Camelia, the Perl 6 bug

IRC log for #parrot, 2009-08-07

Parrot | source cross referenced

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

All times shown according to UTC.

Time Nick Message
00:01 tetragon joined #parrot
00:02 whoppix joined #parrot
00:12 patspam joined #parrot
00:21 zak_ joined #parrot
00:45 Whiteknight joined #parrot
00:47 TiMBuS joined #parrot
00:48 satrac joined #parrot
00:50 Whiteknight I love running an svn up and having lots of files be updated!
00:50 Whiteknight it means things are changing!
00:55 * cotto adds trailing null bytes to all files in svn
00:55 cotto including the checked-in pbcs
00:56 Whiteknight whatever, that's fun to see too
01:09 * Coke made a run through DEPRECATED.pod today, but most things are waiting for feedback.
01:10 eternaleye joined #parrot
01:27 ruoso left #parrot
01:34 kid51 joined #parrot
01:44 tetragon joined #parrot
01:57 Andy joined #parrot
02:00 dalek parrot: r40438 | allison++ | trunk/docs/project/debian_packaging_guide.pod:
02:00 dalek parrot: [debian] Remove no-longer-used step from the Debian packaging guide.
02:00 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40438/
02:00 dalek parrot: r40439 | allison++ | trunk/docs/project/ubuntu_packaging_guide.pod:
02:00 dalek parrot: [ubuntu] Change to more common changelog entry.
02:00 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40439/
02:04 Andy joined #parrot
02:06 petdance joined #parrot
02:08 petdance_ joined #parrot
02:28 treed bacek: Still here?
02:28 purl here is fine, wait until she looks at it
02:29 treed msg bacek Ah, thanks for the confirmation. I wonder why it stopped working: http://gist.github.com/163680 Your help is appreciated.
02:29 purl Message for bacek stored.
02:42 janus joined #parrot
02:43 theory joined #parrot
02:45 Andy joined #parrot
03:02 JimmyZ joined #parrot
03:26 dukeleto joined #parrot
03:42 wayland76 joined #parrot
03:58 dalek TT #905 created by allison++: [RFC] Deprecate Eval PMC
04:00 theory joined #parrot
04:22 bacek_at_work treed: ping
04:24 bacek_at_work msg treed I'll fix parrot today-tomorrow so StringIterator.shift_pmc will work again. OTOH replacing ".local pmc item" with ".local string item" should work right now. And it's faster anyway.
04:24 purl Message for treed stored.
04:25 Coke if I clone a git repo that is a clone of an svn repo, what happens when I git svn dcommit?
04:27 chromatic I don't think it works.  I could be wrong.
04:30 Coke I would think I'd have to 'git commit' and then have someone dcommit on the clone in the middle.
04:31 chromatic I agree.
04:31 Coke msg dukeleto you might want to clarify what's going to happen if someone clones your clone, in terms of dcommits.
04:31 purl Message for dukeleto stored.
04:38 dukeleto Coke: good point
04:39 dukeleto Coke: i don't quite trust git-svn, so I have been using git to keep track of my work and then patching it onto a parrot svn repo. this is probably unfounded paranoia and I need to actually get my git-svn workflow working
04:41 chromatic If you thought Git's documentation was confusing... git-svn will abuse you.
04:41 dukeleto chromatic: indeed. the implementation will torment your
04:41 dukeleto s/your/you/
04:46 dukeleto Coke: if someone clones my clone, dcommiting will do nothing because my git repo knows nothing of the svn repo
05:09 tekky joined #parrot
05:39 Andy joined #parrot
05:41 TiMBuS joined #parrot
06:00 jhelwig joined #parrot
06:02 uniejo joined #parrot
06:12 he joined #parrot
06:17 uniejo joined #parrot
06:29 dukeleto woohoo, just fixed another bug in parrot_debugger. currently it gets a Bus error if you attempt to print string registers which have not come into existence yet
06:30 moritz dukeleto: maybe you should also post the relevant parts of your parrot/.git/config on the wiki
06:30 moritz dukeleto: so that others know how to make git-svn work with it
06:31 dukeleto moritz: i am not quite sure that *I* know how to get git-svn to play nice, but I will try. currently I am using git to organize my work but I end up using "git diff --no-prefix" to generate a patch and just "patch" an up-to-date svn tree and commit directly with svn. but I am starting to feel comfortable with git-svn, so this will change
06:32 moritz dukeleto: I can do it, then
06:32 dukeleto moritz: that would be awesome
06:33 dukeleto moritz: the thing that bothers me is that I would like to use git-svn with an svn branch and that seems to be very shaky ground
06:33 moritz dukeleto: I know that bacek++ does that successfully
06:34 dukeleto moritz: interesting. I would *really* like to know his workflow
06:34 moritz dukeleto: maybe we can convince him to share his experience on the wiki too
06:35 dukeleto moritz: as far as I can see, if I try to use git-svn on a repo that was not started with git-svn, nothing works. git-svn just hangs. other people tell me I am crazy and that it should work, but that is what I know right now. I have been meaning to right a test for git-svn, but I haven't gotten around to it...
06:39 moritz dukeleto: so it seems your parrot git-svn clone doesn't contain all the branches, right?
06:40 moritz Tene++ made a clone that contains the branches, I could upload that to github
06:40 moritz that would be even more awesome, I think
06:40 dukeleto moritz: git fetch will pull in all branches
06:41 moritz all the remote svn branches?
06:41 moritz my clone is still running...
06:41 dukeleto moritz: my full git svn clone took over 3 days to finish
06:42 moritz I meant me cloning your git clone ;-)
06:43 dukeleto moritz: oh. yeah, it seems that I didn't push all the different branches to github, but I will attempt to fix that now.
06:43 dukeleto moritz: just cloning my git repo should only be a few minutes
06:43 moritz right, finished
06:43 moritz pushing all the branches would be great
06:44 dukeleto moritz: i will attempt to that. i think I may need to do it from the original physical repo that I used "git svn" on, which is on a different machine
06:46 moritz afaict there's no git command for pushing all branches, so you might need to iterate over .git/svn/origin/* or so
06:47 dukeleto moritz: interesting. i will attempt to do this when I have access to the machine tomorrow
06:52 moritz it seems that getting git-svn working on a cloned copy involves more than just adapting .git/config
07:02 Tene my parrot git-svn directory is 750M
07:02 Tene that's with the compiled stuff too, though.
07:02 Tene lemme realclean.
07:02 moritz Tene: there's one huge file in your .git/svn/ tree
07:02 Tene 666M after realclean. ;)
07:03 moritz Tene: it's 170M here (based on your tree)
07:03 Tene 514944.git/svn/vtable_morph_change@36392
07:03 wayland76 joined #parrot
07:03 Tene that one file is 503M
07:03 moritz it's not really worth shipping a 400M file with metadata that's only interesting in one particular branch
07:03 moritz oh, 503 even ;-)
07:04 Tene What is it, I wonder?
07:04 moritz I don't know, but the repo seemed to work fine after removing it
07:28 pmichaud Good morning, #parrot
07:32 mokurai left #parrot
07:48 chromatic http://plausible.org/nasal/doc.html
07:53 dalek rakudo: e5a63db | moritz++ | src/setting/Operators.pm:
07:53 dalek rakudo: infix:<eqv> for undefs
07:53 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/e​5a63dbe2a537b88ba6f908567758de7e67499a0
07:58 nopaste "NotFound" at 213.97.96.43 pasted "Ugly fix for TT #218" (21 lines) at http://nopaste.snit.ch/17493
08:00 NotFound This patch shows the weakness of the PMCProxies
08:24 dukeleto joined #parrot
08:24 dukeleto joined #parrot
09:01 cognominal joined #parrot
09:17 MoC joined #parrot
09:37 kj joined #parrot
09:38 HG` joined #parrot
09:49 bacek joined #parrot
09:52 payload joined #parrot
09:55 bacek treed: ping
09:57 wayland76 joined #parrot
09:57 bacek treed: cardinal's string "failures" were fixed in parrot at r40132 and r40142  long time ago... You can just uncomment tests and they will pass.
10:08 mikehh joined #parrot
10:15 donaldh joined #parrot
10:35 tekky left #parrot
10:40 muixirt joined #parrot
10:43 Khisanth joined #parrot
10:45 JimmyZ joined #parrot
10:48 kid51 joined #parrot
11:21 donaldh joined #parrot
12:02 kj joined #parrot
12:05 kj left #parrot
12:05 kj joined #parrot
12:42 tetragon joined #parrot
12:50 ruoso joined #parrot
13:15 HG` joined #parrot
13:25 Khisanth joined #parrot
13:32 Khisanth joined #parrot
14:12 payload joined #parrot
14:38 whiteknight joined #parrot
14:45 Khisanth joined #parrot
14:48 Psyche^ joined #parrot
14:55 Andy joined #parrot
15:10 dalek decnum-dynpmcs: r140 | darbelo++ | trunk/src/pmc/decint.pmc:
15:10 dalek decnum-dynpmcs: Make DecInt's VTABLEs call SUPER and then round the result as appropiate.
15:10 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=140
15:20 donaldh joined #parrot
15:25 jdv joined #parrot
15:36 dalek decnum-dynpmcs: r141 | darbelo++ | trunk/src/pmc/decnum.pmc:
15:36 dalek decnum-dynpmcs: Replace calls to decNumberRemainder() with calls to decNumberRemainderNear(),
15:36 dalek decnum-dynpmcs: which is closer to what we want.
15:37 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=141
15:43 mikehh codetest FAIL, All others PASS (pre/post-config, smolder, nqp_test, fulltest) at r40439 - Ubuntu 9.04 amd64
15:44 mikehh DEPRECATED.pod fails t/codingstd/pod_syntax.t - PATCH in http://nopaste.snit.ch/17491
15:45 cotto somebody's just itching for a commit bit
15:48 cotto mikehh++
15:48 mikehh I am not sure about that :-} - let somebody else take the risks - anyway I submitted a CLA
15:51 dalek parrot: r40440 | cotto++ | trunk/DEPRECATED.pod:
15:51 dalek parrot: [codingstd] fix POD errors in DEPRECATED.pod
15:51 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40440/
15:52 jan joined #parrot
15:52 dukeleto mikehh: congrats, it usually doesn't take long for them to approve you
15:53 mberends joined #parrot
15:56 davidfetter joined #parrot
15:56 hercynium joined #parrot
16:29 treed joined #parrot
16:37 whoppix joined #parrot
16:51 dalek rakudo: 4665d75 | masak++ | src/setting/ (2 files):
16:51 dalek rakudo: [src/setting/Str.pm] a first shot at .encode
16:51 dalek rakudo: Currently, it dies when fed non-ASCII characters, because the relevant
16:51 dalek rakudo: Parrot parts haven't been fully exposed yet.
16:51 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/4​665d75269f43c8255fa4d2b4723c48b53bab077
16:51 dalek rakudo: 7717c4a | masak++ | :
16:51 dalek rakudo: Merge branch 'master' of git@github.com:rakudo/rakudo
16:51 dalek rakudo: review: http://github.com/rakudo/rakudo/commit/7​717c4a91beda30befb34a9999022beb1a748f11
16:54 darbelo joined #parrot
16:59 treed joined #parrot
17:12 chromatic joined #parrot
17:13 dalek TT #906 created by allison++: [PIR] Assignment syntactic sugar restricted to destination parameters
17:15 iblechbot joined #parrot
17:19 dalek parrot: r40441 | allison++ | trunk/DEPRECATED.pod:
17:19 dalek parrot: [cage] Link deprecated item to more relevant ticket.
17:19 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40441/
17:25 treed joined #parrot
17:29 mokurai joined #parrot
17:30 dalek parrot: r40442 | allison++ | trunk/src/library.c:
17:30 dalek parrot: [cage] Remove 'runtime/parrot' from library and include path searches,
17:30 dalek parrot: search under 'runtime/parrot/library' and 'runtime/parrot/include'
17:30 dalek parrot: instead. Resolves TT #511.
17:30 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40442/
17:38 dalek TT #511 closed by allison++: [CAGE] deprecate searching runtime/parrot for libraries and includes
17:38 cotto From C, what's the best way to get the full name of the namespace of a Sub PMC?
17:43 particle as a string?
17:43 particle you'll need to build it
17:47 cotto yay
17:48 particle do you need to worry about aliased namespaces?
17:48 particle that is, can you build towards the root from the sub, or must you build from the root towards the sub?
17:49 chromatic Use Parrot_Context_get_info().
17:53 cotto chromatic, thanks.  That does the trick.
17:53 chromatic I remembered using it when I wrote the proof of concept.
17:57 cotto which proof of concept?
17:58 chromatic The first time I did a profiling core.
17:59 cotto It's coming in very handy.
18:00 cotto inner runloops make for all kinds of fun.
18:01 chromatic Exactly.
18:06 cotto Are inner runloops a common problem with interpreters or is Parrot just special?
18:07 chromatic Parrot's pretty special in that it calls in and out of C so much, but I can imagine similar fun in other VMs when calling back and forth into extensions.  (Shared library callbacks, ugh.)
18:11 Infinoid In libfuse.so (and soviet russia), shared libraries callback into YOU.
18:16 Khisanth joined #parrot
18:17 cotto On a completely unrelated topic, is there any reason a PGO build isn't on the roadmap?
18:20 dalek TT #907 created by allison++: [CAGE] refactor Parrot_mmd_sort_candidates
18:21 dalek tracwiki: v10 | moritz++ | git-svn-tutorial
18:21 dalek tracwiki: mentioned a tar archive including all branches + svn meta data; git svn create-ignore
18:21 dalek tracwiki: https://trac.parrot.org/parrot/wiki/git-​svn-tutorial?version=10&amp;action=diff
18:21 chromatic PGO?
18:21 allison https://developer.mozilla.org/en/Buil​ding_with_Profile-Guided_Optimization
18:21 allison I'm assuming
18:22 allison cotto: because we only have so many tuits
18:22 allison cotto: but anyone is certainly welcome to try it
18:23 cotto PGO is profile-guided optimization
18:23 cotto PGO is also https://developer.mozilla.org/en/Buil​ding_with_Profile-Guided_Optimization
18:23 purl okay, cotto.
18:23 chromatic I tried it; didn't make a huge difference in some naive tests.
18:24 chromatic I think our Makefile trickeries hung it up a bit.
18:24 cotto Our build is special.
18:39 dalek cardinal: 9165470 | treed++ | t/string/ (2 files):
18:39 dalek cardinal: get_pmc_keyed on String is working again, remove todo/skip
18:39 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/91654709532ed73ab1748cfa50c00a2206425455
18:39 dalek cardinal: 31c9263 | treed++ |  (2 files):
18:39 dalek cardinal: Add String.each_char.
18:39 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/31c9263aa388ccfab0198fa27e7cee3bb273aa42
18:39 dalek cardinal: 7f9db1a | treed++ |  (3 files):
18:39 dalek cardinal: Implement String.empty? with tests.
18:39 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/7f9db1a9be93827bf25a459a5e2f9450bdc70ee3
18:39 dalek cardinal: 34e0fdb | treed++ |  (2 files):
18:39 dalek cardinal: Fix String.each to actually work as it should, also have it respect $/. Fixes #17.
18:39 dalek cardinal: review: http://github.com/cardinal/cardinal/commit​/34e0fdb725a46bc43ed4f147131c445287f20e1a
18:40 dalek decnum-dynpmcs: r142 | darbelo++ | trunk/ (3 files):
18:40 dalek decnum-dynpmcs: Move the version() METHOD from DecNum to DecNumContext. It looks like it should
18:40 dalek decnum-dynpmcs: have been there from the beginning.
18:40 dalek decnum-dynpmcs: Update our usage example to match.
18:40 dalek decnum-dynpmcs: review: http://code.google.com/p/decnu​m-dynpmcs/source/detail?r=142
18:40 pmichaud allison: ping
18:42 allison pmichaud: pong
18:42 pmichaud ah.  I was concerned about r40442 (removing runtime/parrot), but I misread the ticket.  <latella>Never mind.</latella>
18:43 pmichaud (I was thinking that runtime/parrot/library was removed from the path... but you just removed runtime/parrot)
18:43 allison pmichaud: ah, aye
18:43 pmichaud we missed you in lisbon.  :-|
18:44 allison yeah, I'm bummed I didn't make it. Hopefully lots of Perl 6 stuff got done. :)
18:44 pmichaud It did.
18:45 pmichaud lots of excitement about the rakudo announcement, and we have a really good plan for getting to next April
18:45 pmichaud jnthn++ and I were reviewing the task list and we think it's very doable.  But we're definitely going to want the calling convention refactors just as soon as Parrot can get them.  :-)
18:48 allison excellent
18:48 allison yes, calling conventions are top on my list
18:48 pmichaud \o/
18:49 allison I've got the extract from the old branch ready, will create the new branch this afternoon
18:49 pmichaud oooooh
18:49 pmichaud I'm definitely looking forward to seeing it.  :-)
18:49 allison the upside to being stuck in the US waiting for a visa is I have absolutely nothing to do for the next few weeks but sit in a hotel and work on parrot :)
18:50 cotto I love that upside!
18:50 * pmichaud realizes that his PIR book should be waiting for him when he gets home.  :-)
18:51 pmichaud I'm eager to learn about PIR... I think I might be able to do some cool things with it.  :-)
18:51 allison heh :)
18:52 allison history shows you're able to do some of the coolest things ever with it :)
18:52 pmichaud Well, I'm hoping that's just the prelude to even more stuff.  :)
18:53 pmichaud I'm also hoping that others are able to do cooler things with Parrot than I've done.
18:53 particle the pir book is surprisingly slim
18:53 particle much smaller than k&r
18:53 * pmichaud wonders if B&N at the airport is carrying the book.  :-P
18:54 allison particle: aye, it's a "Pocket Guide"
18:54 allison (but we can't use that term, since O'Reilly has printed so many)
18:54 cotto Is that a parrot in your pocket?
18:54 particle the firefox colors on the cover are an interesting choice
18:55 particle and the lack of a parrot logo
18:55 pmichaud okay, I should probably release this table in the restaurant for other paying customers
18:55 particle surely you could have gotten written permission :)
18:56 pmichaud back later (maybe)
18:56 particle stick the landing!
19:08 whoppix joined #parrot
19:11 whiteknight allison: ping
19:20 donaldh joined #parrot
19:26 mberends joined #parrot
19:31 dalek parrot: r40443 | NotFound++ | branches/auto_attrs/src (7 files):
19:31 dalek parrot: set auto_attrs in several critical PMCs
19:31 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40443/
19:47 allison whiteknight: pong
19:47 whiteknight hey allison, did you get my email a few days back about the book?
19:47 allison particle: the "firefox colors" are a parrot wing
19:47 whiteknight a "pocket guide" is good, but something bigger would be better :)
19:47 allison particle: (a rather abstract one) actually, it's the background of the logo from parrotblog.org, repurposed
19:48 allison whiteknight: I did, but haven't had time to respond to it
19:48 particle ...knew it looked similar...
19:48 whiteknight that's fine, just wanted to make sure it wasn't lost in the ether
19:48 allison we don't need an expanded book on PIR
19:48 whiteknight also, if you need any help whatsoever in pcc_rewiring I'm at your service
19:49 allison and what most people are really going to want is not a guide to the internals of the VM, but how to write a language, or how to write libraries
19:49 particle i think the most important parrot document eventually will be how to install it
19:50 allison so, I have my eye on two more small guides "Parrot Developer's Guide: Compiler Tools" and "Parrot Developer's Guide: Modules"
19:50 particle hopefull that document is so short, nobody has to read it
19:50 allison particle: aye, but that's a few short lines at the beginning of each book "How to Get Parrot"
19:51 particle yes, i mean to say, our ultimate audience is folks who don't care that parrot is there, just that their hll works
19:51 allison particle: yup, agreed
19:51 whiteknight okay, new books is a good idea too. I'm fine with whatever so long as I have a plan to follow :)
19:52 allison whiteknight: pmichaud is going to work on the Compiler Tools book, but might be interested in help
19:52 allison whiteknight: some of the content for the Modules book could be pulled out of the remaining draft chapters.
19:52 whiteknight yeah, i've drafted a lot of stuff on it that idn't make it into the PIR book, so that's a start
19:52 whiteknight right
19:52 whiteknight so creating these in docs/pct and docs/modules?
19:53 allison docs/book/pct and docs/book/modules
19:53 whiteknight ah yes, that's what I meant
19:53 allison aye
19:53 allison a quick chapter shuffling out of draft could be a good start
19:54 whiteknight Okay, good. I'll see if I can hit that up tonight
19:55 joeri joined #parrot
20:06 ruoso joined #parrot
20:08 ruoso joined #parrot
20:09 tekky joined #parrot
20:09 tekky left #parrot
20:10 ruoso joined #parrot
20:33 dalek parrot: r40444 | allison++ | branches/pcc_arg_unify:
20:33 dalek parrot: Creating branch for refactoring calling conventions, rewiring the internals of pcc dispatch to use a single argument passing strategy.
20:33 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40444/
20:33 cotto arg!
20:44 japhb allison: sorry I didn't respond to your email in the Parrot "standard libraries" thread, but I seem to have never received it.  No matter, reading it in the archives now.
20:48 allison japhb: curious, still having trouble with your ISP filtering?
20:49 japhb allison: I thought I had that turned off, and I did get pmichaud's reply to your message just now (which is how I knew you'd sent one)
20:50 dalek parrot: r40445 | allison++ | branches/pcc_arg_unify (9 files):
20:50 dalek parrot: [pcc] Merging in first round of changes from pcc_rewiring branch to
20:50 dalek parrot: fresh branch.
20:50 dalek parrot: review: https://trac.parrot.org/parrot/changeset/40445/
21:00 japhb allison: I think I understand your objection to a single module repository -- would you instead support a single place to *find* modules?  In other words, a repository of just module metadata, and including in Parrot the necessary tools to create and upload a metadata file when you are ready to "publish" your spiffy new module (wherever it happens to be hosted), and conversely for users to search against that metadata repo, select a metada
21:00 japhb ta file, and call the appropriate tools to download the matching module source from sourceforge, github, etc., and configure/build/test/install?
21:03 darbelo japhb: Wouldn't that be just as easy to implement as an external tool?
21:04 japhb darbelo: what exactly do you mean by "external"?
21:04 particle japhb: you mean freshmeat.net?
21:05 japhb particle: No, I mean real metadata.  Like META.json + MANIFEST + ...
21:05 japhb And I'm thinking of something like masak's 'proto' tool on steroids.
21:06 particle i wonder how much of that could rely on rss feeds from various existing repositories
21:06 darbelo japhb: Not shipped with parrot, and not necessarily maintained by the same people.
21:06 * japhb is trying to find "rough consensus" so that we can then produce "working code"
21:07 particle iwbni all the data were managed externally, and simply aggregated
21:07 japhb particle: Only if said RSS feeds were capable of including only "releases", and also could be taught to contain at least links to the proper metadata files.
21:07 japhb iwbni?
21:07 purl iwbni is "it would be nice if"
21:09 japhb particle: that centralizes the task of writing aggregation tools, but does have the benefit of removing a step from the module author's "publish" workflow ....
21:09 particle correct
21:09 particle the site "discovers" the module, the author doesn't "publish"
21:10 japhb particle: well, remember, there'd be no per-release work for the module author other than 'make publish'
21:11 particle assuming they use make
21:11 particle and a special makefile template
21:11 japhb And we could even give them a reason to do it, by including the ability to squawk about the new release and have that fed directly to a "new modules" planet.
21:11 japhb Hmmm, of course, that could happen in the discover mode too, I guess.
21:12 japhb particle: I was using 'make publish' in the general sense of 'single simple command' rather than that exact syntax.  Sorry for the confusion.
21:12 particle ah, ok
21:13 particle i'm thinking that module authors tag their distros as 'parrot' or similar, and the discovery engine picks out README, CHANGES, AUTHORS, etc
21:14 japhb darbelo: can you explain in more detail why you don't want the Parrot people involved in the module publishing game?  I would think our deprecation cycle means that we may want to have some non-trivial input into that, if not outright control ...
21:14 japhb particle: Would you agree with specifying the naming and format of those magic files?
21:16 Tene joined #parrot
21:17 * japhb also a bit worried that discovery will become a difficult to scale operation, in the sense of pounding on github et al to the point they get a bit annoyed at us.  I wonder if the various source repos have usable push APIs (I'm not talking RSS, I'm talking something that's efficient machine to machine, and pushed rather than polled)
21:17 japhb Oh, and the other problem, is that discovery only works if the source is hosted in a public place.
21:19 japhb Otherwise, the module author has to publish a tarball at least, and then they need some place to put it, and running discovery against every company ftp site won't work, so they need some way to notify the discovery engine to look for the new item.  Which brings us back to 'make_publish'
21:21 darbelo japhb: I ussually think of parrot as a low-level, reusable component. I don't expect the parrot devs to care about the distribution of whatever modules run on their vm anymore than I expect the (extreme example) gcc devs to care about the distribution of the various open source projects written in C.
21:24 mikehh_ joined #parrot
21:32 japhb darbelo: Sure, I can see that ... but Parrot is somewhat special in that a large pillar of its motivation is to promote module sharing, between the HLLs (in the case of HLL-coded modules) and from Parrot to the HLLs (in the case of NCI libraries).  In the C world, the closest equivalent I can think of are the desktop projects (GNOME, KDE, freedesktop.org), which all have some way of marking a project 'official', marking releases and bundle
21:32 japhb s, and so on.
21:35 particle japhb: standard/recommended filenames/formats, yes
21:35 particle japhb: there will always be a darkpan, it can't be avoided
21:36 particle errands &
21:36 japhb particle: The darkpan is not what I meant
21:36 particle if the source isn't public, then what is it?
21:36 japhb particle: I was talking about businesses that do (for example) contribute modules to CPAN, but their VCS repos are behind the corporate firewall.
21:37 mberends joined #parrot
21:37 japhb particle: Thus, they already have to publish a snapshot externally somehow; we need to support that use case.
21:38 japhb Open source != open source *control*, as several large companies are perfectly good (bad?) examples of.
21:48 darbelo japhb: I see that, but what I would expect the modules to have an independant release cycle. The release cycle for the nci wrapper of library FOO should follow the release cycle of library FOO, not that of parrot.
21:49 darbelo Tying both cycles together sounds like a recipe for pain.
21:49 japhb darbelo: Oh sure, I never meant that libraries would follow parrot's release cycle -- I meant that when we cross a deprecation boundary, we need to make it easy for our users to find the version of a module compatible with the version of parrot they have installed.
21:50 japhb In other words, we should make it possible for a Parrot 2.0 and a Parrot 3.0 user to both have an easy time finding the proper module releases they can install.
21:51 japhb And I think we may have to help with that; specifying metadata for compatibility and conflict resolution, for example.
21:52 japhb I mean sure, a 3rd party sufficiently clued into Parrot's workings could indeed get all that right.  But for *now* I suspect having us at least somewhat involved will ease the bootstrap.
21:53 japhb And I also suspect that until one of us JFDIs, no one is going to magically come along and do it for us.  :-)
22:01 darbelo I should probably make clear that I'm not against a centralized Parrot Module Repo. It's making it the One True Parrot Module Repo that makes me have second thoughts.
22:08 japhb darbelo: Just so I'm clear on where your dividing line is ....  In the Perl world, does CPAN bother you?  Or the fact that perl comes with cpan, which only knows about CPAN, and no other module repo?
22:08 japhb (That's an honest question, I'm trying to find the line between "centralized" and "One True")
22:10 darbelo I'd have to say "that perl comes with cpan, which only knows about CPAN".
22:14 darbelo But CPAN has a mitigating factor, in my particular case, for perl modules available in the ports collection I don't have to care that cpan exists.
22:15 japhb BSD user?
22:16 darbelo Yup, OpenBSD.
22:17 japhb I'm in a similar place, because I run Debian -- lots and LOTS of CPAN is packaged.
22:18 japhb Usually, I can find the perl package I want with apt, and if I can't, there's always CPAN search.  At most two places to look -- but even so, I find myself somehow annoyed when I have to go the CPAN search.
22:20 Limbic_Region joined #parrot
22:20 japhb But at $work, where I have to use distros that don't package the world (as Debian tries to), I go nuts trying to find stuff across lots of different RPM repos.  In fact, I often find it easier to badger Dag to package something for RPMforge than to go find it myself.
22:20 japhb I'd be really happy if parrot didn't make people day-to-day unhappy like that.
22:20 japhb Which is why, minimally, a central place to search seems key.
22:21 kid51 joined #parrot
22:23 japhb FPAN: "Federated Parrot Archive Network" ... where the "Federation" is the central metadata & search.
22:25 he left #parrot
22:26 japhb This idea is actually a win in terms of allowing the Parrot HLLs to see the Parrot NCI libraries right alongside the pure-HLL libraries from PEAR, Eggs, etc.
22:28 darbelo Hmm, but where do you draw the line for HLL-specific stuff. I would expect Rakudo to have it's own module repo, separate from whatever parrot provides.
22:28 Tene One statement I'd rather question from Allison's recent mail... "there's very little value in
22:28 Tene creating a CPAN-like module repository"
22:28 Tene CPAN is a CPAN-like module repository.  Does that mean that there's very little value in CPAN?
22:29 japhb darbelo: I meant, if we federate just the metadata, then Rakudo users don't need to know or care whether they are pulling from a Parrot repo, or from a Perl 6 repo, or a Rakudo-specific repo, or directly from some github project ...
22:33 Tene japhb: you ever use cpanspec to package your own rpms from cpan?
22:35 japhb Tene: nope.  Usually the stuff I need from CPAN has sufficiently finicky prereqs that a knowledgeable person figuring out library version interactions is helpful -- I'm not really an RPM guy, I can just cargo-cult a SPEC occasionally.
22:39 rg joined #parrot
22:41 kid51 cotto?
22:41 purl cotto is probably Christoph Otto <mailto:christoph@mksig.org>
22:44 japhb On #toolchain, it was pointed out that CPAN mirrors are often far more reliable than, say, github.  So offering mirroring isn't such a bad idea after all ....
22:46 chromatic I (personally) don't want to be in the business of managing/administering such a thing.
22:48 tetragon joined #parrot
22:50 japhb chromatic: "such a thing"?  mirrors or central meta data repo or both?
22:51 cotto kid51?
22:51 purl kid51 is now going to live dangerously.
22:51 chromatic Mirrors.
22:51 purl mirrors are like that
22:54 japhb chromatic: Ah OK.  Well, I don't see mirrors as something we need to provide, merely that we should support the possibility that a VCS site might be mirrored by a third party, and let the market take it from there.
22:54 japhb Does that make sense?
22:55 kid51 cotto:  Just sent you mail re your new config step question.
22:56 cotto kid51, great.  Thanks.
22:57 chromatic Sure, I much prefer that we allow this to happen rather than trying to do it ourselves up front.
22:58 cottoo joined #parrot
22:58 Casan joined #parrot
22:58 japhb As long as we make it possible, I think that's good enough for a start.
22:59 chromatic Exactly.
23:03 allison Tene: I wouldn't say CPAN has little value, but I would say that CPAN is part of why Perl is losing relevance in the wider world. People only go there if they already use Perl.
23:04 japhb OK, sounds like some consensus is forming in general.  Now for the next level of detail: what tools and standards do we need to provide?  I'd say at a minimum we need to standardize metadata format and semantics, and the names (and possibly formats) of any special files such as README, CHANGES, etc.  And we (by which I mean the community, not the Parrot core team) need to provide a search server and a web and API interface to it.
23:05 japhb I would *like* to support at least a CLI tool (like a smarter cpan shell) that makes search and download/build/install easier than the manual method.
23:13 hercynium joined #parrot
23:18 darbelo japhb: (CLI tool)++
23:20 donaldh joined #parrot
23:24 jhelwig joined #parrot
23:25 dalek TT #908 created by allison++: [CAGE] Relocate xconf/ examples to examples/ directory
23:38 mberends joined #parrot
23:52 mikehh examples_tests FAIL, All Others PASS at r40445 (pre/post-config, smolder, nqp_test, fulltest) - Ubuntu 9.04 amd64
23:53 mikehh t/examples/pir.t - Failed test:  16
23:53 mikehh lokking at it now
23:59 mikehh looking

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

Parrot | source cross referenced