Parrot 2.9.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Long live Git! github.com/parrot/parrot | git clone git://github.com/parrot/parrot.git
Set by moderator on 10 November 2010.
00:06 bacek_at_work left, bacek_at_work joined 00:15 dngor_ joined
dalek nxed: r674 | NotFound++ | trunk/winxedst1.winxed:
simplify null checkers in stage 1
 00:16
00:16 dngor left 00:32 theory left 00:38 theory joined
cotto_work new proposal: implement Lorito in Minecraft - www.minecraftforum.net/viewtopic.ph...mp;t=69299 00:41
seanstickle Worst steganography ever 00:49
01:00 theory left 01:01 theory joined 01:03 seanstickle left, theory left, theory joined
atrodo cotto_work> +1 from me. I'll get right on it 01:08
cotto_work If you thought Rakudo was slow before, just you wait. 01:12
01:13 theory_ joined, theory left, theory_ is now known as theory, seanstickle joined, seanstickle left 01:15 theory_ joined 01:17 theory left, theory_ is now known as theory 01:41 dngor_ is now known as dngor 01:55 dmalcolm left
cotto `` 01:57
cotto is shiftless 02:04
02:10 theory left
GeJ pushes @cotto, 'stuff'; 02:11
02:14 dngor left 02:16 whiteknight left
chromatic I probably won't have time to merge the GC MS2 SF branch before next Tuesday. Volunteers welcome. 02:17
02:18 dngor joined
GeJ chromatic: out of curiosity, what's the branch about? Is it related to bacek's generational_gc branch? 02:25
02:26 bluescreen left 02:34 rurban_ joined 02:35 chromatic left 02:36 rurban left, rurban_ is now known as rurban 03:06 chromatic joined
chromatic GeJ, it's merely tunings to trunk's existing GC. 03:08
03:11 theory joined
bacek_at_work chromatic, speaking of which. Can you take a look at github.com/parrot/parrot/commit/a6...e303411b5f please? 03:19
chromatic Looking. 03:20
Thinking about it for a while. 03:22
bacek_at_work chromatic, ok 03:23
03:25 lidden left 03:33 theory left, theory joined 03:55 plobsing joined
chromatic bacek_at_work, it'd be nice to get rid of sweeping. 04:32
04:33 theory left, theory joined
dalek rrot: b63472c | plobsing++ | config/gen/config_pm.pm:
add pkg-config equivalents to parrot_config
 04:34
rrot: 748cb9d | plobsing++ | docs/embed.pod:
change docs to use parrot_config

pkg-config has problems (TT #1842). autoconf isn't used by any embedders ATM, so is purely speculative.
rrot: b63472c | plobsing++ | config/gen/config_pm.pm:
add pkg-config equivalents to parrot_config
 04:37
rrot: 748cb9d | plobsing++ | docs/embed.pod:
change docs to use parrot_config

pkg-config has problems (TT #1842). autoconf isn't used by any embedders ATM, so is purely speculative.
plobsing OMG double commit report. what does it mean? 04:39
04:39 theory left
simon_ haha 04:43
04:43 Kulag left 04:44 Kulag joined, theory joined
GeJ plobsing: double karma? 04:44
sorear plobsing: it means somebody forgot to remove parrot from the poll list after setting up push notifications
cotto GeJ, yup. If you're lucky, you too can trick dalek.
sorear I don't know who to blame though
bacek_at_work chromatic, it is. And recursive marking too. But I'm not able to figure out exact way how to do it. 04:45
GeJ purl, always blame purl.
cotto even when absent, it's her fault
GeJ exactly. 04:46
Pardon my ignorance, but how would work a M&S GC without the sweep part? (I don't know zilch about GC in general) 04:49
bacek_at_work GeJ, trac.parrot.org/parrot/wiki/GCSweep...ementation 04:51
Basically - we gather all live objects into separate list. We still have "sweep" but iterate only actually dead objects, not whole lot. 04:52
05:04 theory left 05:12 theory joined 05:14 Kulag left
chromatic I still think there's a way to combine generational GC with sweep-free. 05:16
Every freshly allocated GCable goes into a (skip list?) called the newest generation.
Any time we do a full GC to promote from that generation, we know we can immediately collect everything in that newest generation. 05:17
05:17 Kulag joined 05:22 preflex left 05:26 preflex joined 05:34 theory left, theory joined 06:30 theory left 06:41 simcop2387 left, simcop2387 joined 07:16 chromatic left 08:27 fperrad joined
dukeleto 'ello 08:52
moritz oh hai 09:30
nopaste "moritz" at 192.168.1.3 pasted "compare git-describe output" (28 lines) at nopaste.snit.ch/25560 09:31
moritz I've written the code in the nopaste for rakudo's configure system, so that we can compare git-describe output with a reference value in the same format 09:32
09:42 fperrad_ joined 09:46 fperrad left, fperrad_ is now known as fperrad 10:30 masonkramer left 10:33 rurban_ joined 10:36 rurban left, rurban_ is now known as rurban 10:40 masonkramer joined 12:24 lucian joined 12:32 mikehh left 12:37 bluescreen joined
dalek rrot/tt855_debugging_options: f0e1b5c | jkeenan++ | / (5 files):
Eliminate sub _is_debug_setting_valid() in config/inter/progs.pm. Adjust 3
 12:44
rrot/tt855_debugging_options: b1959be | jkeenan++ | / (4 files):
Move processing of 'ccwarn' to location of other options. Rename sub
rrot/tt855_debugging_options: 019dac0 | jkeenan++ | / (3 files):
Eliminate profiling option.
12:46 bacek left 13:34 whiteknight joined
whiteknight good morning, #parrot 13:35
13:51 s1n joined 14:25 chromatic joined 14:26 bacek joined
dukeleto moritz++ 14:29
moritz: the question of the hour: Where should that code live? 14:30
moritz dukeleto: that's not the question of the hour
dukeleto: the question of the hour is "when will git-describe output be available, so that I can actually use it?" 14:31
my original plan was to put it in a module in rakudo 14:32
but I'm open to other options too
dukeleto moritz: i am at a conference and travelling, so I have not been as quick as I would like. But i have a branch for git_describe. It is close 14:34
moritz: I promise I will give you something to use soon 14:35
moritz oh, I missed the branch
moritz takes a look
dukeleto moritz: github.com/parrot/parrot/tree/git_describe
moritz: it doesn't yet put the key in parrot_config, that code adds the config step to detect it 14:36
moritz: so it is half done
moritz: and i wrote a few tests
moritz is it avaialable as config key already? 14:38
dukeleto moritz: i don't think so. But it is a matter of writing a few lines. I just don't know where those lines live. 14:41
moritz :-) 14:42
PerlJam Hmm. 14:45
dukeleto How do i add something to parrot_config 14:46
PerlJam I think the regex on line 35 of config/auto/git_describe.pm is incorrect. It uses dots between the numbers and the tags have _ between the numbers.
rurban Note that the config hash is already too big
moritz rurban: adding a few bytes won't kill it 14:47
dukeleto rurban: No trolling right now. We are trying to make something actually work.
PerlJam dukeleto++
dukeleto PerlJam: the . is not as strict. it will match a _
rurban all the big instal strings can be stripped, really
moritz rurban: I think it's more a question if we actully need to store the list of config steps, for example
PerlJam dukeleto: no, it's literal dots: /^(RELEASE_|REL_)\\d+\\.\\d+\\.\\d+~g[a-z0-9]+$/i
dukeleto rurban: feel free to make a branch to remove the cruft from parrot_config, that would be awesome
rurban the version string of course is fine and small enough
dukeleto PerlJam: touche! Thanks. 14:48
rurban already did but rejected ages ago
dukeleto rurban: that will really help on small memory systems
moritz dukeleto: mayb use [_\\W] instead
dukeleto rurban: we are in the age of git, try again :)
rurban ok, when I have my p5 jit working finally
dalek rrot/git_describe: 79cf2c7 | dukeleto++ | config/auto/git_describe.pm:
[config] Fix the regex that matches a git-describe string, PerlJam++
 14:50
14:50 C0ke joined
dukeleto Parrot::Config::Generated ! 14:50
dalek rrot: 8123d2d | moritz++ | lib/Parrot/Configure/Compiler.pm:
fix typo in lib/Parrot/Configure/Compiler.pm
 14:51
dukeleto this rabbit hole is deep 14:52
Parrot::Configure::Data seems to be it 14:54
dalek rrot: 8123d2d | moritz++ | lib/Parrot/Configure/Compiler.pm:
fix typo in lib/Parrot/Configure/Compiler.pm
moritz did I accidentally commit this in two branches? or was one a commit without a branch? 14:55
moritz confused
PerlJam perhaps it's just dalek that is confused.
whiteknight everybody is confused!
moritz right, it's the same hash twice 14:56
PerlJam overlapping RSS feed info?
moritz too distracted to investiage
dukeleto tests something 14:57
sorear: ping!
sorear: we are getting dupes, see above. Is this a bug on the github side, or on the hook side?
PerlJam aha! lib/Parrot/Configure/Step/List.pm 15:05
dukeleto PerlJam++++++++++ 15:08
dalek rrot/git_describe: c11d574 | dukeleto++ | config/auto/git_describe.pm:
[config] This almost adds the 'describe' key to the config hash
PerlJam parrot config system is a maze of twisty passages.
dalek rrot/git_describe: b18101c | dukeleto++ | lib/Parrot/Configure/Step/List.pm:
[config] Add git_describe to the config step list
 15:09
dukeleto PerlJam: all alike
the key is called 'git_describe', my previous commit msg is misleading
dalek rrot/git_describe: 4899abb | dukeleto++ | config/auto/git_describe.pm:
[config] Fix bugs and make the git_describe config key work
 15:12
15:13 s1n left
dalek rrot/git_describe: 0d855e0 | dukeleto++ | config/auto/git_describe.pm:
[config] Remove a dupe config key for our describe string
 15:16
15:16 davidfetter joined 15:17 Andy joined 15:26 bluescreen left, bluescreen joined 15:59 davidfetter left 16:04 Koven left
whiteknight is ready for this long week to be over 16:06
16:06 seanstickle joined 16:12 dmalcolm joined 16:15 Kovensky joined, lidden joined 16:17 theory joined 16:26 lucian left
dukeleto why is Parrot not mentioned? developers.slashdot.org/story/10/11...re-of-Java 16:30
moritz dukeleto: it is in one of the comments 16:31
and I can understand it well, because parrot isn't (yet) great for statically typed languages
and that's what the article is about
though I do admit that mentioning Go as an alternative to Java seems weird 16:32
dukeleto moritz: comments don't exist 16:35
moritz: but yes, i just saw that :)
16:37 seanstickle left 16:50 lucian joined
NotFound As usual, lot of people mix java language and java VM 16:55
I'm tempted to add a comment suggesting winxed as One language to rule them all. 16:57
Then other suggestng perl6, and start a flame war. 16:59
whiteknight We need to improve Parrot performance significantly if we want to get into that conversation. We need to increase execution speed by 50% at least
And when you look at something like firefox which had 3x performance improvements in its newest JS engine, that is something that we can do
atrodo NotFound> Oh, a fake flame war to increase parrot exposure. Nice idea 17:00
NotFound The funny thing is that in spite of all comments about the speed of compiled and strongly typed java, all java based applications are slow. 17:01
17:01 lucian left
NotFound So the thinhg looks like; "We need a fast VM in order to run horribly slow libraries and frameworks" 17:02
whiteknight not "all", but you're right that performance of java applications is never as good as it should be
atrodo NotFound> I think that's it right there 17:03
whiteknight many java applications are written as a simple mashup of existing libraries
NotFound Some time ago I read about how they make python in java look faster. They showed a prompt before finishing initialization X-) 17:05
atrodo Microsoft did that for years with office
NotFound I'm not opposed to use tricks to avoid abusing user patience, but using that tricks to support "java is not so slow" claims... 17:06
17:11 Kovensky left
whiteknight We don't need tricks. We need to do some serious hard work 17:18
17:18 C0ke is now known as Coke
jnthn fwiw, I have built pretty fast response stuff atop of a JVM before. Wasn't using a load of GUI frameworks and stuff though. :) 17:19
Coke writes code that compiles down to java that compiles down to bytecode every day and it's more than fast enough.
(and much faster than anything I do on parrot)
17:22 Kovensky joined
whiteknight right. Parrot has some major work to do 17:22
but we've known that
Coke aloha: tell bacek that aloha would be much more useful as a messagebot if you knew how old the messages were.
aloha?
dukeleto Coke: we need bacek to work on the GC, not irc bots ;)
Coke aloha: msg bacek that aloha would be much more useful as a messagebot if you knew how old the messages were. 17:23
aloha Coke: OK. I'll deliver the message.
Coke aloha: msg cotto: I have no idea how old your reparrot request is, but I just got it, and I added it on 10/24 17:24
aloha Coke: OK. I'll deliver the message.
Coke whiteknight: I just asked robert to update your creds... but I see you already got privs.
whiteknight Coke: yeah, he forwarded me a copy of an email he sent to me a while back that I had lost
stupid me was using the wrong password to get in 17:25
I forgot to use my svn.perl.org password, and hilarity ensued
Coke wonders if it's worth hacking on partcl again any time soon. 17:26
whiteknight what is the current state of partcl? 17:27
dukeleto Coke: sure, why not? Random segfaults?
17:29 lucian joined
dukeleto Lorito? 17:29
dukeleto shakes fists that there is no infobot in here
aloha, Lorito?
aloha dukeleto: Lorito is "little parrot" in spanish or trac.parrot.org/parrot/wiki/Lorito
Coke whiteknight: slow, changes in parrot causing me to fail tests, switch to git breaking my config script, 17:30
dukeleto Coke: i can help you with your config script
Coke: let me at it
Coke github.com/partcl/partcl 17:31
whiteknight Coke: there is space on smolder for partcl smoke reports to be uploaded. Would it be beneficial for us to get that set up?
dukeleto Coke: i have a commit bit. where is the file?
Coke er, github.com/partcl
atrodo aloha, Lorito is also github.com/atrodo/lorito
aloha atrodo: Okay.
atrodo aloha, Lorito is also github.com/ekiru/yalp-asm 17:32
aloha atrodo: Okay.
Coke dukeleto: the main script is the top level Configure.pl
atrodo aloha, Lorito?
aloha atrodo: Lorito is "little parrot" in spanish or trac.parrot.org/parrot/wiki/Lorito or github.com/atrodo/lorito or github.com/ekiru/yalp-asm
Coke there is also build/gen_parrot.pl, though that is copied from rakudo and I don't use it.
whiteknight: ideally, yes. realistically, no, because when tests fail, nothing happens. 17:33
whiteknight what do you mean nothing happens?
Coke I mean I report a parrot bug and nothing happens. 17:34
dukeleto sounds like something a product manager wants to hear...
whiteknight that's why I'm asking
Coke: Which is the current "master" for partcl? The old one or the new PCT-based one? 17:35
in the older version, I remember having a hell of a hard time debugging things 17:36
Coke no new work is being done on the old one. I only patch it so that "make test" continues to work and parrot gets exercised.
you can debug the new one? that makes one of us. ;) 17:37
17:37 nwellnhof joined
Coke as far as keeping the old one working with parrot, the last defect that was introduced I didn't bother fixing, I just skipped those tests. 17:37
(there's still an open parrot ticket, I think, but I gave up on it.)
whiteknight Coke: I'm on much more familiar ground debugging something pure-pir than I am when custom PMC types are mixed in 17:38
Coke: Any new and outstanding tickets for partcl can be assigned to me 17:39
Coke whiteknight: any tickets for partcl are marked with "lang: tcl"
enjoy. 17:40
s/are/should be/, to cover myself.
whiteknight well, all tickets marked tcl and without a current owner are now owned by me 17:44
in the future, when you have tickets for partcl, assign them directly to me
Coke whiteknight: you missed #1811 17:45
whiteknight: if I remember to, sure.
whiteknight oh, I didn't hit submit
plobsing ping dukeleto
tadzik atrodo: ping 17:46
atrodo tadzik> pong
Coke whiteknight: looks like you also missed #760, #966, #1028, #731...
#691 hasn't been touched in 16 months. 17:47
#721 in 13...
whiteknight I own all those
Coke #748 in 17.
tadzik atrodo: Lorito is meant to be a minimal set of opcodes, with ones more complex build from the core ones. Why is there say and write then? Isn't say supposed to be this “complex” one here, or it's just so commonly used it's core?
Coke whiteknight: not according to trac.parrot.org/parrot/report/16 and trac.parrot.org/parrot/ticket/760, e.g.
whiteknight tadzik: that's a good question. I don't think we should have write or say at that level. Those are methods on PMCs 17:48
atrodo tadzik> Those are in there more as a bootstrapping opcodes
not meant to stay long-term
whiteknight Coke: I'm looking at report 16 now and my name is next to all those tickets. Refresh your cache
Coke whiteknight: fix the cache at trac.parrot.org 17:49
whiteknight regardless, you're going to have to trust that I own them
Coke yup, the server cache is only aggressive for non-logged in users. Have fun. ;) 17:50
17:50 Kovensky left, Kovensky joined
cotto_work tadzik: yes. They're temporary ops to make debugging easier. They aren't expected to be in the final version. 17:51
whiteknight dukeleto: If you can fix partcl's configure script to work with git, I can get cracking on some tickets
Coke; I'm not going to fix the cache on the trac site. That's well beyond my purview 17:53
moritz whiteknight: I'm currently working on the rakudo configure system, branch parrotgit 17:56
I think much of that can be stolen for tcl
whiteknight moritz: yeah, dukeleto has already expressed an interest in doing it
and I've got enough on my TODO list without stealing tasks from him
dukeleto whiteknight: hokey dokey 18:07
moritz: yes, it can. i will steal your code for partcl :)
18:10 ash_ joined
Coke whiteknight: (cache) I don't expect you to fix it, just be aware of it. 18:12
dukeleto Coke: i am running the partcl test suite now 18:18
Coke dukeleto: is the work on parrot needed for this in trunk or in a branch? 18:20
18:21 chromatic left 18:22 ash_ left, ash_ joined, nwellnhof left
dukeleto Coke: it passed. I wasn't expecting that 18:22
Coke (i'm not seeing anything in parrot_config --dump on master, which is why I ask)
dukeleto: for old partcl? yah, because I skipped the last set of failing tests. 18:23
dukeleto Coke: git_describe branch
Coke ... weeks ago. you're right, parrot should have broken something new by now. ;)
whiteknight dukeleto: what's the ETA of that branch?
dukeleto whiteknight: when it gets done
whiteknight: it is in a state that Rakudo can use it, but it needs a lot more tests 18:24
whiteknight: perhaps, if the parrot community thinks it is important enough, i can merge the branch and add more tests in a branch. Doesn't matter to me.
dukeleto goes to the airport
cotto_work now you can commit on the plane 18:25
should you so desire
18:28 ash_ left
whiteknight dukeleto: considering the timing of everything, I do think it would be prudent to merge the branch as-is and work out the kinks later 18:29
if it doesn't break anything, and adds necessary functionality, mark it experimental and merge it 18:30
moritz +1 to merging now 18:31
cotto_work If it's going in before the release, it shouldn't happen after today. 18:32
btw, has anyone walked through the release process?
18:33 rurban_ joined 18:34 ash_ joined
whiteknight I have not walked through the process yet. I will try to do that soon 18:35
18:36 rurban left, rurban_ is now known as rurban
cotto_work cotto ought to take a shot at it too. For a dry run, he can skip some of the more time-consuming parts. 18:41
whiteknight we need to make sure that tcurtis has access to the ftp server with his ssh key
I'm not even sure if I still have ftp access to it 18:42
whiteknight being stupid, keeps deleting his ssh key
ah, RESPONSIBLE_PARTIES needs major updating 18:45
sorear seen dukeleto 18:46
aloha dukeleto was last seen in #parrot 22 mins 5 seconds ago saying "whiteknight: perhaps, if the parrot community thinks it is important enough, i can merge the branch and add more tests in a branch. Doesn't matter to me.".
sorear dukeleto: do you know who set up push notifications for Parrot? 18:47
cotto_work I can give him access to the ftp server.
whiteknight sorear: why? are the push notifications wrong? 18:53
18:53 davidfetter joined
sorear whiteknight: yes 18:54
whiteknight: specifically, he or she forgot the "de-configure pull notifications" step
which is why every commit to master is being reported twice
I could fix this, but I'd rather be certain that he or she won't do it again, first 18:55
18:57 M_o_C joined 19:11 ash_ left 19:21 ash_ joined
whiteknight blah. I have to install yet another version of VisualStudio on this damned machine 19:42
brings me up to 3
cotto_work whiteknight: where's the new $dayjob? 19:43
dukeleto sorear: ping
sorear: i set up the push notifications 19:44
whiteknight cotto_work: center city philadelphia. A web-dev place called weblinc
dukeleto sorear: de-configure pull notifications?
19:50 seanstickle joined 19:53 mikehh joined 19:54 silug joined
moritz dukeleto: you can do that too 19:58
sorear dukeleto: there is an entry for parrot master in perl6/mu/misc/dalek-conf.json; 20:00
it should have been removed when push for parrot was set up
I'll do that now
dukeleto: What repos did you set up push on? 20:02
dukeleto sorear: i only set it up for parrot.git 20:03
sorear++
sorear: i didn't realize there was a dalek-conf file to fiddle with.
whiteknight dukeleto: a lesson I've learned over the years: there is ALWAYS a conf file to fiddle with 20:11
the first half of the fun is finding the damn file. The second half is scraping through extremely concise and poorly-written online documentation to figure out how to do even basic things with it 20:13
sorear whiteknight: there has only been a conf file for a month or two 20:14
for most of dalek's history, it was configured by writing Perl modules
atrodo whiteknight> Just to find out the documentation on the website is old because the author was halfway through refactoring it 20:16
so option "DTRT_now" is no longer supported because no one used and it's no longer important, but it's the single option that makes it work perfectly for you 20:17
20:22 seanstickle left 20:26 hercynium joined 20:50 bluescreen left 20:56 M_o_C left
moritz aloha: msg dukeleto the parrotgit rakudo branch now works, in the sense that success has been reported from two different machines so far. Eagerly avaiting the merge of the git_describe branch in parrot :-) 20:59
aloha moritz: OK. I'll deliver the message.
21:02 perlite_ joined 21:06 perlite left, perlite_ is now known as perlite 21:27 whiteknight left 21:43 fperrad left 21:57 ash_ left 22:00 ash_ joined 22:03 ash__ joined 22:04 ash_ left, ash__ is now known as ash_ 22:28 hercynium left 22:31 masonkramer left 22:33 masonkramer joined 22:45 Andy left 22:59 ash_ left, seanstickle joined 23:12 TonyC_ joined, TonyC_ left 23:46 whiteknight joined 23:51 dngor left, dngor joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!