Camelia, the Perl 6 bug

IRC log for #parrot, 2013-02-14

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 kid51_ joined #parrot
00:18 benabik joined #parrot
00:36 kid51 joined #parrot
01:08 Reini joined #parrot
01:15 Eddward joined #parrot
01:30 kid51 joined #parrot
01:44 kid51_ joined #parrot
03:49 Coke ugh. go to remove an unused PMC. the PMC has a copy of sorts in t/ that is used for testing things that are not functional, but instead test infrastructure scripts that are only used to modify source code that we check in (and then test)
03:50 Coke eh. I'll leave the copy for the dubious test.
04:11 Reini joined #parrot
04:12 dalek parrot/sixparrot: 4b0d213 | coke++ | / (14 files):
04:12 dalek parrot/sixparrot: Remove FixedBooleanArray and ResizableBooleanArray
04:12 dalek parrot/sixparrot: review: https://github.com/coke/parrot/commit/4b0d21305e
05:53 dalek rakudo/nom: fdbba70 | moritz++ | / (2 files):
05:53 dalek rakudo/nom: RT #107992: make $_ writable in -n/-p code
05:53 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/fdbba70ad4
05:53 dalek rakudo/nom: 8369e63 | moritz++ | / (2 files):
05:53 dalek rakudo/nom: speed up eqv for Buf
05:53 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/8369e63a8f
06:06 cotto Coke: is there any reason not to work on a branch of parrot/parrot?
06:07 sorear hmm?  isn't that what he's doing?
06:09 cotto it's a separate fork
06:10 sorear #perl6 has expressed a complaint that the Parrot full history takes too long to download, and the idea of rebooting the repository has been raised
06:10 sorear maybe that's what Coke is doing
06:14 cotto Throwing away history seems like the wrong way to solve that particular problem.
06:15 cotto It's a valid concern though.  I noticed it too.
06:15 benabik It's a normal fork.
06:15 benabik There's no particular reason to _not_ work in an individual repository though.
06:15 cotto Sorry.  I was referring to the suggestion that sorear brought up.
06:16 benabik Yes.  I was saying he hasn't thrown away history.  Github reports it as a normal fork, which I rather doubt it would do if he had started afresh history-wise.
06:17 benabik It also has the normal pile o' branches.
06:34 zby_home joined #parrot
07:07 Mike-PerlRecruiter_ joined #parrot
07:46 bouncy joined #parrot
08:25 Reini joined #parrot
08:36 xcombelle joined #parrot
08:51 cotto draft blog post:  I'll post it tomorrow.  https://gist.github.com/cotto/4951406
09:24 not_gerd joined #parrot
09:24 not_gerd cotto++
09:25 not_gerd cotto: perhaps one should mention that while supporting all dynamic languages is no longer an immediate goal, it might be put back on the roadmap long-term
09:25 not_gerd and in unrelated news: I'm keeping a list of stuff to work on: https://gist.github.com/gerdr/4754609
09:25 TonyC joined #parrot
09:41 cotto not_gerd: one problem with migrating opsc to nqp is that nqp has a non-trivial runtime.
09:42 cotto I was leaning toward resurrecting the old p5 ops mangler, if one ends up being necessary at all.
09:42 cotto The irony is not lost on me.
09:43 cotto It may also be the case that we can remove enough ops that maintaining the C code manually becomes feasible.  Whether that's a good idea is left as an exercise to the reader.
09:44 cotto Ah.  pm already commented
09:45 not_gerd cotto: Parrot currently checks in the generated C code for it's core ops
09:45 not_gerd I'm not sure if that can be done with dynops as well
09:45 not_gerd but you can just move the necessary dynops to core
09:46 cotto Are you talking about git add, because that works fine.
09:46 cotto It's late.
09:46 not_gerd cotto: I was thinking of checking in the generated dynop sources will lead to problems with ops numbering or so
09:46 * not_gerd has no idea what he's talking about
09:47 cotto not really.  dynops are numbered numbered when they're loaded unless I'm forgetting something
09:48 not_gerd so jst check-in the generated NQP dynop files and you've solved the bootstrapping problem
09:49 not_gerd unless I'm missing something here, of course ;)
09:50 cotto that plus nqp's runtime libraries
09:53 cotto I wonder if there's any reason to keep dynops around other than for testing.
09:53 cotto (in Parrot core)
09:54 cotto either nqp needs an op or it doesn't
09:54 not_gerd my reasoning as well
09:54 not_gerd however, teh Rakudo ops should probably stay dynamically loadable
09:55 cotto yes
10:12 * not_gerd needs to leave, but will keep an eye on the backlog
11:33 MikeFair_ joined #parrot
12:02 Reini joined #parrot
13:04 brrt joined #parrot
13:25 Coke we need to keep the ops2c code, or we'll have to regen the C based code from the ops code, and no one wants that.
13:25 Coke or rather, we'll lose the ability to write .ops files.
13:25 Coke I'm working in a fork because the fork predates the bandwagon.
13:25 Coke (by ... minutes, probably0
13:44 dngor joined #parrot
13:55 ilbot2 joined #parrot
14:07 PacoAir joined #parrot
14:16 bluescreen joined #parrot
14:21 Psyche^ joined #parrot
14:25 atrodo I dreamed last night about removing parrot code.  Vividly, actually staring at the screen editing files...
14:31 uvtc joined #parrot
14:35 uvtc cotto: in your blog post gist, line 31, "isn't attainable" --> "isn't realistically attainable given Parrot's resources" ?
14:38 uvtc cotto: oxford comma missing on line 38.
14:45 uvtc cotto: unsure what you mean by "induced complexity" on line 52. Maybe "without increasing complexity elsewhere" or "without requiring an increase in complexity of Rakudo"?
14:45 uvtc Oh, what's that word ... not "residual" but ...
14:48 uvtc maybe "without inducing collateral complexity".
14:48 Reini joined #parrot
14:48 Reini joined #parrot
14:53 uvtc Curious (actually, I'm *most* curious to hear about allison 's electric car mentioned the other day), ... anyhow, curious: why have monthly releases anyway? Why not instead just let Rakudo releases refer to a particular Parrot git commit?
14:54 uvtc (see also gist lines 66-68.)
14:54 PerlJam uvtc: and we'd call that commit "a release"  :)
14:56 uvtc Oh, so releases would simply be determined by what Rakudo is using in *its* releases. Makes sense.
14:58 PerlJam Though it could make a strange inversion ...  currently, Rakudo times its releases to the Parrot releases; they're always a couple of days after the Parrot release.  In the near future it might be more like "Rakudo wants to release, let's tag the commit they're using and release it"  :-)
14:58 uvtc cotto: line 24, s/and even it/and even *it*/  .  Also, line 71 s/time that I've/time that *I've*/.  (Seems like those sentences need to pick a part to be emph --- IMO).
14:59 uvtc Heh. "See rule #1". :)
14:59 uvtc PerlJam: ^^
15:01 uvtc cotto: line 78: add comma after "reader".
15:02 PerlJam uvtc: you know you can fork the gist and edit it right?
15:02 uvtc cotto: line 83: maybe mention repo and branch here? (Somewhere maybe there should be instructions for building and using sixparrot head with the latest Rakudo+nqp.)
15:03 uvtc PerlJam: Didn't think I'd be making so many suggestions. Also, I thought some suggestions, if placed here, might be useful for discussion.
15:05 uvtc What, you don't like this channel getting clogged up with editorial minutae?! :)
15:06 PerlJam It doesn't bother me a bit, but I figured given the number of changes you've offered so far, I figured it would be easier to just edit the doc and let people look at the diffs.
15:08 uvtc cotto: last one: if you think Parrot can offer a simple "good enough" Perl 5 (and C library) interop story, they could stand to be mentioned, IMO.
15:09 uvtc PerlJam: For the minor edits, will do that next time. Thanks.
15:13 uvtc Parrot marketing request: would be neat to see graphically how the size of Parrot is changing over time. Particularly over the coming months.
15:14 PerlJam That doesn't seem like a particularly good marketing tool unless it's juxtaposed with some other improvements.
15:14 PerlJam "Parrot shrinks 20%, gains 50% speed increase"
15:14 PerlJam stuff like that
15:18 uvtc Maybe even better. Overlay a plot of performance on the same graphic. Watch as size goes down and perf goes up over time. :) Though, my hunch is that "spring cleaning", or "removing cruft" or whatever would attract potential contributors on its own, to some extent.
15:18 pmichaud good morning, #parrot
15:18 pmichaud (housecleaning can wait a bit... interesting backscroll to read :)
15:26 uvtc Oh, phoo. It's spelled "minutiae".
15:32 dngor joined #parrot
15:34 dmalcolm joined #parrot
15:42 Reini joined #parrot
16:13 Reini joined #parrot
16:41 pmichaud a couple of comments on cotto's draft
16:42 pmichaud in the section at the bottom identifying the two goals of judging any potential changes -- there perhaps needs to be an anti-goal that says that if a change is significantly detrimental to nqp/rakudo, don't make it.
16:42 pmichaud e.g., any change that moves a parrot maintenance activity to become a nqp maintenance activity might not be accepted (e.g., moving ops2c into the nqp repo)
16:44 pmichaud I'm not intending this to be a "don't break nqp" edict, as we'll accommodate changes where we can easily do so.  Just don't make it too hard on nqp/rakudo, lest we decide not to follow parrot
16:45 pmichaud I do recommend that parrot continue to make monthly releases.  Having a cadence is a huge developer benefit.
16:48 pmichaud keeping some form of ops2c is vital.  I've looked at the C code that is generated and there's no way we (nqp) want to be maintaining that by hand.
16:50 uvtc Reminds me of Asimov's Three Laws of Robotics. :)
16:51 uvtc "Parrot may not injure Rakudo/NQP, or through inaction allow Rakudo/NQP to come to harm..."
16:51 pmichaud from a "what can we do to attract developers" perspective, I suspect that "clean up parrot" doesn't get the project very far.  Right now Parrot has a very uncertain short-term future, because of performance bottlenecks.  Until those are addressed, Parrot's future continues to be highly uncertain and will be a disincentive for new developers to come in.
16:56 cotto ~~
16:56 uvtc I've seen that a few times here. What does "~~" mean?
16:57 cotto waving hello
16:57 PerlJam smartmatch *rimshot*
16:57 uvtc hehe
17:00 uvtc My guess is that there are a wide range of things that attract a contributor to a VM project, only one of which is the VM's performance.
17:01 pmichaud uvtc: that's not the point.
17:01 pmichaud uvtc: how many people want to contribute to a project that has a short lifetime?
17:03 uvtc pmichaud: Few, I agree.
17:04 pmichaud so, parrot currently has one major client (rakudo/nqp), and rakudo/nqp are looking for other platforms primarily because parrot doesn't provide adequate performance.
17:05 pmichaud (there are other reasons, but that's the biggie)
17:05 pmichaud so, until parrot provides adequate performance, its future is highly uncertain.
17:08 uvtc Possibly related: I just started up a Clojure repl. It took at least 5 seconds to come up. In contrast, it takes about 2 seconds on my machine for perl6 to come up.
17:09 uvtc If folks want mult-threaded high performance, I'm sure Rakudo~JVM will be compelling. But Rakudo~Parrot may not end up serving the same audience.
17:12 mtk joined #parrot
17:13 Reini joined #parrot
17:16 PerlJam uvtc: if Parrot retrains itself to be the best ever VM-for-Rakudo, it might accidentally serve the same audience
17:18 pmichaud my point is I'm wondering what audience Rakudo-Parrot will serve without significant performance improvements.
17:19 uvtc Rakudo-JVM would in particular serve the audience who are tied (for whatever reason) to the JVM. Maybe that's what the company uses and supports, or existing codebase is in Java, ...
17:24 PerlJam So ... there's a parrot release scheduled for next week.   What will that parrot look like?  Or will it be skipped to get parrot in better shape after trimming the fat?
17:24 uvtc pmichaud: Something I mentioned the other day here: one problem (concern?) with using a language on the JVM is that there's continuous gentle pressure to use Java-alternatives to the libs you'd actually rather use. "Want a GUI? Eh, tricky to use GTK -- go with Swing. Want SQLite? Eh, you're better off using H2 or Derby." Etc.
17:24 uvtc pmichaud: some folks just don't want to use the JVM.
17:25 uvtc pmichaud: also, as I alluded to previously, Rakudo-Parrot would appeal to folks who want that quick start-up time for scripts.
17:28 pmichaud uvtc: you seem to be positing a world where people have already decided to use Rakudo, and have the choice of Rakudo-JVM or Rakudo-Parrot.  That's a pretty small world from where I sit, and fairly short-term.
17:30 pmichaud at any rate, we don't have to dissect this to the nth degree.  I just maintain that until Parrot can solve its performance problems, it's going to steadily be less and less relevant to Rakudo/NQP.
17:35 uvtc pmichaud: this topic comes up in the Clojure community because it's a language that's quite symbiotic with its platform, and because there are already at least 2 implementations on different platforms (jvm and js). Each impl has different strengths, and you can choose which impl to use based on what's important to you. Performance isn't the only determining factor.
17:39 * pmichaud is reminded why he doesn't hang around #parrot that much.
17:39 uvtc pmichaud: Oh, sorry. :( I hope that didn't come off as rude.
17:40 pmichaud I'm writing about what Rakudo/NQP sees as being most important from Parrot, and you're telling me it's not really that important.  Been there, done that.
17:40 cotto pmichaud: is there a list of specific pain points anywhere?
17:41 uvtc Oh, I see. I thought you were writing from an end-user's perspective, rather than from Rakudo/NQP's perspective.
17:42 pmichaud cotto: calling conventions continues to be the main suspect.
17:42 moritz too many GCables per invocation
17:43 pmichaud well, the number of gcables wouldn't have to be a problem, if gc were faster
17:43 cotto Great.  We have a starting point.
17:43 rurban gc is fast enough
17:43 Reini joined #parrot
17:44 rurban too many GCables is the problem (waist management)
17:44 rurban oh, waste management
18:13 Reini joined #parrot
18:14 Psyche^ joined #parrot
18:27 PerlJam Now that all is silentish again ...
18:27 PerlJam what will next week's parrot release look like or will the release be skipped to get Parrot in better shape?
18:27 Coke I see no reason to discontinue the monthly releases at this point, unless there is no activity on trunk. (no point in releasing something that is identical to the previous version)
18:28 cotto +1
18:28 cotto I'm quiet because I'm technically at $dayjob.
18:30 uvtc Does coke/parrot/tree/sixparrot currently work with Rakudo HEAD?
18:30 PerlJam Any projections on when the parrot feature ejection will reach its end?  (just curious)
18:31 PerlJam er, I guess they aren't "features" really if they're not needed :)
18:31 Psyche^ joined #parrot
18:42 Psyche^ joined #parrot
18:44 Reini joined #parrot
18:49 moritz uvtc: there's an easy way to find out
18:50 uvtc I was going to re-phrase the question as, "Can anyone please point me to instructions for building coke's sixparrot for use with Rakudo HEAD?", but
18:51 uvtc then recalled that someone here mentioned the other day that changes may currently be going on in more than just that repo/branch.
18:54 moritz you can build the sixparrot parrot just like any other, install it, and then configure rakudo with --with-parrot=$path_to_parrot_binary
19:00 pmichaud (repeating a parrot release comment from #perl6):
19:00 pmichaud PerlJam: (parrot release)  fwiw, NQP is currently tied to Parrot 4.10.0, and I doubt we'll bump it this month.
19:00 pmichaud so a Parrot 5.1.0 release (or the lack of one) isn't likely to affect Rakudo/NQP much.
19:01 cotto pmichaud: last I checked nqp built fine against parrot master
19:01 cotto so the bug is either fixed or re-hidden
19:02 pmichaud cotto: that's good.  we tend to not bump revisions until there's a compelling reason to do so
19:02 not_gerd joined #parrot
19:02 pmichaud otoh, moritz++ points out that parrot can handle spaces in pathnames better now, which would be worth a bump
19:02 cotto The last time I checked was either yesterday or the day before.
19:03 not_gerd pmichaud: also https://github.com/parrot/parrot/commit/​1d6430d658d010aeaedd6f9943403827d4237f06
19:05 pmichaud these days, PARROT_REVISION tends to note the minimum revision on which Parrot will run.
19:05 pmichaud If we bump to 5.1.0, for example, then recent rakudos aren't eligible for inclusion in linux distros
19:05 pmichaud s/Parrot will run/Rakudo will run/
19:05 pmichaud (sorry, typing while distracted)
19:06 moritz pmichaud: I can merge the branch, and simply state in the ChangeLog that it'll only work with a newer parrot
19:08 pmichaud yeah.  effectively, until we know Parrot's future direction, we need to keep Rakudo/NQP tied to no later than 5.0.0 for a while, or else distro packagers won't migrate to newer rakudos.
19:08 Mike-PerlRecruiter_ joined #parrot
19:11 uvtc Oh, unlike Rakudo, when building Parrot and running `make install`, it puts it into /usr/local rather than into its very own local install directory.
19:12 not_gerd uvtc: unless you specify your own --prefix
19:12 uvtc not_gerd: gotcha. Thanks.
19:14 moritz oh, it seems I forgot to push my parrot commit anyway :(
19:14 Reini joined #parrot
19:14 moritz no, it's there
19:16 dalek nqp: 8282ac5 | moritz++ | tools/lib/NQP/Configure.pm:
19:16 dalek nqp: Small preparation to allow paths with spaces
19:16 dalek nqp: review: https://github.com/perl6/nqp/commit/8282ac5755
19:16 dalek nqp: ae3b47d | moritz++ | tools/ (2 files):
19:16 dalek nqp: more fixes to allow spacey paths
19:16 dalek nqp: review: https://github.com/perl6/nqp/commit/ae3b47dabe
19:16 dalek nqp: 1db6167 | moritz++ | tools/ (2 files):
19:16 dalek nqp: Merge remote-tracking branch 'origin/spacey'
19:16 dalek nqp:
19:16 dalek nqp: This enables NQP to build with source path and/or prefix containing
19:16 dalek nqp: spaces, given a sufficiently new parrot (RELEASE_5_0_0-8-g6ab9fdc).
19:16 dalek nqp: On older parrots, nothing should be broken. I hope. (Famous last words).
19:16 dalek nqp: review: https://github.com/perl6/nqp/commit/1db6167740
19:17 uvtc moritz: re. building sixparrot, is this correct so far? : http://www.unexpected-vortices.com/misc-n​otes/perl-6/build-install-sixparrot.html
19:17 pmichaud I don't mind if at some point we add a --gen-sixparrot option, fwiw.
19:18 moritz uvtc: well, 'make install' is missing
19:18 pmichaud I'll even do that tomorrow.
19:18 moritz uvtc: and you should configure parrot with --optimize
19:18 uvtc moritz: right. I'm at `make test` right now. :)
19:18 pmichaud that way we can have both --gen-parrot (traditional parrot) and --gen-sixparrot available
19:19 uvtc moritz: Oh, didn't know about --optimize.
19:19 uvtc moritz: am looking at http://docs.parrot.org/parrot/latest/html/d​ocs/book/pct/ch02_getting_started.pod.html .
19:29 uvtc moritz: after I `git clone git@github.com:rakudo/rakudo.git`, `git branch` tells me I'm on "nom". That's the newest, and what I want, correct?
19:31 not_gerd uvtc: correct
19:31 not_gerd parrot master, nqp master, rakudo nom
19:32 uvtc So then, the next step is `perl Configure.pl --with-parrot=/home/john/parrot/bin/parrot` , correct? Or do I need to include --gen-nqp in there?
19:33 uvtc (sorry -- that should be /home/john/opt/parrot/bin/parrot)
19:36 not_gerd uvtc: --gen-nqp won't hurt, but might be unnecessary
19:36 uvtc Trying w/o --gen-nqp...
19:37 uvtc It tells me: "Cloning into 'nqp'..."
19:37 not_gerd that's the right thiing to do if you haven't built a custom nqp yet
19:37 not_gerd also fyi, nqp and rakudo will get their prefix from parrot
19:38 uvtc not_gerd: Sorry, don't know what you mean by that.
19:39 uvtc Wait. nqp comes with Rakudo though, doesn't it?
19:39 not_gerd uvtc: nqp and rakudo will be installed in the same place as parrot is if you do a `make install`
19:40 uvtc not_gerd: I ask because, above, you wrote "parrot master, nqp master, rakudo nom", as if I'd be git cloning all three.
19:40 not_gerd uvtc: if you don't do it manually, Rakudo's Configure.pl will do it for you
19:40 uvtc I've built R* in the past, and everything installed right there into the cwd --- the rakudo directory.
19:41 not_gerd uvtc: that happens if you use `--gen-parrot` - if you don't you'll have to specify a custom prefix when configuring parrot
19:43 mtk joined #parrot
19:44 Reini joined #parrot
19:45 uvtc Wait, do I use the `--optimize` flag when configuring Parrot, Rakudo, or both?
19:45 moritz parrot
19:46 uvtc moritz: thanks.
19:47 mtk joined #parrot
19:50 mtk joined #parrot
19:51 mtk joined #parrot
20:00 uvtc When I `perl Configure.pl...` Rakudo, what does that "Cloning into 'nqp'..." indicate?
20:01 Coke wrong window?
20:02 Coke it's git cloning the nqp repository into the nqp directory - prereq for rakudo.
20:02 Coke if you had an nqp already installed, it would use that.
20:02 not_gerd uvtc: it clones https://github.com/perl6/nqp into the 'nqp' subdirectory in the rakudo....
20:02 not_gerd what Coke  said
20:02 uvtc Ah. *There* it is. I'd thought it came with Rakudo. Thanks.
20:03 uvtc Coke: no, I didn't type it into the wrong window. I figure it's on-topic here, since I'm building Rakudo to use sixparrot...
20:05 uvtc Coke: my notes are at http://www.unexpected-vortices.com/misc-n​otes/perl-6/build-install-sixparrot.html .
20:10 Coke ok. we can simplify this greatly if we move my sixparrot branch back to parrot's repo.
20:10 Coke then you can use --gen-parrot=sixparrot
20:11 Coke instead of having to clone my fork separately.
20:11 Coke Would that be helpful?
20:11 cotto Coke: +1
20:11 Coke Pretty sure moritz has a CLA on parrot.
20:12 * moritz is pretty sure too
20:15 Reini joined #parrot
20:16 Coke if I do a pull request, will the commits from branch A end up on branch A in the pre-forked project?
20:17 preflex joined #parrot
20:17 uvtc "The following step can take a long time, please be patient." ... this stage takes some serious horsepower ... shields weakening ... gravity dampeners at full power ... hull buckling ... must .. hang .. on...
20:18 cotto Coke: that sounds right
20:18 atrodo But I'm giving her all she's got cap'n
20:18 Coke I did a pull request. I am not sure I can approve my own pull request. cotto?
20:19 uvtc Spock: damage report!
20:19 uvtc We seem to have passed through unscathed captain.
20:21 * atrodo seems to finally be making headway on ejecting nci
20:21 cotto Coke: you can approve it.
20:21 cotto so much red
20:21 dalek parrot: 22b6b11 | coke++ | ports/ (37 files):
20:21 dalek parrot: Remove ports
20:21 dalek parrot:
20:21 dalek parrot: (lots of false positives when ack-ing)
20:21 dalek parrot: review: https://github.com/parrot/parrot/commit/22b6b11c7e
20:21 dalek parrot: ac9dafe | coke++ | / (11 files):
20:21 dalek parrot: Remove hllmacros and anything that uses it.
20:21 cotto or you could have if I hadn't just done so
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/ac9dafe6ad
20:22 dalek parrot: 1fab5cd | coke++ | / (7 files):
20:22 dalek parrot: remove covers, exsec, hav, vers, and fact opcodes
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/1fab5cd3be
20:22 dalek parrot: 2aa386d | coke++ | / (9 files):
20:22 dalek parrot: remove pin, unpin opcodes
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/2aa386d26f
20:22 dalek parrot: 0a85e4c | coke++ | / (12 files):
20:22 dalek parrot: Remove Complex PMC
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/0a85e4cbd1
20:22 dalek parrot: 258652d | moritz++ | / (3 files):
20:22 dalek parrot: remove addrregistry PMC
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/258652d70b
20:22 dalek parrot: 5b96c3e | moritz++ | / (3 files):
20:22 dalek parrot: Revert "remove addrregistry PMC"
20:22 dalek parrot:
20:22 dalek parrot: AddrRegistry is used in TGE
20:22 dalek parrot: This reverts commit 258652d70b17a89a5609ea0f76b02238ba0c16c1.
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/5b96c3eaaa
20:22 dalek parrot: c3bed91 | moritz++ | t/op/vivify.t:
20:22 dalek parrot: remove vivify opcode
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/c3bed91ea9
20:22 dalek parrot: 6227946 | moritz++ | sixparrot-TODO.md:
20:22 dalek parrot: add TODO file for sixparrot, mostly by Coke++
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/62279463f0
20:22 dalek parrot: 23f7d67 | moritz++ | / (2 files):
20:22 dalek parrot: more "vivify" removal
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/23f7d67904
20:22 dalek parrot: 7282466 | moritz++ | / (4 files):
20:22 dalek parrot: update bootstrap
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/728246633a
20:22 dalek parrot: 4b0d213 | coke++ | / (14 files):
20:22 dalek parrot: Remove FixedBooleanArray and ResizableBooleanArray
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/4b0d21305e
20:22 dalek parrot: 50c7090 | cotto++ | / (88 files):
20:22 dalek parrot: Merge pull request #932 from coke/sixparrot
20:22 dalek parrot:
20:22 dalek parrot: Sixparrot
20:22 dalek parrot: review: https://github.com/parrot/parrot/commit/50c709079d
20:22 atrodo 16k deletions. Not bad
20:22 Coke cotto: I was in the process of editing the commit message. whew. ;)
20:23 Coke atrodo: tip of the iceberg.
20:23 moritz wait, did that go into parrot/master ?
20:23 pmichaud that looks to me like it went into master :-(
20:23 Coke oh, so glad cotto pulled that trigger and not me. ;)
20:23 Coke Easy enough to fix.
20:24 pmichaud I think I would much prefer sixparrot to be a separate repo instead of a branch... unless there's a strong reason for the two to track.
20:24 pmichaud also, a separate repo would give an opportunity to trim the history a bit.
20:24 Coke sent a message to the list.
20:25 Coke pmichaud: the only reason I suggested putting it in parrot was to simplify the configure step for folks testing it out. Doesn't matter to me where it lives for now.
20:25 Coke cotto: want me to revert those commits?
20:25 pmichaud Coke: I suggested a little earlier that we create --gen-sixparrot
20:26 moritz maybe you want to push --force instead
20:26 moritz otherwise 'git blame' becomes very annoying, for example
20:26 not_gerd moritz: and rewrite public history? *HERETIC*
20:26 not_gerd ;)
20:26 Coke I would not feel comfortable editing history, but am ok if you want to. :)
20:26 Coke I agree it screws up things like blame.
20:26 moritz not_gerd: well yes, public for 4 minutes
20:27 uvtc Cloning parrot takes a while.
20:27 Coke anyone mind if moritz does it the hard way?
20:27 cotto So, why shouldn't sixparrot be master?
20:27 pmichaud there's also the possibility of leaving 'master' alone, and creating a branch for 'oldparrot'
20:27 Coke cotto: it's an experiment. master is too much pressure.
20:28 moritz cotto: dunno, is there a consensus? did anybody test rakudo+parrot on sixparrot?
20:28 Coke moritz: I did.
20:28 moritz Coke: ok
20:28 Coke but that's not enough. Anyway, I'm not comfortable with it being master, and it's mostly my commits atm.
20:28 atrodo my vote is to keep master where it was, not as sixparrot
20:28 pmichaud I'm okay with declaring sixparrot as 'master', if that's the consensus.
20:28 Coke moritz: you doing the force push?
20:28 moritz ok, will do the force push
20:28 cotto Coke: "pressure" is a strong word.
20:29 uvtc Yay! `~/opt/parrot/bin/perl6 -v` works. :)
20:29 cotto The 5.0 release will always be available.
20:29 moritz unless sombody objects in the next minute
20:29 cotto -1
20:29 uvtc It tells me: "This is perl6 version 2013.01-117-g8369e63 built on parrot 5.0.0 revision RELEASE_5_0_0-19-g4b0d213"
20:29 moritz ok, I'll leave it to you then
20:29 Coke It's early days, yet, anyway, and branches and forks are cheap.
20:29 pmichaud since cotto is the remaining leader, he gets to decide :)
20:29 cotto whee
20:30 Coke moritz: you're still doing the push, though.
20:30 atrodo With great titles comes great vetos
20:30 * moritz feels torn
20:31 cotto I'm not seeing the point of keeping the sixparrot work out of master.
20:31 Coke let's cleanup master for now and figure out the rest later.
20:31 Coke cotto: we have -two- people with commits at the moment.
20:31 Coke so stuff is ripped out with no benefit. for something this big we'd do it in a branch first anyway.
20:32 * not_gerd agrees with coke
20:33 not_gerd you can always pull a nom and declare sixparrot the new master
20:33 moritz ok, force-push done
20:33 Coke Ok. I'll warn the masses.
20:33 cotto Coke: ok.  So you're just suggesting that we not merge into master too quickly.
20:33 moritz Coke: thanks
20:33 Coke moritz++
20:33 Coke cotto: I could see this work ending up on master eventually, sure.
20:34 cotto Coke: gotcha.  That's sensible.
20:35 cotto The pull request wasn't you saying that the branch was ready for master but that you wanted everything in parrot/parrot.
20:38 uvtc How can I ask my fresh new ~/opt/parrot/bin/perl6 which nqp and parrot versions it's using? Is there some way to determine that?
20:38 Coke sent.
20:39 Coke cotto: yes.
20:39 pmichaud uvtc: perl6 -v
20:39 Coke but I may have been supposed to click something to make that easy when applying, I don't know. Sorry.
20:39 uvtc `perl6 -v` tells me "built on parrot 5.0.0 revision RELEASE_5_0_0-19-g4b0d213". Is that correct? I just built from sixparrot...
20:40 Coke sixparrot is based off that, so it's plausible.
20:40 pmichaud all perl6 can report is whatever revision parrot reports
20:40 Coke though admittedly unhelpful.
20:40 uvtc pmichaud: Ah, right. Thank  you.
20:40 Coke I can update that fork to say something else.
20:40 not_gerd uvtc: your notes are now obsolete - the build process just got simplified to `perl Configure.pl --gen-parrot=sixparrot && make`
20:40 cotto uvtc: that means "RELEASE_5_0_0 plus 19 commits"
20:40 moritz but rakudo doesn't report the NQP revision :(
20:40 Coke not_gerd: no, that didn't happen.
20:41 Coke not_gerd: it was pushed to master, then force wiped.
20:41 moritz cotto: but it doesn't show *which* 19 commits :-)
20:41 Coke and then we decided to leave it as is.
20:41 not_gerd Coke: then, I vote for adding that branch
20:41 moritz shall I do that, without modifying master again? :-)
20:42 pmichaud it means "19 commits since the RELEASE_5_0_0 tag", but doesn't say along what branch or fork.
20:42 Coke from the email: "And now pmichaud has volunteered to make it easy for rakudo to test against a fork, so the pull request is unneeded.
20:42 Coke "
20:42 moritz that's too much back and forth for me
20:42 pmichaud if people feel a branch is better than a separate repo, I'm okay with that.
20:43 pmichaud but it's really simple to create a --gen-sixparrot, too.
20:43 moritz it's simple either way
20:43 Coke I officially don't care, but someone else has to reply to my email if we do that. :)
20:43 perlite joined #parrot
20:43 uvtc Seems like it would be very useful for debugging porpoises if I could ask the perl6 binary exactly which branch/commit it's using for nqp and parrot.
20:43 Coke I think politically it might look better if it's in parrot/parrot.
20:44 Coke as long as people treat that branch as I would expect.
20:44 moritz uvtc: it tells you the exact parrot commit already
20:44 pmichaud uvtc: the perl6 binary only knows whatever the nqp/parrot binaries tell it.
20:44 not_gerd pmichaud: a new repo might make sense eventually (eg so you could create a git submodule for parrot in the nqp repo which doesn't pull n the whole history), but I think that should wait until we know if 'Parrot: Resurrection' actually happens
20:44 pmichaud not_gerd: +1
20:44 not_gerd and if the branch is in main parrot, no new logic is needed
20:44 not_gerd for now, that is
20:44 pmichaud and by the time perl6 is a binary, the "nqp" compiler part isn't around anymore.
20:45 atrodo uvtc: 4b0d213 is the commit
20:45 Coke If folks want to do really crazy stuff ripping things out, just make a new branch. ;)
20:45 moritz uvtc: you can go into the parrot repo and do a 'git log RELEASE_5_0_0-19-g4b0d213'
20:45 pmichaud afk, kid pickup
20:45 Reini joined #parrot
20:45 uvtc atrodo, moritz, pmichaud : Cool. Thanks.
20:46 dalek parrot/sixparrot: 22b6b11 | coke++ | ports/ (37 files):
20:46 dalek parrot/sixparrot: Remove ports
20:46 dalek parrot/sixparrot:
20:46 dalek parrot/sixparrot: (lots of false positives when ack-ing)
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/22b6b11c7e
20:46 dalek parrot/sixparrot: ac9dafe | coke++ | / (11 files):
20:46 dalek parrot/sixparrot: Remove hllmacros and anything that uses it.
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/ac9dafe6ad
20:46 dalek parrot/sixparrot: 1fab5cd | coke++ | / (7 files):
20:46 dalek parrot/sixparrot: remove covers, exsec, hav, vers, and fact opcodes
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/1fab5cd3be
20:46 dalek parrot/sixparrot: 2aa386d | coke++ | / (9 files):
20:46 dalek parrot/sixparrot: remove pin, unpin opcodes
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/2aa386d26f
20:46 dalek parrot/sixparrot: 0a85e4c | coke++ | / (12 files):
20:46 dalek parrot/sixparrot: Remove Complex PMC
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/0a85e4cbd1
20:46 dalek parrot/sixparrot: 258652d | moritz++ | / (3 files):
20:46 dalek parrot/sixparrot: remove addrregistry PMC
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/258652d70b
20:46 dalek parrot/sixparrot: 5b96c3e | moritz++ | / (3 files):
20:46 dalek parrot/sixparrot: Revert "remove addrregistry PMC"
20:46 dalek parrot/sixparrot:
20:46 dalek parrot/sixparrot: AddrRegistry is used in TGE
20:46 dalek parrot/sixparrot: This reverts commit 258652d70b17a89a5609ea0f76b02238ba0c16c1.
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/5b96c3eaaa
20:46 dalek parrot/sixparrot: c3bed91 | moritz++ | t/op/vivify.t:
20:46 dalek parrot/sixparrot: remove vivify opcode
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/c3bed91ea9
20:46 dalek parrot/sixparrot: 6227946 | moritz++ | sixparrot-TODO.md:
20:46 dalek parrot/sixparrot: add TODO file for sixparrot, mostly by Coke++
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/62279463f0
20:46 dalek parrot/sixparrot: 23f7d67 | moritz++ | / (2 files):
20:46 dalek parrot/sixparrot: more "vivify" removal
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/23f7d67904
20:46 dalek parrot/sixparrot: 7282466 | moritz++ | / (4 files):
20:46 dalek parrot/sixparrot: update bootstrap
20:46 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/728246633a
20:47 dalek parrot/sixparrot: 4b0d213 | coke++ | / (14 files):
20:47 dalek parrot/sixparrot: Remove FixedBooleanArray and ResizableBooleanArray
20:47 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/4b0d21305e
20:48 Coke ok. who is going to reply to the email with the new status? ;)
20:48 moritz I'll do it
20:50 Coke moritz++
20:51 Coke ok, NOW your instructions can be simpler.
20:52 Coke killing the fork.
20:52 moritz WITH FIRE
20:53 uvtc Ok. Will have another go. What part of http://www.unexpected-vortices.com/misc-n​otes/perl-6/build-install-sixparrot.html needs to change?
20:53 not_gerd uvtc: you no longer need to build your own parrot
20:54 not_gerd just check out rakudo and do a `perl Configure.pl --gen-parrot=sixparrot`
20:54 Coke check out a rakudo: perl Configure.pl --gen-nqp --gen-parrot=sixparrot
20:54 uvtc Oh, very nice. Will go try that presently.
20:56 uvtc Er, though, I should include the `--gen-parrot-option=--optimize` as well, correct?
20:57 uvtc This is, btw, on Xubuntu 12.04, which comes with gcc 4.6.3.
20:57 cotto I thought that nqp did that by default
20:59 uvtc I'm looking at http://rakudo.org/how-to-get-rakudo/ .
21:00 uvtc moritz: is `--optimize` only needed when using those specific versions of GCC (4.1.2 & 4.1.3)?
21:00 moritz uvtc: no
21:00 moritz well
21:00 moritz it's not *needed* otherwise
21:01 moritz but if you want a fast rakudo, you'd better include it
21:01 moritz and yes, rakudo uses --optimize for parrot if no --gen-parrot-option was specified
21:01 moritz tools/lib/NQP/Configure.pm:    my @opts       = @{ $options{'parrot-option'} || ["--optimize"] };
21:03 cotto That's what I thought.
21:10 uvtc Note, when I Configure.pl Rakudo, and it grabs nqp, it tells me:
21:10 uvtc Cloning into 'nqp'...
21:10 uvtc {snip}
21:10 uvtc Note: checking out '2013.01-160-gc41c8f4'.
21:10 uvtc You are in 'detached HEAD' state.
21:11 * atrodo enjoys a good soap opera
21:11 Coke yes, because the required version isn't HEAD
21:11 Coke if you want to play with nqp master ( a good idea in this case), use --gen-nqp=master
21:12 atrodo ignore me. ENOTMYTTY
21:12 uvtc Oh. Well, for folks who want to try out the latest changes of Rakudo, nqp, and sixparrot, all together, they'd want --gen-nqp=master, right?
21:13 uvtc I mean, in order to make useful and up-to-date bug reports here.
21:14 uvtc Since now, I'm guessing that changes to nqp and sixparrot are probably going to correspond pretty closely, no?
21:15 Reini joined #parrot
21:17 uvtc Coke: which is what you said. :)
21:18 Coke Aye.
21:19 moritz uvtc: nqp/master won't start to respond to sixparrot immediately, I think
21:19 Coke nothing to respond to at the moment.
21:23 Coke utvc: --parrot-make-option='--jobs=4' is also nice.
21:23 dalek rakudo/nom: 1e85ffd | moritz++ | Configure.pl:
21:23 dalek rakudo/nom: [Configure] stop on unknown command line options, [Coke]++
21:23 dalek rakudo/nom: review: https://github.com/rakudo/rakudo/commit/1e85ffd2cb
21:24 uvtc Coke: That's an additional option to the Rakudo Configure step?
21:24 uvtc Coke: what does it do?
21:24 Coke yes, tells parrot to build with "make --jobs=4" instead of the default 1.
21:25 uvtc Ah. Thanks.
21:25 Coke (set it to a reasonable number for your machine, but it can speed up the build process)
21:35 uvtc Hm. That optimization stage didn't pin my machine like last time.
21:35 uvtc Wait. Come to think of it, that was probably firefox.
21:42 TonyC joined #parrot
21:44 particle joined #parrot
21:46 Reini joined #parrot
21:47 uvtc Thanks for the help, all. It builds, and `perl6 -v` reports:
21:47 uvtc This is perl6 version 2013.01-117-g8369e63 built on parrot 5.0.0 revision RELEASE_5_0_0-19-g4b0d213
21:47 uvtc I updated my build/install notes, but am having a problem with my webhost at the moment.
21:54 uvtc Updated http://www.unexpected-vortices.com/misc-n​otes/perl-6/build-install-sixparrot.html .
21:56 uvtc So, since the output of `perl6 -v` includes "on parrot 5.0.0 revision RELEASE_5_0_0-19-g4b0d213",  I went looking for the "g4b0d213" commit at https://github.com/parrot/parrot/commits/sixparrot , but don't see it there. Am I looking in the wrong place?
21:56 atrodo take off the g
21:56 uvtc hahaha. :)
21:56 uvtc Thanks, atrodo .
21:56 atrodo uvtc: no problem
21:57 Timbus joined #parrot
22:01 not_gerd bye, #parrot
22:01 not_gerd left #parrot
22:16 Reini joined #parrot
22:16 uvtc Man, slavorg, you are *on the ball*.
22:17 cotto it's a bot
22:17 uvtc I mean, you can't slip one past that guy.
22:17 uvtc ;)
22:32 uvtc left #parrot
22:37 Reini joined #parrot
22:40 mtk joined #parrot
23:06 dalek parrot/sixparrot: 1f1d086 | rurban++ | src/pmc/bigint.pmc:
23:06 dalek parrot/sixparrot: [codingstd] fix bigint c_arg_assert.t test
23:06 dalek parrot/sixparrot: review: https://github.com/parrot/parrot/commit/1f1d08638a
23:12 benabik joined #parrot
23:13 benabik_ joined #parrot
23:39 davidfetter joined #parrot
23:43 sivoais joined #parrot

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

Parrot | source cross referenced