listed in the MANIFEST -- only its 2 templates should be -- and it should not be subjected to the codingstd tests -- only the templates should.
22:46 dukeleto wagle: yeah, just using an authormap that git-svn reads
22:46 kid51 But, currently, it *is* listed in the MANIFEST, which creates problems.
22:46 kid51 Peter is aware of this.
22:46 dukeleto wagle: i just need the filter-branch invocation to use that authormap
22:46 chromatic Subjecting templates to coding standards seems like a recipe for disaster.
22:46 kid51 He said he was unsatisfied with the design, but has other things to get to first on that branch.
22:46 wagle dukeleto: trying to setup a test now..  give me an hour or so..  (maybe 15 minutes)
22:47 dukeleto wagle: thanks. My current clone with git-svn will probably go another 5-10 hours :)
22:47 kid51 chromatic:  If you look at them, you'll see that they are, in effect, two different versions of the .pmc file.
22:47 kid51 This creates a problem for other testers.
22:47 dukeleto kid51: one for libffi, and the other for without it?
22:48 kid51 If you don't have libffi on your machine, when you do 'svn up', you'll get an update which comes from a machine that *does* have libffi.
22:48 kid51 ... meaning massive conflicts.
22:48 kid51 dukeleto: Yes
22:48 dukeleto kid51: generated code shouldn't be checked into the repo
22:49 dukeleto kid51: so the generated PMC being committed to the repo is the bug
22:49 kid51 Some of these problems *might* go away with git, but the overall design is problematic ... which I think plobsing agrees.
22:49 chromatic Compile-time generation isn't a good approach either.
22:49 dukeleto kid51: git doesn't make ponies appear to fix bad design
22:49 dukeleto as much as I would like that.
22:49 kid51 I also don't know how generating a .pmc will play with config/auto/pmcs.pm
22:50 kid51 dukeleto:  At $job, we transitioned to git last week in my section.  It doesn't pretty up the source code ;-)
22:50 dukeleto Are we under the impression that our codetests are not learned enough to deal with this code?
22:50 kid51 We could fix the codingstd/filenames.t failure in that branch easily.
22:50 kid51 But I think we should hear from plobsing more about what the status of the branch is.
22:51 dukeleto This code is important to get merged. Sooner the better. If our codetests need to be smarter to grok this branch, then maybe we need to TODO those codetests, and fix them in a branch.
22:51 kid51 dukeleto:  True, but this is a new approach to our source code:  .pmc files which are generated rather than part of the Parrot distro.
22:52 kid51 In this instance, the codingstd failures I was getting were the canaries in the coal mine.
22:52 chromatic What about the crypto PMCs?
22:52 dukeleto kid51: i don't think this is that new of a technique
22:53 dukeleto kid51: parts of many pmc's are autogenerated, and this will probably happen more in the future
22:53 dukeleto kid51: what status report do we need from plobsing again? To see if he has more fixes for it?
22:53 kid51 chromatic:  What crypto PMCs?  We took the crypto steps out of Configure.pl some time ago
22:53 chromatic We had crypto dynpmcs which were mostly generated.
22:54 kid51 First, take a look at the libffi config step and the two variations on the .pmc file (templates).
22:55 donaldh Whatever happened to compilers/ncigen ?
22:55 kid51 then, ask him if he intends to keep the "copy_if_diff" way of generating the .pmc.
22:55 dukeleto donaldh: i think that got put to pasture
22:55 dukeleto donaldh: or rather, it flew up to the big parrot branch in the sky
22:55 * donaldh doesn't have libffi 5 on Cygwin. Saves for another day.
22:56 kid51 I ran tests on gsoc_nci several times on both my machines, and every time I did 'svn up', I got a completely different .pmc file.
22:57 kid51 ... because I was pulling something appropriate for a machine that did have libffi, which mine do not (currently)
22:57 dukeleto kid51: like i said. Committing generated PMC's is the bug.
22:57 dukeleto kid51: that needs to get fixed before the branch is merged
22:58 kid51 Well, all I can say is that I think that was on Peter's list of things that still needed to be done.
22:59 dukeleto kid51: good to know. Thanks for working on gsoc_nci
22:59 dukeleto in other news, PL/Parrot is close to becoming an official Fedora package
23:00 dukeleto gerd++ for working on that.
23:03 donaldh Is there an emerging DBD definition? Is that happening in Rakudo?
23:04 * donaldh is hacking on a Sqlite NCI binding.
23:04 jnthn___ donaldh: See MiniDBI, also see Zavolaj.
23:04 jnthn___ both linekd from modules.perl6.org
23:04 jnthn___ is now known as jnthn
23:05 donaldh jnthn: thanks
23:05 tadzik jnthn: you grew quite a tail there :)
23:08 wagle i found the solution, but it doesnt work
23:12 donaldh left #parrot
23:13 wagle git filter-branch --env-filter 'export GIT_AUTHOR_EMAIL=foobar@baz.com'
23:14 dukeleto wagle: i will teach you to sweep, grasshopper
23:14 dukeleto wagle: i need a commit-filter that uses the authormap, and sets all the proper variables for usernames that are in the authormap
23:15 dukeleto wagle: your command filters history and changes every author to foobar@baz.com, which is not as useful ;)
23:17 wagle dark bash magick
23:18 wagle hand me your rewriter, and i'll hand you the black bash magic to do it
23:18 kid51 is now known as kid51_at_dinner
23:19 dukeleto wagle: i need the rewriter. I know how to use bash ;)
23:19 wagle hmm..  alternatively, see http://stackoverflow.com/questions/392332/​retroactively-correct-authors-with-git-svn
23:20 dmalcolm left #parrot
23:20 wagle if you knew bash, you wouldnt have made fun of my vital piece of info..  8-p
23:21 dukeleto wagle: that stackoverflow link is useful tho
23:21 dukeleto wagle: that has the core of what i need
23:21 dukeleto wagle++
23:22 wagle same info, just expanded..
23:23 wagle git filter-branch --env-filter "$(perl-script-to-generate-​huge-author-remap-statement remappings-file)"
23:24 dukeleto wagle: yeah, i needed to see a bit more detail
23:24 dukeleto wagle: we have a script called tools/dev/mk_author_map.pl
23:24 dukeleto wagle: it reads CREDITS and spits out an authormap
23:25 lucian left #parrot
23:25 wagle can you handle this, or is bash environment variable hacking biting you?
23:26 * wagle is a programming language weenie, tranmogrification of programs is his bread and butter
23:27 dukeleto wagle: with that stackoverflow link, i think i am good.
23:27 dukeleto wagle: i am still going to wait to see what the output of my current git-svn clone looks like
23:29 dukeleto the parrot repo was quite a bit bigger before most of the language left the nest.
23:35 whiteknight joined #parrot
23:45 theory left #parrot
23:48 * wagle bets filter-branch would be done by now
23:49 wagle for all branches..  8)
23:49 preflex left #parrot
23:50 wagle oh..  unless your rewriter rewrites some A to B and B to C, etc
23:53 preflex joined #parrot

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

Parrot | source cross referenced