Parrot 3.10.0 "Apple Pi" | parrot.org | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 2 December 2011.
dalek rrot/release_4_0: 02324c9 | Whiteknight++ | / (10 files):
Get all the metadata in place for Parrot 4.0, 'Hyperstasis'
 00:01
whiteknight dukeleto: branch pushed, feel free to modify NEWS and whatever
...and make fulltest again with the tarball 00:10
even with all the wonderful improvements that have happened, this whole thing still cries out for more automation
cotto yup 00:11
whiteknight dukeleto: any changes to ChangeLog forthcoming? 00:18
I've made the tarball and refulltested it
dukeleto whiteknight: give me a minute 00:26
whiteknight dukeleto: all the time in the world
cotto whiteknight: did you see that mozilla released a metacircular js interpreter? 00:29
whiteknight yeah
00:30 bacek joined
dalek rrot/release_4_0: c0cd33d | dukeleto++ | ChangeLog:
[doc] Give the changelog some love
 00:30
dukeleto whiteknight: there you go
cotto: i knew about narcissus (js in js), what is the new part? 00:31
cotto dukeleto: nothing new
whiteknight I'm re-doing the tarball with duke's changes 00:42
00:45 kid51 joined
whiteknight tarballs are uploading now 00:47
...and they're online 00:49
dukeleto whiteknight++ 00:51
cotto whiteknight++ 00:52
whiteknight now for the release announcements, which are by far my least favorite part 00:53
cotto ~~ 01:38
dalek rrot: 02324c9 | Whiteknight++ | / (10 files):
Get all the metadata in place for Parrot 4.0, 'Hyperstasis'
 01:57
rrot: c0cd33d | dukeleto++ | ChangeLog:
[doc] Give the changelog some love
02:35 whiteknight joined
whiteknight ...I'll get these release announcements out eventually, if my computer will stop freezing up and being stupid 02:39
dalek website: Whiteknight++ | Parrot 4.0.0 "Hyperstasis" Released! 02:46
website: www.parrot.org/news/2012/Parrot-4.0.0
whiteknight ...and with that, I'm off to bed. Give all your complaints to aloha, I'll sort through them in the morning 02:54
dukeleto whiteknight++ # releasing 03:15
03:43 Psyche^ joined 04:07 ruoso joined
dalek kudo/nom: 42068aa | moritz++ | src/core/Range.pm:
remove debugging statement, moritz-- kboga++
 06:03
alvis cotto: ping 06:45
06:48 bacek joined
cotto alvis, pong 06:50
alvis hello. i understand the credentials concern and OAuth2 is not a solution, but ... what about a dummy repo for issues that throws issues events?
cotto: sorry, i'm still thinking about the parrotbug thing. 06:51
cotto hmmm 06:52
fundamentally, we're trying to allow anonymous access to github issue creation without allowing spamming
which won't work 06:53
I do want to get parrotbug resurrected, if we can find a sane way to do so. 06:54
alvis ok. well, what i'm thinking - though it's a'bit hacky - is a dummy repo that accepts issues ...
cotto perhaps pulling information from git on the local machine
yes, but that adds another layer that may not be necessary 06:55
alvis yes, that's true. we could do a local machine ... hmmm, let me think about that a'bit... 06:56
cotto are issues stored in a git repo that's in any way externally accessible?
I don't think that's the case, but I'd enjoy being wrong.
alvis let me check a few more things and i'll get back to you ... 06:57
like you ... i'd like to make parrotbug useful. 06:58
i'll let you know, hopefully, tomorrow what i come up with. thanks.
cotto If github allowed ssh keys to be used for api authentication, we could go that route
but they don't
alvis right.
cotto If we could set up a sufficiently locked-down account, we could use that and put the credentials in git. I'm not sure if an account can be sufficiently locked-down thouhg. 07:00
alvis i don't think we can, unless it's just an issues-type repo and, then, as issues come in, it throws an event, which we can then catch. 07:01
cotto which was your original suggestion 07:02
alvis yeah, that's all of got for now. ...
cotto If that's all that's possible, +1.
alvis but i'll dig around a'bit more. (the staff is unhelpful; they just pointed me to the API and said OAuth2) 07:03
not to criticize. they're a helpful bunch. they just don't have an alternative solution for now.
07:05 jsut_ joined
alvis and just to be clear: i mean a repo *only* for parrotbug submissions (but i really like the pull from git idea). 07:21
moderator Parrot 4.0.0 "Hyperstasis" | parrot.org | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC 07:35
cotto alvis, feel free to start experimenting 07:41
sorear hope I did that right 07:42
cotto sorear, wfm 07:43
08:25 mj41 joined 09:34 lucian joined
dalek kudo/nom: ed943e6 | moritz++ | src/core/ (2 files):
stringify stuff before passing it to the regex engine

Otherwise lots of tests break; though not sure if this is the right solution		 09:34
10:37 donaldh joined 10:38 donaldh joined 11:51 seme joined
seme hi guys... 11:51
11:54 seme left, seme joined
seme hey guys I'm trying to walk through the parrot documentation and when trying to implement the try statement I get the following error, Symbol '$S0' not predeclared in statement:sym<try>.... not sure exactly why though 11:55
ok I got it... someone may want to update the documentation... I'll see if there is an appropriate bug list 12:03
12:13 benabik joined
benabik I exist again! 12:22
12:55 preflex joined
dukeleto ~~ 13:14
seme: showing us the exact script you tried to run and the full error would help 13:15
seme: as well as parrot --version
13:24 bluescreen joined
dukeleto Woke up this morning and there is only decaf coffee in my hotel room. Pure Evil 13:24
benabik decaf-- 13:25
dukeleto benabik: exactly 13:35
benabik cold coffee no good either. Bleck. 13:36
Iced coffee good, hot coffee old enough to be room temp bad. 13:37
13:47 jsut joined, bluescreen joined
seme thanks dukeleto I figured it out. I sent an email to the mailing list (it may be an error in the tutorial) 13:48
13:57 bluescreen joined
dukeleto seme: if you find errors in our docs, please submit patches :) 14:02
donaldh parrot is not detecting x86_64 on MacOS Lion
I'd like to fix it but am worried about breaking older MacOS versions 14:03
What's the ususal approach to fixing configure problems like this? 14:04
An additive patch will fix my specific problem but make the code messier. I'd prefer a cleaner fix but have no way to verify it 14:05
moritz the usual approach is to try, and wait until somebody whines
donaldh Sure, okay.
I could also gather sample data from mac owners I guess. 14:06
seme dukeleto: where? I looked on the docs page but didn't find anywhere to download or submit bugs?
moritz submit bugs at github.com/parrot/parrot/issues/new
donaldh moritz: what about patches? Same place, or pull requests? 14:07
moritz donaldh: pull requests
donaldh moritz: is the process described anywhere? 14:09
moritz donaldh: help.github.com/send-pull-requests/ 14:11
donaldh moritz: thanks. 14:12
dalek rrot/github-links: b95dba6 | moritz++ | docs/submissions.pod:
simplify docs/submissions.pod by linking to githubs help page for pull requests
 14:15
rrot/github-links: b10eec1 | moritz++ | docs/ (2 files):
update commit bit management to use github
 14:20
14:24 donaldh joined
aloha (parrot/parrot) Issues opened : 696 (The '"' quote prevents it from compiling) by johnrizzo1 : github.com/parrot/parrot/issues/696 14:33
seme moritz: thanks.. I think I submitted a fix on github directly
14:38 PacoAir joined
dukeleto donaldh: what do you mean it is not detecting x86_64 on Lion? 14:50
dalek rrot/release_4_0: 0ad70dc | dukeleto++ | examples/languages/squaak/doc/tutorial_episode_4.pod:
Fix bug in squak tutorial noticed by John Rizzo
 14:52
dukeleto blarg. wrong branch
moritz makes the trac -> github issues mapping
dukeleto moritz++
dalek rrot: a8aab9f | dukeleto++ | examples/languages/squaak/doc/tutorial_episode_4.pod:
Fix bug in squak tutorial noticed by John Rizzo
 14:53
dukeleto deletes the release_4_0 branch, since it has been cut
14:53 whiteknight joined
dukeleto whiteknight: mornin' 14:56
whiteknight: i just removed the release_4_0 branch, since it is out 14:57
aloha (parrot/parrot) Issues closed : 696 (The '"' quote prevents it from compiling) by johnrizzo1 : github.com/parrot/parrot/issues/696 14:58
moritz moritz.faui2k3.org/tmp/parrot-trac-github.txt is the mapping from github to trac 15:01
benabik moritz++ 15:03
dalek rrot/github-links: 0d4aed0 | duff++ | docs/submissions.pod:
More text updates
 15:04
rrot/github-links: 4880d10 | duff++ | src/string/api.c:
Replace trac ticket url with github issue url
PerlJam moritz: I've noticed that sometimes the trac ticket URLs are http, sometimes https 15:06
moritz comes up with a very inefficient substitutions script that would make Don Knuth turn in his grave if he were dead
benabik Efficiency doesn't matter too much for a one-off script. 15:07
moritz PerlJam: you mean, in the repo?
donaldh dukeleto: Configure.pl detects i386 as the architecture on Lion. I am running 64 bit. It uses uname -p which reports i386 15:08
moritz benabik: it makes 450 passes through every file in the parrot repo 15:10
benabik moritz: That is... notably inefficient.
Hm. My parrot_config reports cpuarch => 'i386' but HAS_INT64 => '1' 15:11
PerlJam moritz: aye
donaldh benabik: on MacOS ? 15:12
benabik donaldh: On Lion, yes
donaldh benabik: same problem as me. uname -m correctly reports x86_64 but uname -p annoyingly says i386 15:13
benabik donaldh: What problem does that cause, exactly? parrot says it's using 64b ints and file reports that it's a x86_64 executable 15:14
(My parrot)
donaldh I only notice the problem far downstream when trying to use zavolaj on rakudo to dynamically load libraries.
Parrot might be building correctly but it's reporting its own config wrongly to HLLs 15:15
benabik Is zavolaj using the value of cpuarch to affect compilation? I thought it just accessed existing libraries. 15:16
donaldh (assuming that cpuarch should = x86_64 or amd64 or somesuch on a 64 bit OS.
PerlJam Are there any plans for importing the closed tickets from trac into github?
donaldh benabik: nope, zavolaj uses nqp functionality that doesn't link correctly, due to architecture mismatches at link time. 15:17
The symptom is missing symbols at runtime.
benabik donaldh: Try `file /usr/local/lib/libparrot.dylib`... If that's reporting itself as x86_64, then that's probably not the issue. 15:18
IIRC, Zavolaj uses dlopen which should open things based on the architecture of the program not a value in parrot_config 15:19
(Although I would be interested to know if parrot reports cpuarch x86_64 in 64b Linux)
donaldh As I said, I don't know whether parrot is building itself correctly or not. I do know config/auto/arch detects i386 even though it is meant to report amd64. 15:20
Then rakudo and nqp build against a reported arch of i386 so they build wrongly.
nqp silently fails to link a few necessary files because of architecture mismatch. 15:21
At runtime Zavolaj tries to use dlLoadLibrary which wasn't linked due to architecture mismatch. 15:22
benabik I don't see NQP doing anything based on cpuarch...
15:24 japhb joined
dalek rrot/github-links: 8621563 | duff++ | src/ops/sys.ops:
add another github issue url
 15:26
donaldh benabik: perl6 nqp Configure.pl needs to. It doesn't just now. The only thing that _can_ correctly report the architecture from parrot is cpuarch.
benabik Ahhh... I see it now 15:27
donaldh :)
dalek rrot/github-links: de92e5e | moritz++ | / (3 files):
update ticket links from trac to github

Since many tickets have been closed before migrating, many trac URLs remain
moritz PerlJam: that's all that my automated search has found
PerlJam moritz++
moritz a rather poor result for 20 minutes of scraping and fiddling
benabik Why is dyncall using -arch i386 anyway? 15:28
dukeleto ~~
moritz dyncall has its own, simplistic configure system, it seems
donaldh dyncall is also broken.
but it's third party so I chose to feed it the correct cpuarch (on my branch)
benabik I'm confused because it looks like dyncall is trying to build universally by default. 15:30
PerlJam moritz: There are still RT ticket URLs in the parrot repo that currently link to no where (in the small spot checking that I've done). There are many more trac ticket URLs that don't exist in github. So, if we want to maintain their informational content, we'll need to keep track running (but with copious notes that point to github) or move over all of the remaining trac tickets even if they are closed.
moritz: What do you think?
donaldh when nqp calls configure, it reports 'guessing i386' or somesuch. I'm at work and don't have my Mac handy.
PerlJam s/track/trac/ 15:31
moritz PerlJam: the right thing to do is to make trac read-only, and put a huge notice on top saying that all new discussion should happen on github
whiteknight Trac should be effectively read-only at this point, except to admins 15:32
moritz that's good
benabik HAH!
donaldh benabik: oh, that's right, dyncall tries to infer 64bitness from the OS major version 10.
PerlJam except that trac would be another service to maintain
benabik dyncall is being very dumb.
donaldh But Lion reports 11
benabik Right.
PerlJam another thing to worry about during upgrades.
benabik It should be $VERSION_MAJOR -ge 10
whiteknight there's an argument to be made for not including specific ticket numbers/urls in the source code at all
donaldh yep
moritz whiteknight: I found them generally useful 15:33
donaldh But such an unreliable test. I'm sure there are 32 bit kernels still.
benabik Or just use uname -m
donaldh quite.
PerlJam whiteknight: so how would you link to further discussion? Or would you replicate the results of discussion in comments in the code?
moritz parrot>ack -aQ 'trac.parrot.org' |wc -l
83
donaldh It's one of those bugs that leaves you patching 4 products :( 15:34
PerlJam moritz: only 56 for ack -aQ 'trac.parrot.org/parrot/ticket' |wc -l
benabik There are some wiki pages that haven't migrated. 15:35
moritz so 83 - 56 left that we can tackle easily
aloha (parrot/parrot) Issues opened : 697 (The statement_list rule was spelled) by johnrizzo1 : github.com/parrot/parrot/issues/697 15:38
benabik Hm. I wonder what uname -m reports on powerpc Mac. 15:40
moritz any objections to merging branch github-links into master?
I only did the stuff in a branch because we were shortly before a release
benabik We run uname -p on everything != windows, but only use it on darwin. 15:41
Coke benabik: ask kid51.
he's our lone powerpc guy (after my laptop died with no hope of repair.) 15:42
donaldh benabik: yep. My preferred patch would be to change that to uname -m
But afraid of breaking older macs
Coke so make a patch and have him test it? 15:44
donaldh coke: sure thang
PerlJam moritz: I have no objections :) 15:45
donaldh coke: that was moritz's advice an hour ago ;-)
benabik Does 10.6 compile 32b by default?
hmmmm...
seme can anyone tell me if by episode 5 of the squaak tutorial I should be able to "print"? there are a couple of references but as far as I can tell print doesn't work. if I do I get an syntax error which isn't very helpful. pastebin.com/TEJrq62b 15:46
benabik uname -m supposedly reports i386 on 10.6 even on 64b systems because the kernel is 32b
dukeleto moritz: merging is fine with me
seme: perhaps it is supposed to be "say" ?
dalek rrot: 0d4aed0 | duff++ | docs/submissions.pod:
More text updates
rrot: 4880d10 | duff++ | src/string/api.c:
Replace trac ticket url with github issue url
rrot: 8621563 | duff++ | src/ops/sys.ops:
add another github issue url
rrot: de92e5e | moritz++ | / (3 files):
update ticket links from trac to github

Since many tickets have been closed before migrating, many trac URLs remain
rrot: 8b3aa4a | moritz++ | / (34 files):
Merge remote branch 'origin/github-links'
seme dukeleto: I don't think so. Early on in the tutorial it had me remove say... let me check that again 15:47
benabik donaldh: Yes. Just using uname -m won't be sufficient to fix 10.6 15:48
dalek rrot: fd87a61 | dukeleto++ | examples/languages/squaak/doc/tutorial_episode_5.pod:
Merge pull request #697 from johnrizzo1/patch-2

The statement_list rule was spelled
seme yeah in episode 3 it has you remove the say rule 15:49
aloha (parrot/parrot) Issues closed : 697 (The statement_list rule was spelled) by johnrizzo1 : github.com/parrot/parrot/issues/697
donaldh benabik: oh, ok.
dukeleto Something faster than the FFT: arxiv.org/abs/1201.2501v1 15:56
moritz funny, I just had a paper open that talked about parallel FFT implementations 15:58
dukeleto it is for sparse data, but still, looks like some very nice speedups 16:00
benabik donaldh: If you can add -arch_errors_fatal in there, that would be useful.
masak heh, I also just found this: web.mit.edu/newsoffice/2012/faster-...-0118.html
benabik thinks that should bloody well be the default.
masak (which is about the same paper)
benabik Silently not linking things, blah.
donaldh benabik: where is there? In nqp ?
benabik donaldh: It's a option for ld. If you can make nqp use it when it's linking in dyncall, it won't silently fail. 16:01
donaldh benabik: Cool. Is that a Darwin specific flag?
benabik donaldh: Likely.
donaldh: `gcc -dumpmachine` may be the most reliable. 16:07
donaldh benabik: what does that give you on Darwin? 16:15
benabik Oh, that just gives i686.
benabik grumbles.
16:20 dmalcolm joined
donaldh benabik: a link on stackoverflow suggests using 'sysctl hw.optional.x86_64' and 'sysctl hw.cpu64bit_capable' 16:22
benabik: perhaps hw.cpu64bit_capable is sufficient for us ? 16:25
benabik donaldh: Probably... I just fear that some vernon of the OS reports that a 64b processor is available but then compiles to 32b
donaldh does parrot use the default target of the compiler? 16:26
benabik I think so.
16:41 jsut_ joined 16:49 dmalcolm_ joined
whiteknight we're using mercurial for some projects at work now 17:15
and I have to say that my initial impressions of it so far are...underwhelming
although that may be related to the fact that we're on windows and the tortoiseHG that we're using as the client feels extremely flimsy 17:16
PerlJam whiteknight: when I first switched away from svn, I looked at bzr, hg, and git. git won me over by a wide margin. 17:17
seme hey guys... I found another error in squaak episode 6 but I'm not confident I know what I'm doing... if you follow up to episode 6 there will be a method "stat_or_def" defined... up until now that was fine since we didn't have defined sub's... now that we do it breaks things unless you remove it. I think that is all that needs to be done but I'd like someone else to check before I submit a bug... 17:19
PerlJam notes that even the php community voted to switchto git instead of hg :)
seme or should I just submit a bug?
PerlJam seme: what breaks? 17:26
seme PerlJam: basically prior to episode 6 we had a rule stat_or_def { <statement> } defined and we had an action defined for it which just returned a make $<statement>.ast but now that we changed the rule to rule stat_or_def { | <statement> | <sub_definition> } we either need to update the stat_or_def method or remove it(I think) 17:28
removing it works as far as I can tell but I've never done anything like this before so I'm not very confident
PerlJam seme: seems like updating the action method would be the right thing. 17:31
seme odd... i'm looking at the code for the example now and it doesn't really match the documentation....
there are some inconsistencies in the naming of things 17:32
confusing for a newbie
benabik Squaak is not the best written tutorial. :-/
seme he hee
well I've gotten this far so I'm not giving up :)
PerlJam seme++
seme ok it looks like the docs forgot to mention that you need to update the method... if you change it to check for $<statement> and if that doesn't exist just assume that it is a sub_definition... hrm... dubious 17:35
ok I'm heading home... I'll submit a patch when I get there
PerlJam looks at the docs
It says "Appropriately modifying the action method is simple." right after the updated stat_or_def rule 17:36
that's to imply that you should do so.
seme he hee I see... I didn't read that as something I should do... I read it as indicating that later it would be described.... but I see what you mean 17:37
PerlJam seme: a patch to improve the wording can only help :)
seme perhaps in addition to trying to learn how to write a programming language I should try to learn how to read English ;)
PerlJam seme: and I see your point too. It could look like an introduction to the modification that they're about to show you, only the mod isn't there. 17:38
seme ok I've put in an update. I'm going to try and finish the tutorial tonight and also make the code consistent with the tutorial. 17:42
17:43 bluescreen joined
dalek rrot: 9e4d1ca | (John Rizzo)++ | examples/languages/squaak/doc/tutorial_episode_6.pod:
It was not clear that this was an action for someone following the tutorial. I've changed the language to clearly state that the follower of the tutorial should update that action method.
 17:45
rrot: b2f3e44 | dukeleto++ | examples/languages/squaak/doc/tutorial_episode_6.pod:
Merge pull request #698 from johnrizzo1/patch-3

Documentation is not clear that the user should perform a change necessary for completing the tutorial.
aloha (parrot/parrot) Issues opened : 699 (cleanup cygwin build) by gerdr : github.com/parrot/parrot/issues/699
benabik dukeleto: Beat me to it.
dukeleto benabik: :)
17:45 fperrad joined
dukeleto benabik: i found an espresso machine 17:45
benabik seme++
dukeleto: Awesome! 17:46
17:46 bluescreen_ joined
dukeleto goes for a sunny run 17:46
17:46 bluescreen joined 18:05 not_gerd joined
not_gerd hello, #parrot 18:05
benabik hello, not_gerd 18:06
not_gerd msg alvis according to your config_lib.pir, you *did* use the cross-compiler, but with an incorrect -I flag - no idea where that came from... 18:09
aloha OK. I'll deliver the message.
Coke seme++
not_gerd msg alvis if you have some more time, could you also take a look at github.com/parrot/parrot/pull/699 ? 18:10
aloha OK. I'll deliver the message.
not_gerd needs to restart cygwin 18:17
18:17 not_gerd left 18:26 not_gerd joined 20:22 donaldh joined 20:37 mj41 joined 20:38 contingencyplan joined 21:10 not_gerd left
dalek p: 76377d0 | jonathan++ | src/Q (2 files):
Try to fix the hangs that implementing backtracking into subrules introduced.
 21:14
p: baf0b9b | jonathan++ | src/QRegex/Cursor.nqp:
Fix for !BACKREF, which fixes test 42.
 21:37
22:01 dmalcolm__ joined 22:11 dmalcolm_ joined 22:33 seme joined
bacek_at_work ~~ 22:58
23:14 PacoAir joined 23:26 donaldh 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!