Perl 6 - the future is here, just unevenly distributed

IRC log for #parrotsketch, 2009-09-29

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

All times shown according to UTC.

Time Nick Message
01:33 japhb joined #parrotsketch
12:01 kid51 joined #parrotsketch
12:02 kid51 kid51 reporting:
12:02 kid51 Done:
12:02 kid51 * Closed 3+ RT tickets
12:02 kid51 * Eliminated configuration step auto::funcptr, which was doing nothing in the absence of JIT
12:02 kid51 * Reported failures observed in my own testing and Smolder reports
12:02 kid51 * TT #989 was nominally mine once it was reopened, but I don't know how to resolve it; reassigned to nobody
12:02 kid51 Doing:
12:02 kid51 * In library_files branch, contemplating where best to define lists of test files (TT #1061)
12:02 kid51 Blocking:
12:02 kid51 * Have repeatedly requested, but not received, feedback on TT #509.  Need feedback from people with non-symlinkable systems in order to apply patch from wayland
12:02 kid51 EOR
14:44 Tene Done: git and svn fetch stages for Plumage, and the start of a pure-parrot HTTP client.
14:45 Tene Doing: More plumage work, possibly IO, maybe a Select PMC, Austin asked for import/export for NQP.
14:47 Tene EOR
16:14 japhb Done:
16:14 japhb * http://use.perl.org/~geoffrey/journal/39682
16:15 japhb * Edit https://trac.parrot.org/parrot/wiki/ModuleEcosystem#MetadataProposal for rev 1
16:15 whiteknight joined #parrotsketch
16:15 japhb * Review work from Tene++ and darbelo++
16:15 japhb Will do:
16:15 japhb * Continued Plumage coding
16:16 japhb Blocking on:
16:16 japhb * Lack of sleep and surplus of headache.  Bleah.
16:16 japhb EOR
16:20 NotFound joined #parrotsketch
16:43 mikehh What I did in the last week:
16:43 mikehh * building and full testing parrot on amd64 and i386
16:43 mikehh * fixing test failures
16:43 mikehh * testing language builds on latest parrot, mainly rakudo and partcl
16:43 mikehh * worked on make fulltest and make cover:
16:43 mikehh * they now run all tests run by make test as well as the run-core tests,
16:43 mikehh * codetest, benchmakk_tests, manifest_tests, examples_tests and distro_tests
16:43 mikehh What I intend to do in the next week:
16:43 mikehh * continue testing
16:43 mikehh * write some tests
16:43 mikehh .eor
16:49 cotto_work What I did:
16:49 cotto_work * <this space intentionally left blank>
16:49 cotto_work What I intend to do:
16:49 cotto_work * more than this week
16:49 cotto_work Blockers:
16:54 darbelo joined #parrotsketch
16:55 whiteknight WHAT I DID:
16:55 whiteknight * Looked more at JIT stuff, especially LLVM documentation
16:55 whiteknight * Got a great response to a blog post, a libJIT-based NCI frame builder, looking at that to see if it's feasible to get it into trunk.
16:55 whiteknight * Following along with some GC changes
16:55 whiteknight * Digging into PCC refactor branch test failures. Am certain that adding support for :named and :slurpy return arguments will resolve large numbers of failures
16:55 whiteknight * Tried adding said support but failed and ran out of time.
16:55 whiteknight * Updated the list of TODO items for ongoing IO cleanups to make sure other people know what needs to happen there
16:55 whiteknight WHAT I WANT TO DO:
16:55 whiteknight * Continuing on JIT documentation-reading and planning. Especially trying to get documentation/examples for the LLVM C bindings (the main API is C++)
16:55 whiteknight * Trying to get an LLVM-based NCI thunk generator set up
16:55 whiteknight * Looking over some recent GC changes to see if we can do more optimization/cleanup
16:55 whiteknight * Got a lead on a new memory allocator that might improve Parrot performance, going to be testing that out.
16:55 whiteknight * Debugging on PCC refactors branch. Big push to get that ready.
16:55 whiteknight WHAT I AM BLOCKING ON:
16:55 whiteknight * Severe tuit shortage.
16:55 whiteknight EOR
17:21 darbelo Did:
17:21 darbelo (decnum dynpmcs)
17:21 darbelo * Did some makefile cleanups.
17:21 darbelo * Replaced Configure.pl with Configure.pir
17:21 darbelo (parrot)
17:21 darbelo * Attempted some strstart removals. Found dragons.
17:21 darbelo * Removed _next_for_GC.
17:21 darbelo * Made pmc2c error out on PMCs with pmclass != filename.
17:21 darbelo * Opened ticket to review pmc2c tests.
17:21 darbelo (plumage)
17:21 darbelo * Got sucked into it. Wrote some code.
17:22 darbelo Will do:
17:22 darbelo (decnum dynpmcs)
17:22 darbelo * A DecQuad PMC is imminent.
17:22 darbelo * Might remove DecInt as a failed experiment.
17:22 darbelo (parrot)
17:22 darbelo * Make pmc_freeze.c stop abusing string internals.
17:22 darbelo Might block:
17:22 darbelo * Tuit shortage has been announced for next week.
17:22 darbelo .eor
17:22 darbelo q2q
17:24 pmichaud joined #parrotsketch
17:28 NotFound What I did:
17:28 NotFound * Renamed list.c functions that conflicted with mysql client lib.
17:28 NotFound * NCI parameter handling fixes
17:28 NotFound * Moved example Mysql module out of the parrot repo.
17:28 NotFound * Deprecated Parrot_oo_get_namespace.
17:28 NotFound * Added type safe versions of mark alive.
17:28 NotFound * Miscellaneous fixes, cleaning, and test coverage.
17:28 NotFound What I will do:
17:28 NotFound * No plan.
17:28 NotFound EOR
17:29 pmichaud What I did (past two weeks):
17:29 pmichaud * Had some $otherjob tasks to take care of
17:29 pmichaud * Mainly working on regex engine refactors
17:29 pmichaud * Rewriting the engine in pct-rx branch
17:29 pmichaud * New engine is completely PAST/POST based, unlike PGE
17:29 pmichaud * It should also be quicker, and with a lot less GC pressure
17:29 pmichaud * Have basic operations in place for concatenation, alternation, quantifiers
17:30 pmichaud What I'm doing this week:
17:30 pmichaud * More grammar engine work
17:30 pmichaud * Will be adding full Cursor and Match capabilities soon
17:30 pmichaud * Then protoregexes
17:30 pmichaud * Then parsing/compiling P6 regexes using the new engine
17:30 pmichaud * Then flesh out the rest of the regex engine capabilities
17:30 pmichaud * Then redesign NQP's grammar and implementation to use the new engine
17:30 pmichaud * (Don't know how far I'll get with the above... I just keep plugging away at it)
17:30 pmichaud What I'm blocking on:
17:30 pmichaud * Insufficient usable hours per solar day
17:30 pmichaud EOR
17:39 Util joined #parrotsketch
18:08 jrtayloriv joined #parrotsketch
18:12 jrtayloriv WHAT I DID:
18:12 jrtayloriv -- Started working on a Scheme compiler, "phemer"
18:12 jrtayloriv -- Bombarded Austin++ with hundreds of questions about PAST/PCT
18:12 jrtayloriv BLOCKERS
18:12 jrtayloriv -- Not knowing Scheme well enough
18:12 jrtayloriv -- Just barely understanding PCT
18:12 jrtayloriv EOF
18:14 chromatic joined #parrotsketch
18:14 chromatic I fixed some bugs.
18:14 chromatic I helped tune the GC and clean up some of its code.
18:14 chromatic I found yet more evil in IMCC.
18:14 chromatic This week I will do as much as possible to fix IMCC's line number tracking.
18:14 chromatic I will also post some ideas on pruning some of our structs.
18:18 allison joined #parrotsketch
18:23 barney joined #parrotsketch
18:30 Util # Done
18:30 Util * TT#600 (bytecode testing framework) - Wrote up a first draft. Posted just now.
18:30 Util # Not Done:
18:30 Util * No progress on TT#994 or TT#389 (git-svn auto-props, and keeping subs with :method out of the namespace).
18:30 Util # Plan for next week:
18:30 Util * Resume work on TT#994 or TT#389.
18:30 Util # Blockers:
18:30 Util * None.
18:30 Util .end
18:30 chromatic Hello.
18:30 whiteknight hello
18:30 cotto_work hi
18:30 Tene HI!
18:30 barney hi
18:30 mikehh hello
18:30 pmichaud hello
18:30 allison - Created a git branch for merging trunk with pcc branch. That's done (and pushed to a github repository), but I haven't created a new subversion branch yet because I'm getting a segfault on Mac OSX (which may have been there before the merge).
18:30 allison - Was in Portland this past week for LinuxCon, so spent some time talking with chromatic about project blockers and roles and responsibilities.
18:30 allison - Gave a talk at a Bay Area Python group and met with the Unladen Swallow developers (an implementation of Python using the LLVM JIT).
18:30 allison EOR
18:30 allison and hello
18:31 darbelo Hola.
18:31 jrtayloriv greetins'
18:31 NotFound hola
18:32 chromatic Let's talk about last week's priorities.
18:32 chromatic Exception testing?
18:32 whiteknight nothing that I am aware of
18:33 chromatic MultiSub testing?
18:34 mikehh there were some additions to exception tests
18:34 dukeleto joined #parrotsketch
18:34 chromatic I saw someone mention coverage numbers.  Was that NotFound?  Can whoever's done that document how to get those numbers on the wiki?
18:34 whiteknight I say we roll those two over. Need more tests on MultiSub and Exception in any case
18:35 pmichaud I vote we table them.
18:35 cotto_work didn't we get some PMC to 100% coverage?
18:35 dukeleto yes, i added some coverage for exception, but it needs more
18:35 NotFound chromatic: make cover, or http://tapir2.ro.vutbr.cz/cover/cover-results/
18:35 dukeleto equality of exception objects is a big chunk of coverage that is needed
18:36 NotFound We have now 100% on some Array PMCs
18:36 chromatic That's a useful link.
18:37 NotFound Don't know who does it, somone put the link on #parrot a few weeks ago
18:38 cotto_work mj41 iirc
18:38 chromatic Move on to new PMCs or keep working on these two?  Votes?  Suggestions?
18:38 pmichaud I'd like to see us prioritize PCC.
18:38 NotFound pmichaud: me also
18:39 chromatic Alright then, let's talk about PCC.
18:39 chromatic Allison?
18:40 allison I have a fresh trunk-applyable patch ready for a new subversion branch.
18:40 allison I'm doing some testing now, but could apply it to a new branch right after #parrotsketch.
18:40 darbelo Ship it!
18:40 NotFound +10
18:40 allison A show of hands who'd rather work in subversion versus git on this one?
18:40 Util +1 SVN
18:41 darbelo +1 either way.
18:41 NotFound I'm unable to work with git
18:41 pmichaud I think svn will win this one.
18:41 allison subversion it is, I'll take all the help I can get
18:41 whiteknight yay!
18:41 * whiteknight giggles with excitement
18:41 NotFound Go, go, go!
18:41 pmichaud I'd prefer git, but I'm not likely to be extremely active on committing at first, and I can certainly work okay with svn.
18:41 chromatic What's the status of the patch, allison?
18:41 whiteknight Git in baby steps
18:42 pmichaud I'd rather keep the svn users (NotFound, allison, Util) productive on this one.
18:42 mikehh I can go either way
18:42 allison chromatic: cleanly applies to  trunk, segfaults in the config PMC generation on OSX
18:42 whiteknight allison: I've got a lot of test failures traced back to a lack of :named and :slurpy returns
18:42 whiteknight we get those added, I think we'll have a much cleaner test output
18:42 allison chromatic: haven't been able to test on Linux because my linux laptop bit the dust on the plane to London
18:42 whiteknight I'll test on Linux tonight
18:43 mikehh I can test that
18:43 whiteknight it does build there, fails lots of tests but does build
18:43 chromatic Any other discussion of that branch and our prioritization?
18:44 whiteknight nope
18:44 pmichaud +1, thanks everyone
18:44 chromatic Okay.  Other priority suggestions for the week?
18:44 Tene There are several people who can provide linux boxes for testing, I'd expect, including me.
18:45 chromatic Does anyone know the status of pirc?  I'm running into places I can't patch IMCC without a severe refactoring.
18:45 allison chromatic: stalled, AFAIK
18:45 whiteknight last I heard, it was having PBC problems
18:45 whiteknight floating point values, mostly. Needs testing
18:45 Tene chromatic: Incomplete.  What's there works okay, but it needs several more things.
18:45 whiteknight otherwise, kjs was optimistic
18:45 allison yes, mainly just blocking on his time
18:46 allison there may be a way to spread that work out
18:46 whiteknight I think a concerted team effort oul get it moving again
18:46 Tene It was fairly easy for me to get started on when I looked at it.
18:46 chromatic I'd like to suggest that we focus on that soon.
18:46 whiteknight agreed
18:46 chromatic Otherwise I suspect it'll slip past 2.0.
18:46 chromatic It may anyway, but I don't want IMCC to live any longer than it has to.
18:46 allison chromatic: we already said it would slip past 2.0
18:46 allison 2.6 at the earliest
18:47 darbelo Somebody needs to check in the generated c for those of us without bison.
18:47 Tene I tried to fix some of the problems, but I ran into issues about knowledge of parrot internals and what some of the things mean.
18:47 chromatic 2.6 is the earliest supported release in which it can be the default, or the earliest any kind of release in which it can be the default?
18:47 whiteknight that's what I get blocked on too: esoteric internals details
18:47 particle tene: did you take notes, or comment the code?
18:47 cotto_work sounds like a good next target after pcc
18:47 Tene particle: I made a couple of commits, iirc... lemme check.
18:47 allison chromatic: it's the earliest any kind of release where it can be used as a full replacement for imcc
18:47 mikehh I think a concentrated effort can get it done sooner
18:48 allison mikehh: probably, but we would have to balance that against other roadmap items currently listed for 2.0
18:48 whiteknight I agree, I think we could have it as a viable alternative by 2.0 if we focused
18:48 allison (i.e., we can't do everything)
18:48 whiteknight (2.0 is a busy milestone)
18:48 Tene particle: looks like no.
18:49 chromatic My suggestion for this week: let's a couple of us figure out its current state and make a list of tasks needed to make it the default.
18:49 whiteknight I would much rather we focus on JIT, if we have to choose
18:49 allison hll interop is a higher priority than pirc, for example
18:49 particle (lost knowledge)--
18:49 mikehh I think rirc is a priority, just as jit was/is
18:49 particle tene++ for checking
18:49 allison mikehh: indeed, it is
18:49 allison mikehh: they're all priorities, we just have to rank them
18:50 Tene Is there any interest in starting to have regularly scheduled hackathons?  "On X datetime, everyone work on Y" for items like pcc, pirc?
18:50 whiteknight Tene: I would be very much in favor of that
18:50 chromatic +1, Tene
18:50 darbelo +1
18:50 allison Tene: that could be a good motivator
18:50 pmichaud we're doing mini-hackathons here in the north Dallas area and it's worked out well.
18:50 chromatic Suggestions for a time?
18:51 whiteknight Weekends are most permissive, I think.
18:51 NotFound Did we really need pirc? If we can have a PCT based pir compiler, that may be good enough.
18:51 Tene I'm always able to be available every saturday, any time.
18:51 whiteknight I personally have a hack marathon on Saturday
18:51 allison NotFound: that has bootstrapping problems, and the fact that PCT is SLOW
18:51 chromatic PCC hackathon this Saturday, then.
18:51 whiteknight +1
18:52 pmichaud allison: what part of PCT is slow, here, ooc?
18:52 chromatic Let's move the PCT discussion to #parrot.
18:52 kurahaupo joined #parrotsketch
18:52 allison NotFound: so, not necessarily pirc, but a low-level fast PIR/PASM compiler as an alternative to imcc
18:53 pmichaud note that if we can get POST->PBC translation in place, then the need for PIR compilation becomes much smaller.
18:53 Tene NotFound: iirc, one of the things pirc was still needing was some part of pbc generation, which any pir compiler would need.
18:53 particle tell that to the hll folks not using pct
18:53 allison pmichaud: indeed, though still needed (since PCT is written in PIR)
18:53 allison pmichaud: for now :)
18:53 pmichaud allison: PCT can be rewritten... :)
18:54 pmichaud it's explicitly designed to be nqp-able
18:54 allison PIR or PASM will never go away though
18:54 pmichaud sure, I agree
18:54 chromatic Shall we review our milestones now?
18:54 chromatic HLL interop?
18:55 allison needs someone to focus on it for a week or two, as I understand it
18:55 allison as in, close to done, but not getting attention
18:55 pmichaud I'm listed at the point person for that, but I doubt I'll be the person to focus on it in October.  If someone else wants to take the lead, I'm happy with that (and will help as the PCT-related issues come up)
18:55 chromatic Anyone willing and able?
18:56 whiteknight I don't understand it either
18:56 Tene Take the lead on HLL interop?
18:56 whiteknight can we get some good documentation about the outstanding tasks?
18:56 NotFound Or even not so good?
18:56 Tene I'm pretty sure I'm supposed to be the lead on HLL interop issues.
18:56 pmichaud if nobody does it I *will* get around to it before 2.0, but regex refactor has nearly all of my attention at the moment, since so much (parrot, nqp, rakudo) is blocking on that at the moment.
18:57 chromatic Tene, are you volunteering or revolunteering?
18:58 Tene chromatic: Both!
18:58 allison I'd like to see someone else working on it, even if pmichaud can come back to it later
18:58 pmichaud Tene is the logical choice.
18:58 allison (just a matter of spreading effort)
18:58 allison sounds like it's in good hands
18:59 Tene If anyone else wants to be brought up to speed on HLL interop, they can ask me.
18:59 chromatic dukeleto, Parrot debugger documentation?
18:59 allison Tene: if you can break the work down into simple tasks on a wiki, you may get more help fast
18:59 Tene allison: I'll look into it.
19:00 allison (it's not always possible)
19:00 dukeleto chromatic: i think if someone could tell me what they wanted to see in the debugger docs, i could 'finish' the task, but right now, it is ambiguously specc'ed
19:01 Tene AFK; I need to lunch now, before I run out of time. :P
19:01 chromatic Do you have a list of tasks you can perform in the debugger?  We could brainstorm some use cases.
19:01 allison dukeleto: focus on helping people getting started in using it
19:01 dukeleto allison: ok, I think we definitely need more of a debugger "user guide" than a manual page. good idea
19:02 pmichaud dukeleto++
19:02 chromatic japhb, seed libraries?
19:02 allison dukeleto: a short manpage is also good (so we can ship it with the distros), but that should be fast
19:02 japhb chromatic, sorry, still spending all effort on Plumage.
19:03 japhb On the plus side, do have a couple helpers now.  :-)
19:03 NotFound I moved Mysql from examples to an external repository
19:04 japhb NotFound, have you made a Plumage metadata JSON for that?
19:04 NotFound japhb: not yet, lacked the time to learn the format.
19:05 chromatic Bytecode testing, Util?
19:05 japhb NotFound, no problem.  I can help you when you have spare time.
19:05 Util I put my thoughts into TT#600, but I am not claiming the ticket (yet?). The real work looks to be in answering the outstanding questions on how it should function, which I expect IRC discussion to resolve.
19:06 chromatic Can you lead a discussion of those issues on #parrot?
19:06 Util Yes
19:06 chromatic Excellent.
19:06 chromatic Finally: pruning C data structures.  I'll publish some notes on that soon.
19:07 chromatic Next question: what's blocking us in the long term, either technically or socially?
19:07 allison q2qs
19:07 whiteknight A "where to start" point for JIT
19:07 whiteknight although I'm getting closer to that
19:08 chromatic Other thoughts?
19:08 NotFound strings
19:08 allison whiteknight: I had a good conversation with the unladen swallow team, will send you a summary
19:09 particle chromatic: revision of roadmap/milestone items is due
19:09 whiteknight allison: awesome
19:09 chromatic Strings, NotFound?
19:09 NotFound No work in the string branch for months
19:09 chromatic Okay.
19:09 allison you mean implementing NFG?
19:10 NotFound And renaming functions, and deleting deprecated ones, and fix lack of documentation...
19:10 whiteknight I really should look into that problem more. Bit twiddling and buffer managment is right up my alley
19:11 allison NotFound: the big branch there was merged in
19:12 NotFound allison: I'll write a ticket with the main current pitfalls.
19:12 chromatic particle, what did you mean?
19:12 allison NotFound: (the renaming interface cleanup, that is)
19:12 allison NotFound: excellent
19:12 particle the ROADMAP developed at pds2008 has proven invaluable in setting and achieving our near-, mid-, and long-term development goals.  incremental updates to the ROADMAP over the past ten months have met our near- and mid-term needs, but no longer meet our long-term needs.  it's time for a full review of the ROADMAP.
19:12 chromatic Okay, good point.
19:12 chromatic Other blockers?
19:12 allison PDS 2009
19:12 pmichaud will there be a PDS 2009?
19:12 Util A "where to start" point for really hacking Parrot; e.g. understanding the Parrot and PCT subsystems. Is the .str thinking even documented anywhere?
19:13 Util We have a PIR book, but if all the non-book .pods were spliced together and read, would one get a coherent picture of *every* corner of Parrot and PCT?
19:13 chromatic Good one, Util.
19:13 particle pmichaud: yes, however it may be virtual, if we can't get funding to make it face-to-face.
19:13 allison pmichaud: I list it as a blocker because it needs someone to plan it. We should have one,  yes.
19:13 pmichaud I was going to suggest a virtual PDS
19:13 pmichaud if we can't have a f2f one
19:13 allison that works too
19:13 chromatic Anything else?
19:13 darbelo q2q
19:14 pmichaud Util: I plan to draft someone to do a PCT book when I'm done with the current set of refactors
19:14 chromatic Let's move on to questions.
19:14 chromatic Allison?
19:14 Util pmichaud: very glad to hear it
19:14 dukeleto 1q1
19:14 allison Longish first q:
19:14 NotFound BTW we have a new platform: PacoLinux has been able to build parrot on Haiku.
19:14 dukeleto or q1q rather
19:14 allison chromatic and I are both reading a book called "The E-Myth Revisited". It's about running a small business, but a lot of the concepts apply equally well to an open source project.
19:15 allison Talking about that and about Parrot, we hit on the idea of adding a new role in Parrot, called something like "Development Coach". The role would involve keeping an eye out for pain points and blockers in the development process, "Are there ways we could do things better, cleaner, more efficiently?"
19:15 allison This would be chromatic's role, he's already doing it now, but it's valuable to mark it as a specific role, both so people inside the project recognize what he's doing and why it's important, and so people outside the project see the value we place not just on good code but on good practices.
19:15 allison It has an added advantage that it allows me to focus more on the architectural vision role, where we're going, why we're going there, and what are the most important steps along the way.
19:15 chromatic (and so I *remember* to put on a hat and do it)
19:15 pmichaud +2
19:15 whiteknight +1
19:15 dukeleto +1 to dev coaches
19:15 allison What do others think of adding a Coach role? Does this sound like a good idea?
19:15 NotFound +1
19:15 darbelo +5
19:15 Util +1
19:15 particle *2
19:15 pmichaud and congratulations, chromatic.
19:16 dukeleto what about "new developer" coach, is that a different role?
19:16 chromatic I'd like to think anyone and everyone can coach new developers.
19:16 allison dukeleto: that's mentoring
19:16 particle dukeleto: we have individual mentors for that
19:16 allison and, we have a number of people doing it
19:16 mikehh excellent
19:16 allison but, worth recognizing them too
19:17 dukeleto mentoring++
19:17 chromatic Next question, allison?
19:17 allison Sounds like we have general approval, excellent!
19:17 allison Second q: On git, chromatic and I propose to table the migration discussion for now.
19:18 Util +1
19:18 allison We can't migrate until after 2.0 or maybe even 3.0 for stability of development process anyway.
19:18 cotto_work for now long?
19:18 particle agreed, even though it's not a question :)
19:18 allison But, we have a process modification to consider as a compromise.
19:18 allison (the compromise is the question)
19:18 dukeleto is there a time-frame decided for when it will be de-tabled for discussion?
19:18 dukeleto 2.0 ?
19:19 pmichaud wait, let's hear the compromise first.
19:19 pmichaud (proposed compromise)
19:19 particle ...people are jumpy today...
19:19 allison What's the general thought on opening the parrot branching strategy up, so a particular development task could be done either with an SVN branching techique, or with a pure git branching technique?
19:19 allison The developers involved can decide what branching strategy to use.
19:20 pmichaud I'm not quite sure how that would work (from a technical perspective)
19:20 particle whatever works, but i'd prefer all external branches become svn branches before trunk merges
19:20 cotto_work allison, immediately?
19:20 allison Git branching is quite painless if the main developer takes a branch from trunk, and the other developers fork off that main branch and collaborate directly through git.
19:20 pmichaud sure
19:20 dukeleto allison: +1 for each developer choosing svn branch or git branch. it is easy enough to provide a diff of a git branch for svn folk to test
19:20 pmichaud but how do those get back into svn?
19:21 pmichaud just a diff of the branch point?
19:21 Util Having the branches be public, rather that chance them being only on someone's laptop, is worth a lot.
19:21 Tene pmichaud: I can work through a test of it with you tonight.
19:21 allison (It does mean the one developer takes responsibility to merge the main branch into trunk when the time comes)
19:21 cotto_work +1 to the proposed compromise
19:21 Util s/that/than/
19:21 particle i say create svn  branch, commit git branch there, then handle svn trunk merge
19:21 dukeleto pmichaud: by creating a patch
19:21 particle it gives existing smokers something to chew on
19:21 pmichaud dukeleto: so, roll up all of the git commits into a single diff that is then applied to svn (branch or trunk)
19:21 allison But, then, we already have one person taking that responsibility to merge in the SVN branch, so it's not a big change.
19:21 particle we don't have git smoke infrastructure
19:22 dukeleto pmichaud: yes
19:22 pmichaud how does one re-sync the git branch with parrot trunk?
19:22 Tene pmichaud: or apply all of the git commits as svn commits via git-svn
19:22 allison particle: using git on an svn branch actually reduces a good bit of the usefulness of git
19:22 pmichaud Tene: I'm thoroughly unimpressed with git-svn at the moment.  It took 30 minutes for people to try to explain it to me and I couldn't follow it.
19:22 chromatic We can figure out the right workflow depending on the branch.  The important ideas are 1) allowing experimentation with public Git branches and 2) recommending that all committers try Git at least once or twice.
19:22 dukeleto pmichaud: git pull origin
19:23 particle allison: when it's an integration candidate with parrot trunk, i'd like to be able to take advantage of the smokers, and to have the history in one repo
19:23 dukeleto pmichaud: we should avoid git-svn if posssible
19:23 allison pmichaud: if it's used well, it's no trouble, but the current "recommended" technique for using git-svn on the parrot wiki is not ideal
19:23 pmichaud please, if I may have the soapbox for just a second
19:23 * particle just dropped git-svn at $work, due to continued branch merging pain
19:23 dukeleto particle: shall I create a TT for investigating git smoking infrastructure?
19:23 pmichaud I have a very real application for this today
19:23 particle dukeleto: sure!
19:23 allison pmichaud: application for what?
19:24 pmichaud git branch versus svn branch
19:24 pmichaud current I'm doing my development on the new regex engine in an svn branch
19:24 pmichaud I _tried_ to do it with a git branch this last week, but was totally unable to come up with a clean way to do it
19:24 allison a git branch from trunk, or a git branch from the svn branch?
19:24 pmichaud either.
19:25 pmichaud creating the git branch is easy enough
19:25 pmichaud but then resync with trunk isn't obvious
19:25 allison it's worked well for me
19:25 particle the svn barrier prevents git svn merges from being very useful
19:25 allison the easiest is to format-patch and apply that to trunk
19:25 pmichaud and I wasn't sure how to get the git stuff back into svn
19:25 pmichaud sure, I can format-patch and apply to svn trunk
19:25 allison though, you can commit back to trunk directly
19:25 pmichaud but I'm still not sure how to get recent trunk commits into git branch
19:26 allison git fetch
19:26 pmichaud is this assuming that we did a git-svn checkout?
19:26 chromatic Can our resident Git experts come up with a decent workflow we think will work?
19:26 allison git svn fetch
19:26 pmichaud okay.
19:26 allison it's quite clean
19:26 pmichaud I will be happy to work on a workflow with a git expert
19:26 particle no, it's not clean
19:27 particle when filenames change, or big hunks of code are refactored, it totally breaks down
19:27 pmichaud I'd really prefer to use git, but I kept getting conflicting advice
19:27 allison (git rebase, btw, is considered bad news by the linux devs)
19:27 particle because you can't do git hunk-based merging, you're stuck with svn file-based merging
19:27 pmichaud agreed, I tend to want to avoid "git rebase" as well.
19:27 chromatic pmichaud, dukeleto, Tene, can you come up with some workflow for us?
19:28 allison The idea at this point is just to open it up for experimentation.
19:28 allison It may turn out that git is no good for us.
19:28 pmichaud I'm fine with that.  As I said, I have a very real case to work with.
19:28 allison But, it's worth a try.
19:28 pmichaud I disagree with "git is no good"
19:28 pmichaud (for us)
19:28 particle with the understanding that you can't spell 'experimentation' without 'pain', i'm for it
19:28 allison We just don't know yet.
19:29 chromatic Okay.  darbelo, question?
19:29 pmichaud (I have to leave now...back in 30)
19:29 darbelo Can I get some input on the pmc2c tests?
19:30 * darbelo is searching for TT #.
19:30 chromatic I don't find them useful myself.  I've changed the generator a few times and broke the tests and the tests didn't tell me anything interesting other than "you broke the tests".  Parrot continued to pass the tests that mattered to me.
19:31 darbelo TT #1063
19:32 NotFound Tests that looks at minutiae of the code are plain wrong.
19:33 cotto_work I reiterate my reply to the ticket that the tests aren't generally useful and should be deleted except where they can be made agnostic of pmc2c's specific internals.
19:33 cotto_work basically, +1 to clobber
19:33 allison I'm okay with ripping those tests out.
19:33 darbelo 'kay. Clobbering it is.
19:33 allison It is a good reminder how important it is to fully test our core pmcs and dynpmcs, though.
19:34 allison (since they are acting as the tests for the generator too)
19:34 chromatic Next question, darbelo?
19:34 darbelo I hat 2?
19:34 cotto_work better make one up then
19:34 darbelo Oh yeah. The plumage CLA thing.
19:35 darbelo f plumage is going to be (C) The PaFo, that implies that all contributors need a CLA, right?
19:35 allison darbelo: all committers would, yes
19:35 japhb darbelo, I thought we decided that was a "yes"
19:35 allison is it going to be PaFo?
19:36 japhb allison, that's what it is now ... was I wrong to do that?
19:36 japhb I thought it would make eventual merge smoother
19:36 darbelo japhb: Just making sure. We had no 'official' answer there IIRC.
19:36 allison japhb: nope, highly welcome, and will simplify codistribution
19:37 chromatic dukeleto, one question?
19:37 dukeleto i have booked a discussion at the gsoc mentor summit called "dynamic language interoperability"
19:38 dukeleto does anybody want any specific things discussed there?
19:38 dukeleto particle chaired it last year, iirc
19:39 darbelo "When are y'all gonna stop playing and move to our vm?"
19:39 particle wish i could be there to do it again....
19:39 chromatic "Why do you hate America?"
19:39 dukeleto darbelo: exactly.
19:40 dukeleto particle: i will get as much feedback before/during from you as I can.
19:40 japhb chromatic, "'Cause, um, ... it's America, and ... um ... you know?"
19:41 japhb ;-)
19:41 chromatic Any final questions?
19:41 dukeleto where can I get parrot stickers?
19:41 * dukeleto is addicted to stickers
19:42 dukeleto i have a sticker printing shop a few blocks from my house, I can look into it if this has not been researched yet
19:42 allison iirc, pmichaud was setting up a parrot cafepress shop
19:42 Util pmichaud is absent, but was working on some such
19:43 chromatic Sticker printing is likely cheaper than cafepress, with the initial inventory overhead, of course.
19:43 allison they make nice stickers, in additon to tshirts
19:43 allison but, would be interested in prices for printing, too
19:44 allison dukeleto: if you get a quote for a full-color sticker run, send it on to the foundation directors
19:45 chromatic Let's wrap this up then.  Good work, everyone.
19:46 Util parrotsketchers++
19:46 Util left #parrotsketch
19:47 NotFound left #parrotsketch
19:47 chromatic left #parrotsketch
19:50 PacoLinux left #parrotsketch
19:51 darbelo left #parrotsketch
21:49 Whiteknight joined #parrotsketch
22:21 cotto_w0rk joined #parrotsketch
22:22 ascent joined #parrotsketch

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