Parrot 2.9.0 will be released at '2010-10-19 08:00 UTC' | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close configure component tickets
Set by moderator on 18 October 2010.
00:02 ruoso joined 00:36 davidfetter left 00:39 dmalcolm left 00:41 bacek left 00:47 dngor_ joined 00:48 dngor left 01:08 ruoso left 01:26 ruoso joined 01:32 kid51 joined 01:46 treed left 01:55 bluescreen joined 02:07 dngor joined 02:08 dngor_ left 02:17 whiteknight left 02:24 kid51 left 02:35 janus left, janus joined
cotto purely theoretical question: If I were to start a blog, what'd be a awesome name? 02:39
atrodo Insert Awesome Blog Name Here? 02:40
dukeleto cotto: very good question. use "strangely consistent" as inspiration 02:43
cotto dukeleto, can you elaborate?
Ah. That seems to be masak's blog. 02:44
My first thought is "Year of the Newly Exploding Parrot", but the idea probably needs more thought. 02:46
dukeleto cotto: you can go on a wiki binge about parrots and then maybe a solution to your problem will present itself 02:47
cotto: or seek trac.parrot.org/parrot/wiki/ParrotQuotes for enlightenment ;) 02:48
cotto an intentional wikipedia random walk? yes, thanks!
ironically, I read through that earlier today.
I love that page.
dukeleto cotto: it has a very large laugh density 02:49
cotto Sadly, much of the humor comes from bashing on Parrot internals. That portion will decrease as we clean the bird up. 02:50
I guess it's a win-win of sorts.
dukeleto indeed it is. 02:56
cotto: what is the purpose of your new blog? anything in particular?
cotto: why parrot will take over the world? 02:57
cotto It'll be mostly parrot-related. 02:58
I'd rather write about Parrot taking over the world after the fact. Too many people do that prematurely and end up looking pretty silly. 03:04
;)
bacek_at_work Hey, we didn't have updates on ParrotQuotes for 3 months. Is parrot development not funny anymore??? 03:38
03:40 davidfetter joined
cotto You can't force quality humor. 03:51
davidfetter you can force *bad* quality humor :) 03:55
03:56 brianwisti joined 04:00 dngor left
dukeleto cotto: something with the words "feather" or "flock" ? 04:03
cotto some sort of bird vocabulary sounds like a good idea 04:05
cotto likes bikesh^H^H^H^H^H^Hthinking of names for things
brianwisti Naming things can be fun. 04:08
cotto says Mr. Coolnamehere ;)
brianwisti The names I wanted were taken: aardvark.org and artwhore.com
alas. 04:09
cotto I was really lucky to get pipp.org for the eventual PHP on Parrot implementation.
brianwisti I was surprised I got coolnamehere. Although I do occasionally wish I'd gone for something shorter. 04:10
I hope you're pondering something *like* PHP, rather than a full-on bug-for-design-flaw implementation. 04:11
cotto My eventual goal is a compatible but better implementation. I completely agree that the language is a hopeless mess, but it's also hugely popular. 04:13
It's a great example of worse is better.
brianwisti I like a lot of the ideas in PHP. It was my first Web app language. Sorta-compatible-but-better would be awesome. 04:14
cotto It does deployment really well.
It's also brain-dead easy to get started in. 04:15
It might be an attractive feature that anyone can feel like they're contributing to the docs, though that probably does more harm than good. 04:16
plobsing requiring people to download the source to create a patch to submit increases the chances said people will actually *read* the source 04:18
cotto I was referring to the comments they allow on their docs. 04:19
plobsing yes. I've had the displeasure of trying to use those to cobble together a fix before. 04:20
cotto bad idea 04:21
any idiot can post and most do
plobsing exactly. which is why I favour a patch barrier to docs.
cotto too 04:22
davidfetter 3, fwiw 04:32
04:45 dngor joined 05:16 Kulag left 05:19 Kulag joined 05:21 jsut joined 05:26 Kulag left, jsut_ left 05:29 Kulag joined 05:37 Kulag left 05:42 Kulag joined 05:44 Maddingue left, Drossel joined, AzureSto_ left 05:45 AzureStone joined, simcop2387_ joined 05:46 Maddingu1 joined, mj41_ joined, jjore left 05:47 simcop2387 left, simcop2387_ is now known as simcop2387 05:48 cotto left, cotto joined, Kulag left 05:49 jjore joined 05:50 mj41 left, mj41_ is now known as mj41 06:05 brianwisti left, bacek joined 06:09 theory left 06:29 he__ joined 06:32 dngor_ joined, dngor left 07:16 Drossel left, Kulag joined 07:38 gerd joined
gerd I just run the tests for Parrot 2.9.0 with revision 49573 07:43
make fulltest was successful, can I do the "commit" for 2.9.0 now? 07:44
cotto I'd like to give NEWS a once-over. 07:45
gerd Please do I it 07:46
I will do a new checkout 07:47
cotto bacek, ping 07:48
aloha, clock?
bacek cotto, pong
aloha cotto: cotto: LAX: Tue, 00:48 PDT / CHI: Tue, 02:48 CDT / NYC: Tue, 03:48 EDT / UTC: Tue, 07:48 UTC / LON: Tue, 08:48 BST / BER: Tue, 09:48 CEST / TOK: Tue, 16:48 JST / SYD: Tue, 18:48 EST
cotto: I give up.
cotto bacek, what does the new gc in trunk do that the old one didn't? 07:49
bacek cotto, nothing.
It's just foundation for GenGC
cotto just some refactors and cleanups?
Hmm. I'm not especially happy with the NEWS atm. It may just be that not much has happened, but it's not especially impressive atm. 07:53
gerd So I will do the commit now, I think the NEWS are long enough 07:54
cotto Do you mind if I take a few minutes to troll through old #ps logs? 07:55
07:55 Maddingu1 is now known as Maddingue
cotto I'm not looking for length as much as items people who aren't core developers will care about and understand. 07:55
gerd Okay! Let me kown when it is done 07:56
cotto sure 07:57
08:07 contingencyplan left
cotto gerd, done 08:11
er, not quite
gerd the revision is still 49573 08:12
I am still waiting, do want to change the file NEWS 08:15
cotto now done
Our convention of using + and - in NEWS makes the diffs somewhat entertaining 08:16
gerd Okay, I think can do my "commit" without running the test again
cotto I had some conflicts from being a slacker and not updating late enough. 08:17
seems safe ;]
by "commit", do you mean tag the release?
gerd yes
cotto Stupid Configure.pl trick: pmclass Foo extends Foo {} in a PMC will cause Configure.pl to spin in an infinite loop and eat all available memory slowly. 08:18
dalek rrot: r49574 | cotto++ | trunk/NEWS:
[NEWS] various tweaks to try to make the news more interesting to J. Random Hacker
 08:20
cotto Hm. How does this work for a rule of thumb for interesting NEWS items: look for things that are easy to appreciate or that will make users' lives easier 08:22
bacek Is supported release NEWS cumulative since last supported release? 08:25
08:26 dngor_ left, dngor joined
cotto bacek, I don't think we decided on that either way. 08:26
bacek ok
gerd I there a problem that needed to be fixed, before 2.9.0 can released 08:27
cotto I'd like to make the Lorito transition first, but that's probably not realistic without at least 4 extra baceks.
08:27 lucian left
gerd So could I do the commit to 2.9.0 now? 08:30
cotto go for it 08:31
gerd done the revision is 49575 08:32
ls 08:34
dalek rrot: r49575 | gerd++ | trunk (13 files):
update to 2.9.0
 08:35
cotto gerd++ 08:40
I love having the release out early. 08:41
moritz gerd++
gerd I tagged the release (49576) 08:42
cotto so you did 08:43
08:48 bacek left
dalek rrot: r49576 | gerd++ | tags/RELEASE_2_9_0:
tagged release 2.9.0
 08:50
gerd 2.9.0 is copied to the ftp-server and ftp.parrot.org is triggered 08:53
08:54 fperrad joined
gerd "make fulltest" for 2.9.0 tar file just finished successful 08:54
cotto Do you have permissions to update docs.parrot.org? 08:55
gerd No, I do not have.
Can I get the permissions? I would like to update the documentaion. 08:57
08:57 bacek joined
cotto I'll do it now. I think Coke can get you a bit. 08:58
moderator Parrot 2.9.0 Released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close configure component tickets 08:58
gerd Do you need my ssh-key? 09:01
cotto for parrotvm osuosl needs to set you up with your own account 09:04
apparently
gerd Would you create a account for me and e-mail the username and password to me. 09:08
cotto I don't have permissions to do that. 09:09
you can hop on #osuosl on freenode
Coke and dukeleto would know who to talk to 09:10
gerd Okay I will ask Coke for it or only ask him for the documentation update 09:11
I will leave and the write the HTML-page, have a nice day! 09:12
Bye! 09:14
09:14 gerd left
cotto night 09:17
msg gerd I updated docs.parrot.org for 2.9.0 09:18
aloha OK. I'll deliver the message.
cotto msg ijwqeofqlaslkdf test
aloha OK. I'll deliver the message.
bacek cotto, Msg plugin doesn't check existence of user 09:27
dalek rrot: r49577 | bacek++ | branches/generational_gc/include/parrot/gc_api.h:
Remove shortcut in mark_PMC_alive.

This shortcut does break it.		 09:37
rrot: r49578 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Rework pmc_propagate function.
rrot: r49579 | bacek++ | branches/generational_gc/include/parrot/gc_api.h:
Remove shortcut in mark_STRING_alive
rrot: r49580 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
DRY: use already assigned gen
rrot: r49581 | bacek++ | branches/generational_gc/src/pmc.c:
HACK: preserve GC invariant in pmc_reuse
bacek ookey 09:45
gen_gc passing most of test suite now 09:46
10:15 fperrad left, fperrad joined 10:20 zostay left, zostay joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#693) fulltest) at r49581 - Ubuntu 10.10 amd64 (g++-4.5 with --optimize) 10:25
bacek: same read_only failures we had before - make test has a bunch of failures AFAICS all - src/vtables.c:65: failed assertion 'base_vtable' 10:33
got to go out for a bit - bbl 10:34
10:44 dngor_ joined 10:47 dngor left
dalek website: gerd++ | Parrot 2.9.0 "Red-masked" Released! 11:12
website: www.parrot.org/news/2010/Parrot-2.9.0
11:12 ruoso left 11:40 ilbot2 joined
moderator Parrot 2.9.0 Released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close configure component tickets
dalek rrot: r49582 | bacek++ | branches/generational_gc/src/pmc/class.pmc:
Temporary workarond for mussing ro_variant_vtable
 12:13
12:13 dngor_ is now known as dngor
bacek YES! 12:23
bacek going for recharge
12:31 he__ left 12:48 bluescreen left 12:59 bluescreen joined 13:32 dngor_ joined 13:33 contingencyplan joined, dngor left 13:47 davidfetter left 13:56 plobsing left
nopaste "mikehh" at 192.168.1.3 pasted "generational_gc branch - make test" (40 lines) at nopaste.snit.ch/24597 14:06
mikehh msg bacek: make test results nopaste.snit.ch/24597 14:08
aloha OK. I'll deliver the message.
14:13 dngor_ is now known as dngor, szabgab joined 14:34 NotFound joined
NotFound hi 14:34
NotFound is back
14:38 dngor left 14:48 lucian joined, brianwisti joined
mikehh NotFound: where you bin 15:04
NotFound: ain't seen you for a spell 15:05
NotFound Physical world is demanding sometimes
mikehh :-} 15:06
NotFound Nothing importante, but being busy with a mix of things.
mikehh Yeah, know that problem well 15:07
15:08 davidfetter joined 15:19 lucian_ joined 15:22 lucian left 15:41 mikehh left 15:43 davidfetter left 15:48 dngor joined, davidfetter joined 16:01 theory joined 16:02 ruoso joined
pmichaud I'll send email to parrot-dev shortly... but some change in the last four days is causing Rakudo builds to take an incredibly long time 16:04
gist.github.com/634456 # table of revisions and build times
I'm running again and will let the rakudo build on 2.9.0 go to completion so we know how long it's actually taking. 16:05
16:11 PacoLinux left, PacoLinux joined 16:20 dmalcolm joined
pmichaud lists.parrot.org/pipermail/parrot-d...04929.html # urgent message to parrot-dev 16:24
dukeleto will be missing #ps today due to being on a plane to the gsoc mentor summit
pmichaud I'm requesting a hard freeze on trunk commits and (especially) branch merges until this is resolved; otherwise things will get very dicey for Rakudo's build system. 16:25
moritz and maybe a 2.9.1 point release where this is fixed? 16:26
pmichaud I'm sure there will need to be a 2.9.1 point release. 16:27
davidfetter dukeleto, no wifi on airplane?!? 16:33
17:06 dngor_ joined 17:07 dngor left 17:08 dngor_ is now known as dngor 17:19 Andy joined
pmichaud gist.github.com/634604 # build timings 17:22
afk, lunch
17:22 dngor left 17:31 M_o_C joined 17:53 M_o_C left
Util I have verified that change 49557 *alone* is responsible for the abnormal build time. 17:58
If change 49557 is reverted from the 2.9.0 during the rakudo build, the build happens in normal time.
i.e. `cd parrot && patch -R -p2 < ~/Downloads/changeset_r49557.diff` during
`perl Configure.pl --gen-parrot`, after the SVN checkout has occurred, but before gc_ms2.c compiles.
Reverted in r49583. 18:02
18:03 dngor joined 18:06 mikehh joined
dalek rrot: r49583 | Util++ | trunk/src/gc/gc_ms2.c:
Reverting r49557 to resolve Rakudo abnormal build times
 18:07
Util Also, pushed r49557 into PARROT_REVISION on Rakudo's GitHub master 18:09
I mean r49583
18:29 zostay left 18:30 zostay joined 18:52 lucian_ left 19:00 lucian joined
dukeleto says hello from the gates of the PDX airport 19:08
pmichaud wonders if the gates are "perly gates" or if they say "woe to those who enter here"
jnthn ...the gates talk? 19:09
pmichaud I hear them all the time. annoying things.
"Final call for Southwest flight 1234 and if you don't board now we'll give away your seat and foreclose on your house." 19:10
19:11 estrabd left
dukeleto Evidently my flight is getting "serviced" and the boarding has been pushed back to 12:30pm, so I may be at the first few mins of #ps 19:17
cotto is #ps in 10 minutes or 70? 19:20
pmichaud cotto: /topic says it's in 68 19:22
(20:30 utc)
cotto thanks. my irc client likes to mangle topics 19:23
dukeleto i thought it was at 12:30 PST, but it is actually at 1:30pm. I can't seem to remember that. 19:24
this commit trac.parrot.org/parrot/changeset/49583/ reverts the commit that made parrot build on his laptop 19:32
pmichaud dukeleto: that commit also makes Rakudo builds 6x slower 19:34
cotto dukeleto, there was quite a mess over that commit. It increases Rakudo build times something like 4-5x
pmichaud sorry
without that commit, Rakudo builds are 6x slower
cotto perhaps bacek's gc will help once it's ready
pmichaud we got from 4m15s to well over 26m on my box
19:35 estrabd joined
pmichaud afk, kid pickup 19:35
dukeleto pmichaud: understandable. We need to fix it in a better way. 19:36
19:52 tcurtis joined
cotto hi tcurtis 19:52
19:52 luben left
tcurtis Hi, cotto. 19:52
20:00 patspam joined
davidfetter dukeleto, what time does your flight land? 20:02
mikehh got a seg fault in t/pmc/threads.t - test 9 in fulltest/testf, reran the test and it was ok, ran it a bunch of times and it failed once out of 10 runs (the first re-run) 20:07
ran it as perl t/harness --gc-debug -f --runcore-tests (first time) and make testf (bunch of times), re-ran perl t/harness --gc-debug -f --runcore-tests (also a bunch of times) 20:11
so I got a failure in make fulltest and then a second, but it keeps passing now 20:12
20:12 kid51 joined
kid51 notes that the reversion in r49583, while it may enable Rakudo to build once again, is also likely to pose the same problem on small-physical-memory machines as described in TT #1829 in the first place. 20:15
PerlJam kid51: aye 20:16
cotto it's the lesser of two evils for now until we can find a good fix for both
PerlJam Could that number be a macro or something that adjusts to the local system?
kid51 Well, chromatic commented in his commit that it ought to be a command-line option to Configure.pl 20:20
I would go farther.
cotto something dynamic and automatic?
PerlJam kid51: please do :)
kid51 I think it should be a machine-specific value determined by some combination of a 'hints' file and an 'auto' configure step.
chromatic worked on that at the Portland gathering on Saturday. 20:21
cotto does that assume that everyone will build Parrot?
atrodo or that the machine that it's built on will never change memory size?
kid51 I was actually somewhat surprised that he committed my hack (intended to get Parrot to build on my iBook) to trunk ...
... but I assumed that he felt it would have no negative repercussions. 20:22
This needs to be dealt with by people who understand the new GC subsystem ... 20:23
... and who can get it to work both in high-memory footprint situations (like, currently, building Rakudo) and low-resource situations. 20:24
20:29 chromatic joined
mikehh kid51: one of the problems is that the whole gc subsystem is in flux at the moment 20:30
#ps time
20:33 ruoso left 20:34 patspam left 20:35 theory left 20:47 bluescreen left
nopaste "bacek" at 192.168.1.3 pasted "Phys mem calculation is clisp" (11 lines) at nopaste.snit.ch/24608 20:48
"chromatic" at 192.168.1.3 pasted "bacek: hairball of GC tuning; big improvements" (434 lines) at nopaste.snit.ch/24609 20:56
20:58 kid51 left, bluescreen joined 21:01 perlite left, perlite joined 21:14 theory_ joined
chromatic bacek, is there any concrete value to having a pluggable GC now? I'm removing a lot of the plugging for good performance improvements. 21:25
moderator Parrot 2.9.0 Released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GC tuning | remove deprecations | 21:25
21:26 plobsing joined
chromatic bacek, one thing I noticed in gc_ms2: traversing the old generation list to reset colors seems far too expensive. 21:29
It looked very compatible with the sweepless GC scheme.
bacek chromatic, in gen_gc?
chromatic I haven't looked at gen_gc yet. 21:30
bacek chromatic, there is no "old generation" is gc_ms2...
chromatic there's ->objects and ->new_objects.
bacek (pluggable GC) I think we switch to compile time selection 21:31
new_objects == root_objects
chromatic Right.
All of this redirection through function pointers is expensive.
bacek Yes, I know.
chromatic If we had a decent C compiler which could optimize whole programs... ah. 21:33
bacek gcc 4.5? 21:37
icc -ipo?
anyway
chromatic I couldn't get gcc 4.5 to do anything measurable when I tried.
bacek afk # $dayjob preparation
21:41 bluescreen left 21:45 fperrad left 21:46 lucian left 21:49 brianwisti left
allison I'll do a longer post, but for the quick reassurance: me handing off the Chief Architect role is nothing like Dan. I love Parrot, think it's on the upswing and on a good track to take over the world of VMs. 21:56
The meeting on Saturday (and Friday before with Jim), was like a cloud lifting, I've been so worried the past year on how to keep Parrot moving forward. 21:57
21:58 theory_ is now known as theory 22:00 brianwisti joined
dalek rrot: r49584 | cotto++ | trunk (8 files):
update Parrot to 2.9.1
 22:16
22:16 jsut left
22:16 jsut joined
cotto Please hold off committing anything to trunk for the next few minutes while I get this bugfix release out. 22:17
dukeleto bugfix release? 22:18
dukeleto just got to his hotel
cotto yes, to fix the build time issue that Rakudo was seeing 22:19
dukeleto cotto: ah, i saw the commit, but i didn't know it wasn't in 2.9.0 22:20
cotto Yup. 2.9.1 = 2.9.0 + that commit 22:21
dukeleto cotto: this is something that timing automated builds could have detected
cotto How easy is that to do? That information could be very useful.
dukeleto cotto: pretty darn easy. It is something the compile farm could do. we just need the infrastructure to track it usefully 22:23
cotto what infrastructure would we need? 22:26
dalek rrot: r49585 | cotto++ | tags/RELEASE_2_9_1:
tag 2.9.1 release
 22:31
rrot: r49586 | cotto++ | tags/RELEASE_2_9_1:
delete mistagged release
rrot: r49587 | cotto++ | tags/RELEASE_2_9_1:
tag 2.9.1 release (for real this time)
cotto Should I send out an announcement for 2.9.1? 22:37
22:37 brianwisti left
chromatic Please do. 22:39
At least email and website.
cotto ok 22:41
22:47 theory left 22:48 silug left 22:49 theory joined
dukeleto cotto++ 22:51
cotto done 22:54
moderator Parrot 2.9.1 Released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GC tuning | remove deprecations | 22:54
dalek website: cotto++ | Parrot 2.9.1 Released 22:54
website: www.parrot.org/news/parrot-2.9.1-released
dukeleto no codenames for bugfix releases, eh? 22:58
dukeleto likes codenames 22:59
cotto figuring out a codename isn't conducive to getting a release out the door quickly 23:00
I do like them, though
either I've hit my afternoon slump or the release process is pretty draining 23:01
it's done now, though 23:02
recommence breaking stuff!
dukeleto cotto: thanks for doing that. We all very much appreciate it. A new release *does* drain some energy 23:03
cotto: that itemized list of stuff to do is *long*, but well-specified 23:04
cotto ye 23:09
s
If you want to release 2.9.2 with a codename, I won't stop you.
davidfetter "brown paper bag" ? 23:11
cotto you release it, you name it 23:12
dalek rrot: r49588 | cotto++ | trunk/tools/release/templates.json:
[release] clean up crow template grammar a bit
 23:17
23:48 whiteknight joined 23:52 davidfetter left 23:54 Andy left

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!