Parrot 1.7.0 "African Grey" is coming soon, no non-release-related commits to trunk, please | Improve PCC testing this week or the release gets it!
Set by moderator on 20 October 2009.
00:00 Zak joined
Whiteknight incoming 00:06
purl i think incoming is pause.perl.org/incoming/
cotto_work covers ears and ducks 00:07
Whiteknight haha, some of these matrix operations are so slow doing them with nested RPAs in PIR 00:08
darbelo is reading the cblas docs.
That will probably probably unslow them.
dalek trixy: a7571ec | Whiteknight++ | toolbox/matrixyrc.m:
update the matrixyrc.m file to have some explanations and not call functions that don't exist
 00:09
trixy: 9afd842 | Whiteknight++ | t/100-fortran_helpers.t:
the fortran_helpers test file relies on stuff from parrot-linear-algebra, so marking that TODO until we get that project working
shorten dalek's url is at xrl.us/bftcns
shorten dalek's url is at xrl.us/bftcnu
dalek trixy: 61ccc7b | Whiteknight++ | t/300-transpose.t:
transpose doesn't use BLAS (currently) so we can un-TODO this test
shorten dalek's url is at xrl.us/bftcnw
dalek trixy: c79e40e | Whiteknight++ | t/308-loadlibrary.t:
loadlibrary tests rely on the NCITEST library, which isn't installed with Parrot make install. Need to rethink those tests
shorten dalek's url is at xrl.us/bftcny
dalek trixy: 9de8a4d | Whiteknight++ | t/800-loadlibrary-examples.t:
the loadlib-examples test tries to load BLAS, which we don't support just yet. marked as TODO
shorten dalek's url is at xrl.us/bftcn2
Whiteknight yay! I'm giting! 00:10
00:12 Aisling joined 00:13 payload joined
dalek trixy: 73c3d83 | Whiteknight++ | t/009-dispatch.t:
there are 15 dispatch tests
 00:14
shorten dalek's url is at xrl.us/bftcox
darbelo Whiteknight: parrot-linear-algebra has a Configure.nqp now. 00:15
Whiteknight darbelo is a goddamn rock star
dalek rrot-plumage: 84286d9 | japhb++ | :
[TOOLS] import_proto.p6: more refactoring; had to switch := to = because...
 00:16
shorten dalek's url is at xrl.us/bftco3
darbelo To be fair, sir. The car's mostly stolen, I didn't build it all myself.
And there isn't all that much to configure either. 00:17
cotto_work Whiteknight, did you notice TT #1125? It looks like you ran into the same bug in trying to write the matrixy operators. 00:18
dalek rrot-linear-algebra: dfbfd5c | darbelo++ | Configure.pl:
Say goodbye to Configure.pl and make room for it's replacement Configure.nqp
 00:19
rrot-linear-algebra: 84285f4 | darbelo++ | (2 files):
Add Configure.nqp (based on the one from plumage) and it's (somewhat slimmed down) helper library.
shorten dalek's url is at xrl.us/bftcph
rrot-linear-algebra: 69a28b6 | darbelo++ | config/Makefile.in:
Remove #IF()s from the Makefile.in template, Configure.nqp doesn't support them.
shorten dalek's url is at xrl.us/bftcpj
shorten dalek's url is at xrl.us/bftcpm
darbelo I0m guessing the next step for parrot-linear-algebra is lib detection. 00:20
After that we take the detected library and build a PMC arround it. 00:21
Whiteknight: en.wikipedia.org/wiki/BLAS#Implementations
Looks like we have plenty to choose from! 00:22
Whiteknight yeah, I've been looking at that list all day 00:23
dukeleto suggested GSL
that or ATLAS 00:24
ideally, we'll be able to detect any of them
cotto_work but you have to start somewhere
darbelo OpenBSD has cblas and GSL in ports, so my choices are a bit limited on that aspect. 00:27
Whiteknight OpenBSD doesn
t have ATLAS?
look for atlas3
darbelo Nope, no ATLAS. 00:29
You can check at openports.se/ it has a cute web search interface into the ports collection and all.
But I did a make search and it didn't show up.
00:31 Austin joined
darbelo Still, BLAS is pretty damm standard, once we have one going, the rest are a matter of getting the compiler and linker flags right. 00:34
cotto_work darbelo, you're not using NCI? 00:36
darbelo Yes, but we also want an efficient matrix PMC. That's going to have to be made in C 00:37
Also, the cblas bindings are complex enough that any non-frame-building parrot is going to need a recompile. 00:39
cotto_work I'm really looking forward to the point when a Lorito-generated C PMC isn't that much less efficient than a pure-C pmc.
darbelo A non-C binding is likely to cause a greater gc pressure as well, as there's going to be a lot of unmanaged struct flying about. A pure-C PMC can do away with that and manipulate the buffer pointers directly. 00:42
00:50 szabgab joined
darbelo purl: msg japhb Added support for nqp Configure stages to plumage. 00:51
purl Message for japhb stored.
00:52 Eevee joined, abqar joined
darbelo purl: msg japhb parrot-linear-algebra, and soo matrixy use them. It's very basic, but it works for what they need. You might want to extend it. 00:52
purl Message for japhb stored.
Whiteknight BLAS is standard in FORTRAN 00:53
there are no standard bindings in C
dalek rrot-plumage: 7d9b611 | darbelo++ | :
Adjust the configure-type of parrot-linear-algebra to nqp.
shorten dalek's url is at xrl.us/bftc34
dalek rrot-plumage: 7f6a78c | darbelo++ | :
Add support for projects with a NQP Configure stage.
shorten dalek's url is at xrl.us/bftc36
Whiteknight so, for instance, the CBLAS and ATLAS API functions are a little different 00:54
I assume GSL is a little different too
darbelo From what I can see, the basics are the same everywhere. We should be able to suport most with very little #ifdefing. 00:55
dukeleto is back again
is stuff passing on win32 now?
Whiteknight dukeleto: yes, 100% 00:56
purl 1
darbelo dukeleto: I think so.
dukeleto Whiteknight: SWEET
Whiteknight darbelo: the APIs are all similar
dukeleto is going to do 'make release' again, run my tests and kick it out the door
Whiteknight CLAPACK is just a translation of FORTRAN to C, so all parameters are passed by reference
ATLAS is more of an intelligent rewrite, so some things are passed by value
dukeleto++
darbelo dukeleto++ 00:57
darbelo wants to break trunk, dammit!
dukeleto darbelo: soon, my son 00:58
as soon as my tests pass i will tag the release 01:01
and then darbelo can set trunk on fire 01:02
cotto_work Who's planning on syncing pcc_reapply with trunk? That could be done now without messing up dukeleto.
01:02 cconstantine joined
Whiteknight allison suggested she was going to do it today at #ps 01:03
dukeleto yes, r41963 is my release candidate
cotto_work so she did
(not that it matters terribly who does it) 01:04
dukeleto what does 'sh into the parrotvm, and in the webroot for docs.parrot.org, expand these into a release directory 01:06
s/sh/ssh/ 01:07
what is the 'parrotvm' machine?
this is in the instructions for updating docs.parrot.org
darbelo Coke: should know.
Austin Probably the server for parrot.org
Duke, do you have an account? 01:08
Whiteknight that part of the guide must be new
ask Coke 01:09
Austin Boy. I was just going to say "Whiteknight and cotto have both done it recently. Ask them."
darbelo I think I recall Coke adding it today. 01:10
Austin So print out yesterdays version, and keep going.
darbelo let me check the back log.
dukeleto Under "Administer" āˆ’> "Site building" āˆ’> "URL Redirects", change the URL for "release/current" to the FTP directory for the new release
i can't find the 'administer' button on parrot.org, perhaps I don't have the rights? 01:11
Austin: not ssh access, I don't think
dalek rrot-linear-algebra: 3b9beb6 | darbelo++ | LICENSE:
Add LICENSE file as suggested in Whiteknight++'s blog tasklist.
shorten dalek's url is at xrl.us/bftdhm
darbelo Yep, added today: irclog.perlgeek.de/parrot/2009-10-20#i_1622099
Whiteknight awesome!
purl awesome is a window manager or at awesome.naquadah.org or awesome!
darbelo well, yesterday, on some timezones. 01:12
Whiteknight purl msg pmichaud I need to add an initialization sub to my PCT-based parser. I need to call a function before the parser runs, for each instance of it. How's the best way to add that?
purl Message for pmichaud stored.
cotto_work heads out, hoping that pcc_reapply will be merged by the time he gets home several hours later. 01:15
dalek trixy: 9821908 | Whiteknight++ | matrixy.pir:
rearrange the main file a little bit to do what it should
 01:16
shorten dalek's url is at xrl.us/bftdkg
dalek trixy: 264840d | Whiteknight++ | (4 files):
move helper functions out of the matrixy.pir driver file and into src/internals/*
shorten dalek's url is at xrl.us/bftdki
dalek trixy: 9e45f20 | Whiteknight++ | (2 files):
Add a LICENSE file to the project and update the copyright notice in the prompt to reference it
shorten dalek's url is at xrl.us/bftdkk
dukeleto these tests just keep running..... 01:18
what kind of crazy people write all of these tests?
darbelo I guess I won't be breaking trunk today, then.
darbelo has to head out in a few minutes. 01:19
dukeleto darbelo: i am running fulltest on the release tarball, then I will tag and bag and you all are free to nuke trunk from orbit 01:20
Austin Why doesn't parrot unbox as well as boxing? 01:21
darbelo 'cause that would be too easy ;) 01:22
Austin shrugs.
dukeleto svn: Cannot mix repository and working copy sources
Austin I'm looking in the manhattan distance stuff in multidispatch. 01:23
dukeleto (kadath)(~/svn/parrot-1.7.0-release)$ echo $SVNPARROT
svn.parrot.org/parrot
(kadath)(~/svn/parrot-1.7.0-release)$ svn copy āˆ’r41963 -m "tagged release 1.7.0" $SVNPARROT/trunk $SVNPARROT/tags/RELEASE_1_7_0
Austin +1 distance penalty for autobox. +INF for needing unbox.
dukeleto svn: Cannot mix repository and working copy sources
what is up with that?
that is what the release guide says. and that ? is a -. stupid copy and paste encodings
Austin Try ...trunk@41963 ? 01:25
darbelo I don't know, but it worked for me :)
dukeleto Austin: what full command can I do? 01:26
and if someone else can make that tag, please do it
darbelo dukeleto: tagged it for you.
dukeleto darbelo: YOU FREAKIN ROCK
Austin dukeleto: what version of svn are you using?
dukeleto 1.6.5 01:27
darbelo for the record I did: "svn copy -r41963 -m "tagged release 1.7.0" svn.parrot.org/parrot/trunk svn.parrot.org/parrot/tags/RELEASE_1_7_0" with 1.6.5 (r38866)
Austin Ahh. You've probably got a newer version than parrot.org does.
darbelo Maybe your $SVNPARROT env var was wonky?
dalek rrot: r41964 | darbelo++ | tags/RELEASE_1_7_0:
tagged release 1.7.0
 01:28
dukeleto crap, can someone copy the tarball over for me? I can't seem to login to ftp-osl.osuosl.org 01:30
darbelo That one, I can't help with.
maybe Whiteknight or cotto can do it, they've released before. 01:31
Whiteknight eh?
darbelo And on that note, I'm leaving.
01:31 darbelo left
Whiteknight dukeleto: did you add your ssh key? 01:31
or, did you give the key to allison to add for you?
dukeleto Whiteknight: i thought I did. i have ssh'ed before. yes, allison set me up
OH 01:32
Whiteknight pm
dalek trixy: 78ad413 | Whiteknight++ | matrixy.pir:
add some comments and TODO notes in matrixy.pir for thigns that need to change
 01:33
shorten dalek's url is at xrl.us/bftdxk
dukeleto $ ssh -lparrot -i ~/.ssh/parrot.pub ftp-osl.osuosl.org
ssh_exchange_identification: Connection closed by remote host
can someone copy leto.net/tmp/parrot-1.7.0.tar.gz to ftp-osl.osuosl.org ? 01:34
that is the release
Whiteknight ssh -l parrot ftp-osl.osuosl.org 01:35
no -i stuff
(at least, that's how it just worked for me)
dukeleto $ ssh -l parrot ftp-osl.osuosl.org 01:36
ssh_exchange_identification: Connection closed by remote host
this worked before and now it doesn't
before, as in when allison and I set it up
Austin Bah. Does anybody still use ftp anymore? 01:37
dukeleto i have to leave my current position soon, don't know when I wil be able to continue
the release announcement it on parrot.org, just needs to be published and the symlinks for the current release need to be updated 01:38
Whiteknight dukeleto: start putting out he release annoucements. I'll do the ftp
dukeleto Whiteknight: the release announcement links to the FTP site, so should we sync it up? 01:39
Whiteknight: or should I just do it and let the link be broken until you upload?
kinda don't like that 01:40
Austin Slashdot fodder. 01:41
Whiteknight just do it, it takes time for the FTP to sync anyway to the mirrors
just do it
Austin "Ooh, the parrot guys announced a release, but the link was broken for eight whole minutes after the email went out."
+4, snarky
Whiteknight tarball uploaded 01:44
done
SHIP IT
dukeleto OK
www.parrot.org/ <--- COOL STUFF 01:45
be back in a few
Austin I'll try paying my taxes right now. 01:49
dukeleto: I don't see anything on www.parrot.org.
Latest news is 1.6, front page is 1.6 and Svn vs. Git
Whiteknight I'm going to bed now. Goodnight 01:53
Austin Goodnight, Whiteknight. 01:54
Umm, I'm getting a build error. 01:57
Parrot::OpsFile: Unrecognized line: 'inline op end() :base_core :check_event :flow { 01:59
02:03 szabgab joined
Coke Austin: did you reboot^Wrealclean?\\ 02:06
Austin I checked out into a new dir
svn co ...trunk newparrot
Coke parrotvm is parrotvm.osuosl.org
Austin perl Config, make test, error. 02:07
Coke 1.7.0? trunk?
... guess that's the same thing for a little while.
Austin Trunk.Which is 1.7.0
41964 02:08
Coke running a build now. what platform?
Austin austin@andLinux:~/mydocs/Sources/newparrot$ perl ./tools/build/ops2pm.pl src/ops/core.ops
It's xubuntu on andlinux kernel. 02:09
The command above, run by hand, gets the error.
02:10 kid51 joined
kid51 Hmm: Do I correctly infer from the Topic that we haven't released yet? 02:12
Ah, well, VERSION reads 1.7.0. So something's happening. 02:14
Austin kid: Duke had to relocate, I think. 02:15
kid51 thx
Coke Austin: no error here.
Austin Okay.
Coke build is fine on amd64/linux.
Austin I'm pretty sure this isn't a platform thing. ops2pm.pl ?
Coke Austin: silly question: any local mods? 02:16
kid51 some problem with that .pl?
Coke what version of perl?
Austin Nope. Just checkout, configure, make.
This is perl, v5.10.0 built for i486-linux-gnu-thread-multi
Coke /4/86?
huh.
in any case, WFM. Sorry 02:17
dalek website: dukeleto++ | Parrot 1.7.0, "African Grey" Released!
website: www.parrot.org/news/2009/Parrot-1.7.0
kid51 Yeaaaaaay!
PerlJam dukeleto++ 02:18
Austin Ruh roh. Where is Parrot/OpLib? 02:19
Coke lib/Parrot/OpLib/
02:20 cconstantine joined
Austin Not here. 02:20
dukeleto anything on fire?
moderator Parrot 1.7.0 "African Grey" is out! Feel free to nuke trunk | Improve PCC testing this week or the release gets it! 02:21
Austin I don't have a $/lib/Parrot/OpLib 02:21
Coke ah. Austin, did you grab from ftp or svn?
Austin svn 02:22
Coke ok.
that's a generated file.
did you run with make -j ?
(wondering if we lost a build dep)
02:22 Zak joined
Austin just make 02:22
But I do have an installed parrot.
Could it be installed someplace?
Coke I can now duplicate your problem. 02:23
Austin Oh, boy.
And how did you do that?
Coke er, the missing OpLib, that is.
grabbed a fresh copy via ftp. doing a build...
er, no, there it is. hurm. 02:25
Coke does a realclean and starts over.
ah, it's not created by configure, but during the build. no worries, build proceeding fine. 02:26
so, if you're missing that, you failed an earlier step.
Austin Yeah, it looks like I failed the "build OpLib/core.pm" step.
Coke I would do a realclean and log the build.
Austin Which involves ops2pm.pl.
Coke or, if you have the error from that...
dukeleto Austin: you gave my heart a little jump :) 02:28
Austin dukeleto: So far, I've checked out, run configure, run make, and crashed and burned. Keep jumping.
Coke dukeleto: do you have an account on feather? 02:30
dukeleto Coke: yes 02:32
Coke we can share a screen session if you don't yet have an account on parrotvm. 02:33
(I can show you how I do the docs stuff.)
(I just downloaded and built/tested 1.7.0 via ftp.) 02:34
dukeleto Austin: nopaste?
purl i think nopaste is at nopaste.snit.ch/ (ask TonyC for new channels) or poundperl.pastebin.com/ or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/ or paste.scsys.co.uk (for #catalyst, #dbix-class, #moose and others) or gist.github.com/ or paste or gtfo or tools/dev/nopaste.pl or trac.parrot.org/parrot/browser/tru...nopaste.pl
dukeleto Austin: platform?
02:35 janus joined
Austin making distclean 02:36
Platform is ubuntu / andlinux
nopaste "Austin" at 98.235.55.43 pasted "make fails after make distclean, configure, make" (17 lines) at nopaste.snit.ch/18390 02:37
dukeleto Coke: sure, we can try that 02:38
Austin: i got no clue. does the tarball work for you?
Coke: i think i am going out to dinner soon, but I might be sitting on a train tonight on the way to the hotel, so I will have time later. It might be best if you do the docs stuff and record everything 02:39
Austin Okay. It's a unix/win problem. 02:40
Coke line endings?
purl line endings are not different between linux distributions, no.
Austin yep
Coke don't we have "native" set on everything? =-)
Austin Underlying windows filesystem.
svn on windows did the co.
Need to set unix mode, or something. 02:41
Coke ... don't do that. =-)
Austin Can't help it. Ubuntu's svn dist is hella old.
02:41 eternaleye joined
Coke I can no longer ssh in to osuosl, apparently. 02:42
02:42 eternaleye joined
Coke ... at least from feather. wtf. 02:43
02:43 eternaleye joined
Coke *sigh* ID10T 02:43
msg dukeleto docs.parrot.org updated 02:45
purl Message for dukeleto stored.
Austin okay. distcleaned, now let's try again. 02:46
jdv79 searches for beer 02:47
Austin well, that works better.
Austin gives jdv79 a liter mug!
Coke no reason not to use / in a makefile for windows, isser?
Austin Nope. 02:48
It's used all over the place, seemingly without problems.
Whoo. lots of little perl scripts, marching ant-like up the window... 02:49
Much better.
dukeleto Coke++
Austin What's this? A warning? There's a warning in the source code! Who's responsible for this? 02:50
Also, a compile error. :( 02:54
nopaste "dukeleto" at 24.4.68.112 pasted "tools/util/crow.pir doesn't work due to json being renamed" (5 lines) at nopaste.snit.ch/18391 02:55
02:56 JimmyZ joined 03:01 jsut_ joined
Austin Man, they still haven't fixed the symlink problem. 03:08
:(
Coke FAILED SMOLDER: smolder.plusthree.com/app/public_pr...port/29243 03:10
shorten Coke's url is at xrl.us/bfte7y
Coke FAILED SMOLDER:
smolder.plusthree.com/app/public_pr...port/29242
shorten Coke's url is at xrl.us/bfte8c
Austin Coke: /4/86?? 03:11
:)
plobsing ping kid51 03:12
purl I can't find kid51 in the DNS.
Coke Austin: those ain't mine. 03:16
win 4
Austin You're right. paco@javier, whoever that is. But he's running linux... 03:18
Hmm.. 03:19
make smoke requires Tap::Harness::Archive. We should probably ship that.
(If we want people making smoke.)
Coke no. they can get it on their own. 03:22
(should be in the bundle)
Austin What bundle? 03:30
purl well, bundle is just a directory. or a special kind of CPAN thingee that just says "install these other modules"
Coke cpan's parrot bundle.
(we don't want to bundle CPAN modules that aren't required to build) 03:31
03:37 zak_ joined 03:57 Zak joined
dalek kudo: db1e525 | duff++ | build/PARROT_REVISION:
Bump PARROT_REVISION for release
 04:02
shorten dalek's url is at xrl.us/bftf7p
04:03 zak_ joined
dalek rtcl: c725c62 | coke++ | config/makefiles/root.in:
[build] cleanup makefile, remove cruft & duplication
 04:22
shorten dalek's url is at xrl.us/bftggf
04:25 zak_ joined
dukeleto 'ello 04:27
Coke ~~ 04:28
tools/dev/pprof2cg.pl isn't installed 04:36
msg cotto I have a case where the profiler is reporting on subs in the wrong files again. 04:42
purl Message for cotto stored.
Coke msg cotto parrot -Rprofiling tcl.pbc -e "proc a {} {return [expr 3+2]}; puts [a]" 04:43
purl Message for cotto stored.
Coke msg cotto this eventually results in: 315,921,655 src/mathops.pir:tcl;&proc []
purl Message for cotto stored.
Coke msg (&proc is defined in runtime/builtins/proc.pir)
purl Sorry, I've never seen (&proc before.
Coke msg cotto (&proc is defined in runtime/builtins/proc.pir)
purl Message for cotto stored.
cotto hi 04:44
purl> You have 10 messages waiting.
purl cotto: excuse me?
Coke hio. The entire output has mistakes there.
this is with parrot latest and partcl latest.
I can find at least 20 lines mentioning files where the sub listed isn't. 04:45
cotto I'd ask for a minimal case, but you'd probably want to hurt me.
but thanks. 04:46
Coke I gave you the minimal case. =-)
I suspect includes or tailcalls.
cotto the tcl runtime is not a minimal case ;)
Coke, can you make that into a tt? It'll take some digging and I won't have time tonight. 04:47
(and assign to me)
Coke trac sloooow 04:49
Patterner fire pretty 04:51
dalek TT #1126 created by coke++: tools/dev/pprof2cg.pl not installed. 04:57
pmichaud use.perl.org/~pmichaud/journal/39779 # new article about rakudo, nqp-rx, etc. 04:59
dalek TT #1127 created by coke++: profiling core shows incorrect filenames. 05:00
tracwiki: v110 | dukeleto++ | WikiStart 05:02
tracwiki: trac.parrot.org/parrot/wiki/WikiSt...ction=diff
shorten dalek's url is at xrl.us/bftgwn
dalek tracwiki: v111 | dukeleto++ | WikiStart 05:05
tracwiki: trac.parrot.org/parrot/wiki/WikiSt...ction=diff
shorten dalek's url is at xrl.us/bftgx6
dalek rtcl: 779deea | coke++ | (2 files):
use short git revs for readability.
 05:13
shorten dalek's url is at xrl.us/bftg5o
05:27 GeJ joined 05:36 chromatic joined
dukeleto 'ello 05:43
dukeleto is on Caltrain. And on IRC. /me loves tethering 05:44
cotto and now 05:50
sleep
06:03 uniejo joined
dukeleto languages? 06:07
purl languages are tools. just like php programmers.
dukeleto parrot languages?
purl, languages is also trac.parrot.org/parrot/wiki/Languages 06:08
purl OK, dukeleto.
dukeleto shall I include our release quote in the email announcement? I think I shall. 06:11
dukeleto feels like he is talking to himself 06:13
purl, be useful
purl whirrs and clicks in the background.
dalek p-rx: 3de213c | pmichaud++ | src/ (2 files):
Handle :w (quote words) quote modifier .
 06:26
shorten dalek's url is at xrl.us/bfth84
dalek p-rx: 5b85aa3 | pmichaud++ | src/ (4 files):
Simple test of subcalls in NQP.
shorten dalek's url is at xrl.us/bfth86
06:35 fperrad joined
mj41 Hi. TapTinder is offline. Moving servers to new server room. Back probably later today or tomorrow. 07:06
07:17 Zak joined
he_ Did a build and smoke-test on NetBSD/sparc64 5.0, first failure at smolder.plusthree.com/app/public_pr...st_failure 07:17
shorten he_'s url is at xrl.us/bftiqz
07:18 payload joined
he_ Unsurprisingly it's t/pmc/threads.t which fails there as well. 07:18
The failure mode for test 14 is different, though: here it's a "Null PMC access".
07:21 zak_ joined 07:24 donaldh joined 07:26 masak joined 07:42 Zak joined 08:31 bacek joined
bacek o hai 08:56
08:58 zak_ joined
JimmyZ ;) 08:59
bacek dukeleto++ # great job with release
moritz so, has the pcc branch been merged already? 09:00
JimmyZ hadn't
moritz thanks
bacek moritz: I can merge it in few minutes. But would like to hear from allison before doing it. 09:04
09:05 riffraff joined 09:06 zak_ joined 09:16 patspam joined 09:22 Zak joined 09:32 zak_ joined 09:46 payload joined 10:51 kid51 joined 11:00 Zak joined 11:02 cconstantine joined 11:37 cconstantine joined
dalek rrot: r41965 | fperrad++ | trunk/PLATFORMS:
update PLATFORMS
 11:38
moritz running rakudo in valgrind shows some "Use of uninitialised value of size 8" in various mark functions 11:46
is that normal?
nopaste "moritz" at 132.187.31.74 pasted "valgrind output rakudo on parrot-1.7.0" (55 lines) at nopaste.snit.ch/18395
moritz optimized parrot, and most of these messages stem from the startup/compilation, not from running the tests 11:49
jonathan The one in Parrot_Context_mark is worthy of consideration. 11:53
I fixed a segfault a week or two ago that was due to us not init'ing something that got used in there.
dalek kudo: 4e5877c | jonathan++ | src/pmc/perl6multisub.pmc:
Make trait dispatch that depends on a required named parameter a bit more efficient by allowing candidates that could never apply to fail much faster.
 12:00
shorten dalek's url is at xrl.us/bftmd2
12:14 whiteknight joined
whiteknight good morning #parrot 12:15
plobsing hi, whiteknight. 12:17
whiteknight good morning plobsing 12:18
plobsing sorry for not explaining what I wanted tested yesterday
moritz whiteknight: if you happen to fancy some memory debugging: irclog.perlgeek.de/parrot/2009-10-21#i_1626883
I'm doing another valgrind run now, logging all the output this time :-) 12:19
12:19 nopaste joined
plobsing I'm looking for anyone with libjit to apply my last patch on TT1105 against the auto_libjit branch and make test 12:19
whiteknight plobsing: I wanted to do that last night but got busy and ran out of time
will try for tonight
plobsing ok thanks. 12:21
12:39 patspam joined
whiteknight moritz: yes, I think those are normal warnings to get 12:53
moritz whiteknight: ok
whiteknight those functions are tracing the stack, so they are reading uninitialized values from the stack
it's ugly and I want it to disappear, but it's normal
moritz ok
12:53 buildbot joined 12:54 payload joined
moritz colomon on #perl6 has some more interesting traces, will post soon 12:54
whiteknight okay, awesome 12:55
12:57 colomon joined 12:58 colomon joined
colomon www.harmonyware.com/perl/cotanh.t.valgrind.log 12:58
whiteknight okay, I'll look at that 13:00
so pcc_reapply didn't land last night 13:01
colomon, moritz: trace_system_areas and trace_system stack are GC routines to walk the stack. This could produce these "Use of uninitialized value of size 8" warnings you are seeing 13:02
so anything with those function in the backtrace is a false positive
(obnoxious, but harmless)
the way to fix that would be to not allow unanchored PMCs on the C stack 13:03
big project, but not undoable 13:04
the "Conditional jump or move depends on uninitialized value(s)" warnings are much more worrisome
looks like that's a problem in compilers/imcc/parser_util.c:INS() that we could resolve pretty quickly 13:05
dalek kudo: 4c248e5 | masak++ | docs/release_guide.pod:
[docs/release_guide.pod] s/-/_/ in file name
 13:10
shorten dalek's url is at xrl.us/bftmk5
whiteknight in fact, if somebody opens a ticket for that issue and assigns it to me, I'll fix it tonight 13:11
masak src/string/encoding/ucs2.c: In function 'ucs2_decode_and_advance' src/string/encoding/ucs2.c:609: warning: no return statement in function returning non-void
is this something that should be looked at?
whiteknight (can't do it now, work computer doesn't have yacc/lex)
moritz whiteknight: ok, I'll open a ticket for you
whiteknight masak: probably, yes
masak: I'll take a quick look now
masak whiteknight++ 13:12
whiteknight masak: fixed and committed 13:14
masak \\o/
moritz tt #1128 it is 13:15
whiteknight w00t 13:17
dalek rrot: r41966 | whiteknight++ | trunk/src/string/encoding/ucs2.c:
[string] add an unused return statement to shut up warning messages people are getting from code analyzers.
TT #1128 created by moritz++: Possible memory error in compilers/imcc/parser_util.c:INS() 13:19
whiteknight these files in src/encodings/* need a lot of cleanup
I would really like to turn off stack tracing. I think it does more harm then good 13:25
especially since it's so in accurate
whiteknight doesn't know why he added a space there 13:26
13:37 allison joined
nopaste "he" at 158.38.152.119 pasted "Fix to do correctly do rpath for C language tests" (13 lines) at nopaste.snit.ch/18398 13:40
dukeleto 'ello 13:41
he Without that I would get "could not find libparrot.so.1.7.0" from the tests. 13:42
dukeleto is still going to do more release anouncements, after sufficient coffee 13:43
moritz perlmonks! 13:44
purl perlmonks is, like, perlmonks.org, an active community of knowledgeable Perl hackers
he I now have a pkgsrc package for 1.7.0 which installs clenly, and tests cleanly sans test 14 in t/pmc/threads.t.
...with shared libary enabled for NetBSD too.
Now to back-port to the 1.6.0 package people were complaining about...
moritz he++ 13:45
whiteknight dukeleto++
for our mantra of "releases should be small and easy", they never turn out to be particularly easy 13:46
dukeleto whiteknight: yeah, some bumps, but everyone pitched in and I really appreciate that
everybody++
whiteknight we're a team, it's a team effort
plus, I think we all feel guilty for neglecting trunk all month 13:47
he BTW, I got a different failure for t/pmc/threads.t test 14 on NetBSD/5.0 sparc64, there I got a "Null PMC access"
dukeleto we definitely forgot about win32 on trunk. that was slightly painful
he Ref. smolder.plusthree.com/app/public_pr...st_failure 13:48
shorten he's url is at xrl.us/bftiqz
he I'm beginning to have my doubts whether the test is correct...
13:49 ruoso joined
dukeleto he: that test has had lots of issues on many platforms 13:50
he: so that is very possible 13:51
Coke wasn't that test altered in pcc_reapply?
dukeleto Coke: yes, i think so
allison speaking of which, is it safe to merge now?
Coke allison: aside from breaking partcl... 13:52
moritz as safe as can be, I think :-)
Coke: that's why we do it after the release, not before :-)
allison Coke: what are the remaining issues for partcl?
Coke allison: same one as yesterday. the updated "invoke a method that returns an int from c" segfaults. 13:53
(and the old one has been removed.)
dukeleto Coke: "feel free to nuke trunk"
Coke chromatic had poked yesterday, and we had agreed that we needed a segfaulty test /in parrot/. 13:54
allison Coke: I can look into that today, It's a small one, so I'll go ahead and merge, we'll fix it quickly
Coke: yes, exactly
particle1 did 1.7.0 happen? 13:55
dukeleto particle: yes, but I am still doing announcements
particle aha!
dukeleto particle: about to send out an email to a bunch of mailing lists
particle dukeleto++
dukeleto particle: i updated wikipedia, c2, etc 13:56
particle it's probably best that those are done first, to make sure they're done at all.
Coke en.wikipedia.org/wiki/Parrot_Foundation doesn't list the full board, fwiw. 13:58
(though I don't think we did a press release or anything with the election results.)
allison Coke: does it need to be updated with Patrick?
Coke yes 13:59
allison: www.perlfoundation.org/who_s_who is out of date. (ovid's in the wrong place.) 14:01
allison Coke: that looks right, he's an untitled director like me 14:02
Coke: but that has me wondering... ah, yes, I did update the list of directors on www.parrot.org/foundation 14:03
Coke allison: whoops. I saw him only under the GC/Former heading. my bad.
Coke wonders if it would grossly inappropriate for him to add his own wikipedia page to make the link from the pafo page work. =-) 14:04
allison Coke: unfortunately, yes, it's strictly against policy
Coke: but some other kind person can add it for you :) 14:05
Coke: (that's always seemed daft to me, that the person who potentially knows the most is the one absolutely not allowed to write the entry)
Coke: would you like to be added as an owner on the parrot-dev google group? 14:07
Coke allison: sure. I mainly asked so I could see the settings when creating parrot-users.
allison Coke: makes sense
Coke last 4/5 smolder reports? FAIL 14:08
allison Coke: I figure if at least two of us have access to the groups, it ups the bus factor
Coke ayup
i'll return teh favor setting up the other one.
allison I think you have to join as a member before I can add you as an owner 14:09
dukeleto release email just went out
allison Coke: all three are Win32 14:10
Coke: all four, that is
Coke: are those the Win32 failures dukeleto and whiteknight were debugging? 14:12
dukeleto to my knowledge, parrot has no twitter/identi.ca presence. would anybody like for me to change that? 14:13
moritz dukeleto++ # release announce poem 14:14
allison dukeleto: I can see some benefit, but if we aren't posting to it regularly it makes more sense to just post from our own twitter/identi.ca accounts with a parrot tag
dukeleto allison: i can fix that. i will set it up to cotweet, so anybody on the parrot team can post to the parrot account 14:15
allison: cotweet.com
allison dukeleto: looks like it's already taken twitter.com/parrot
dukeleto identi.ca/parrot is still available
i was going to use @parrotvm on twitter
sounds good to whoever is paying attention right now? 14:16
and I just got a fail whale trying to make a new twitter acct
moritz +1 from me
allison dukeleto: can't see any harm
dukeleto allison: sweet. in process 14:17
i can use ping.fm to broadcast to the identi.ca and twitter acct in one fell swoop 14:18
Coke for bonus points, write the poster in PIR. 14:19
... but that can wait. =-0 14:20
first, fix mysegfaults1
dukeleto Coke: now you are talking. that is definitely a good project
Coke: if your segfaults were segfaults 2.0, then I would hack on it
moritz write it in some cross-HLLs
in partcl, with rakudo's IO::Socket::INET
dukeleto is there a way to get an email alias @parrot.org ? 14:21
dukeleto wishes for dukeleto@parrot.org -> jonathan@leto.net 14:22
allison dukeleto: we plan to set those up for members, but haven't yet 14:23
dukeleto allison: no big deal. 14:24
purl Yes it is! I hate you I hate you
dukeleto notes that Parrot is now all web-2.0-ified
purl: I hate you too 14:25
purl dukeleto: excuse me?
dukeleto purl, I hate you is doin' my best to annoy you
purl ...but i hate you is <reply>I'd kick purl's ass but jdube's foot would get stuck...
dukeleto purl, I hate you is also doin' my best to annoy you 14:26
purl okay, dukeleto.
dukeleto i need a smallish parrot logo to use as an avatar 14:31
moritz dukeleto: the favicon from parrot.org? 14:32
allison dukeleto: I can give you a slightly bigger version of the favicon if you need it 14:33
(I don't remember what resolution it was)
moritz or you could downscale the parrot head from the parrot.org front page
dukeleto allison: no worries, I am using mypictr.com to crop the parrot.org logo 14:34
NotFound We need a name, calling it "the parrot logo" is boring. 14:35
allison NotFound: a name for the Parrot? 14:36
like Harvey?
NotFound Yeah
dukeleto Harvey The Virtual Parrot 14:37
Coke -1
dukeleto we could have a "name our mascot" contest. or just call him Bob
purl -1
Coke -1 14:38
purl -1
NotFound Nooooooo. Anythingh but Bob.
This is Bob: en.wikipedia.org/wiki/Microsoft_Bob 14:39
Coke NotFound: exactly.
moritz Harvey > Bob
allison Barvey
14:40 soxet joined
Coke I would prefer to take my chances with Iago. 14:40
allison Coke: that's actually a good name
Coke Though I prefer Sketch.
moritz
.oO( parrotshedding )
 14:41
dukeleto Coke: Norwegian Blue?
allison e's pining for the fjords?
Coke allison: I fear the lawyers.
en.wikipedia.org/wiki/Dead_Parrot can be trolled for more thoughts.
allison Coke: yes
dukeleto wow, p5p's release is synced up pretty close to ours. 14:42
Coke dukeleto: for the 5.11 track?
dukeleto parrot, perl5 and rakudo all get released the same week each month. madness! 14:43
Coke: yes, 5.11.1 came out yesterday
moritz and we're going to have a book release too
dukeleto i am noticing that p5p publishes md5's for tarballs. we should probably do that....
Coke dukeleto: I leave the exercise of determining how far away the 20th is from the 3rd tuesday to the pedants.
(how far away it can get, that is.)
moritz: book? 14:44
purl somebody said book was blog.jrock.us/articles/Catalyst%20Book.html or Philippe "BooK" Bruhat, aka Mr Pink or mailto:book@mongueurs.net or awesome or a dino or a black hat or a galcon player
moritz Coke: Perl 6 book. Will be announced on Friday
dukeleto i never knew about MS Bob. scary. 14:45
Coke dukeleto: dear god, man.
dukeleto moritz: nice!
NotFound dukeleto: I keep a bookmark with that and Apple Newton, to show them to people that says that Microsoft/Apple never failed a product. 14:46
dukeleto NotFound: nice
Coke as the local apple apologist, the newton was pretty cool. =-)
dukeleto twitter.com/parrotvm
identi.ca/parrot 14:47
NotFound Coke: for those not convinced by Newton, I also have the Lisa and the Apple III X-)
Coke NotFound: you win! =-)
NotFound Is nice when people says "I've never heard that" and tell "That's the proof" :) 14:50
Coke does wish he could do more parrot development on his os x box. ah wells. 14:52
dukeleto Coke: the 10.4 box? 14:53
14:55 Psyche^ joined 14:56 Andy joined
Coke dukeleto: yah. 14:57
(lack of install means no partcl dev.)
of course, the box itself isn't alive today. (can anyone ping coleda.mine.nu or tracert to it?)
(feather is blocking tracert!) 14:58
dukeleto Coke: it doesn't respond to pings
Coke: : it=coleda.mine.nu
Coke what about tr? can you get close? =-) 14:59
(oh, tracert vs. traceroute. windows has spoiled me.) 15:00
moritz last hop is gig0-2.albynynsa-ubr01.nyroc.rr.com
15:00 whiteknight_ joined, theory joined
moritz no idea if that's close or not 15:01
15:01 payload joined
Coke think so. I'm in albany, that's rochester. 15:04
bah. digital phone is crapped out too. :| 15:05
15:47 mokurai joined
NotFound Where can .annotate be placed to annotate the "too many arguments passed" point? 15:50
Before .sub is not accepted, before .param the .param is not recognized, after params has no effect. 15:51
dalek rrot: r41967 | allison++ | trunk/docs/pdds/pdd03_calling_conventions.pod:
[cage] Some usage cleanup on "arguments" versus "parameters".
 15:54
15:56 payload joined 15:59 theory joined 16:00 dalek joined 16:03 ttbot joined 16:10 Zak joined, darbelo joined
darbelo he_: ping 16:28
trac slooooooow 16:33
16:35 PerlPilot joined 16:38 nbrown joined 16:42 whiteknight joined, dukeleto joined
dukeleto feather is down 16:42
darbelo Are we getting slasdotted or something? 16:45
PerlPilot I think it's more that just feather. There are several sites that I was suddenly unable to access. 16:46
perlpilot If I could run a traceroute from here, I would.
(I could run it, but the results would be useless due to the nature of the network I'm on) 16:47
whiteknight not slashdotted
he darbelo: here.
darbelo I applied your rpath patch, do you have any other patches to make NetBSD happy? 16:48
Tene I remember hearing that we now work on openbsd, too?
darbelo Tene: Yes. I'm on OpenBSD. 16:49
he Hm, ok. Which rpath patch? I'll nopaste my current set; the patch I had to config/inter/libparrot.pm was wrong -- the resulting executables which were installed still had rpath entries pointing to the build dir (bad)
nopaste "he" at 158.38.152.80 pasted "Current NetBSD patch set" (70 lines) at nopaste.snit.ch/18403
darbelo he: The one for lib/Parrot/Test.pm
he Ah, ok. I think I made changes to it, you can look at that nopaste. 16:50
darbelo Looks the same as what I committed.
he OK, that's good then -- thanks! 16:51
allison here it comes...
16:52 cotto joined
darbelo he: all three patches committed. 16:54
ttbot Parrot trunk/ r41972 i386-linux-thread-multi make error tt.ro.vutbr.cz/file/cmdout/116794.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 16:58
16:58 payload joined
darbelo r41972? I don't think that was me. 16:58
Oh. pcc_reapply merge. 16:59
allison looks like it's missing the changes to build src/call/args.str 17:01
ah, no, the file itself wasn't merged 17:02
(didn't get added by the diff)
whiteknight feather is down and Coke can't get to IRC without it
17:04 nbrown_ joined
allison ah, will revert his fix 17:05
darbelo Odd. it built for me.
allison nope, he beat me too it :) 17:06
to
ttbot Parrot trunk/ r41973 i386-linux-thread-multi make error tt.ro.vutbr.cz/file/cmdout/116846.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
allison waiting for ttbot to reach r41974... 17:09
17:10 nbrown joined
Tene lolcode now builds and passes 'make test' 17:12
whiteknight no dalek?
oh, dalek would be on feather
darbelo Tene++ 17:13
linkk?
ttbot Parrot trunk/ r41974 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/116884.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
darbelo eh, link?
Different platform broken. Sounds like progress. 17:14
purl lolcode?
purl lolcode is HAI CAN HAS STDIO? VISIBLE "HAI WORLD!" KTHXBYE or at lolcode.com/
Tene svn.parrot.org/languages/lolcode/ 17:15
it assumes parrot_config is in your $PATH
darbelo A reasonable assumption. 17:16
Tene and doesn't support 'make install' yet. Needs a moderate amount of config work.
moritz purl: lolcode is also at svn.parrot.org/languages/lolcode/
purl okay, moritz.
cotto_work parallel build continues to be broken 17:17
whiteknight so then don't do the parallel build :( 17:18
ttbot Parrot trunk/ r41975 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/116903.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
darbelo whiteknight: tell that to sun's dmake.
cotto_work darbelo, does dmake automatically do -j? 17:19
darbelo Last time I used it, I think that was the default. 17:20
moritz a make -j 3 just worked for me (linux, GNU make, after the merge)
lots of errors in 'testb'
nopaste "moritz" at 87.176.87.177 pasted "make fulltest failures on r41975" (37 lines) at nopaste.snit.ch/18404 17:21
darbelo Tene: I'm failing the last test in lolcode's t/01-vars.t on latest trunk 17:24
Tene darbelo: yes, known. 17:25
darbelo Ok, then. Ship it! 17:26
whiteknight build problems and test failures? Better then I expected 17:35
cotto_work moritz, did you try realclean? The makefile isn't smart about dynpmcs. 17:36
s/realclean/reconfig/
moritz cotto_work: yes
now trying after git-clean -xfd
cotto_work I'm only seeing one failure in t/pmc/threads.t 17:37
It's too bad that make fulltest doesn't say which test* target it was running when a test failed. 17:40
moritz ok, after thorough cleaning I also only get one failure in t/pmc/threads.t (testr) 17:42
dukeleto cotto_work: patches welcome ;)
cotto_work dukeleto, I welcome them too. 17:44
dukeleto cotto_work: I patch, you patch, we all patch for welcoming patches! 17:45
17:46 nbrown joined
cotto_work dukeleto, I don't patch. cotto patches. 17:46
important distinction
dukeleto cotto_work: duly noted
moritz for running dalek, do I need anything else than cloning the github repo and starting a script? 17:47
darbelo doesn't think so.
Well, you need botnix, but I'm assuming you know that. 17:48
moritz I did not 17:50
and then I need to connect it to the plugins... 17:51
sounds like work :/
NotFound I've fixed a nasty annotations bug. Where are the bots when you need them?
moritz I'll just wait for dalek to show up again
NotFound++ # he needs the karma, it seems ;-)
NotFound More than a week witout any, I was getting an abstinence sindrome. 17:52
cotto_work NotFound, did you add a test for it? 17:53
NotFound cotto_work: I can't figure how to trigger it in a way catchable from pir 17:54
ttbot Parrot trunk/ r41976 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117021.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
cotto_work NotFound, would a parrot debugger test catch it? 17:57
NotFound cotto_work: I can try a perl test, but supposedly we are trying to get rid of them. 17:58
cotto_work NotFound, yes, but we'll need a way to run those kinds of tests anyway.
NotFound Did we have some way in Test/More to run a pir program in other parrot instance? 17:59
cotto_work dukeleto: ^?
darbelo Having a perl test is better than no test. 18:00
dukeleto NotFound: what are you trying to test? 18:01
NotFound dukeleto: the backtrace of an unhandled exception 18:02
dukeleto NotFound: hmm. not quite sure you can easily to that from test_more.pir, yet 18:03
NotFound The code annotations where not properly showed in some cases before my last patch
Some name convention for a perl test companion for a pir one? 18:04
18:05 bacek joined
darbelo NotFound: Feel free to create one. 18:06
ttbot Parrot trunk/ r41977 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117055.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 18:08
18:10 chromatic joined 18:11 perlpilot joined 18:13 hercynium joined
dukeleto NotFound: test-old.t 18:13
NotFound: we have a convention that perl tests are in test-old.t and are removed from there and put in test.t as they are converted 18:14
cotto_work as a bonus, you'll win at irony
18:14 joeri joined
allison This is a weird windows error "nci_test.obj : error LNK2001: unresolved external symbol _PMCNULL" 18:14
dukeleto NotFound: the -old prefix makes sense usually, since we aren't supposed to be adding perl tests, but sometimes you can't help it
NotFound allison: PMCNULL must not be external. 18:15
Someone must be abusing internals on an external program- 18:16
allison NotFound: not in nci_test.c or nci.c... the error message doesn't give any other indication where it might be 18:18
18:19 dalek joined
dalek rrot-plumage: fe869d6 | japhb++ | :
[plumage] Fix variable name in configure_nqp_configure()
 18:20
shorten dalek's url is at xrl.us/bftnx8
darbelo dalek's back! 18:21
dalek TT #1129 created by coke++: segfault in clone_key_arg (due to pcc refactor)
rrot: r41969 | darbelo++ | trunk/lib/Parrot/Test.pm:
Use rpath_blib, when defined, to get at the build libraries in Test.pm. This fixes the "could not find libparrot.so.1.7.0" errors in netbsd. Patch by he++
 18:22
18:23 PerlJam joined
rrot: r41970 | darbelo++ | trunk/t/op/trans.t:
un-TODO the "atan part 2" test on NetBSD, it passes now.
18:23 pmichaud joined
18:23 Coke joined
rrot: r41971 | darbelo++ | trunk/config/init/hints/netbsd.pm:
Update the NetBSD hints file, shared libraries will now work there.
rrot: r41972 | allison++ | trunk (75 files):
[pcc] Merging the pcc_reapply branch into trunk. Reworks the internals of the

arguments and return values.
18:23 dukeleto_ joined
rrot: r41973 | coke++ | trunk/MANIFEST:
[build] fix MANIFEST after branch mergeback
rrot: r41974 | allison++ | trunk/src/call/args.c:
[pcc] Add file that didn't make it into the merge.
rrot: r41975 | coke++ | trunk/MANIFEST:
[build] re-fix manifest now that the file is back.
rrot: r41976 | NotFound++ | trunk/src/debug.c:
Fix a bug in showing annotations when backtracing an uncatched exception
rrot: r41977 | allison++ | trunk/src/call/args.c:
[pcc] Set file metadata on added file from branch.
NotFound Someone has dropped the external version of PMC_IS_NULL??? 18:24
Coke whiteknight++ # feather status updates
dalek p-rx: 6cd6c1a | pmichaud++ | src/HLL/Grammar.pm:
Add default <ws> rule.
 18:25
shorten dalek's url is at xrl.us/bftny7
dalek p-rx: 9924922 | pmichaud++ | src/NQP/ (2 files):
Reorganize code a bit, initial term:sym<identifier> (subcall) implementation.
shorten dalek's url is at xrl.us/bftny9
dalek p-rx: b079e0d | pmichaud++ | src/NQP/ (2 files):
Add basic statements and terminators.
shorten dalek's url is at xrl.us/bftnzb
dalek p-rx: a4bfb16 | pmichaud++ | build/Makefile.in:
Refactor Makefile in preparation for nqp tests.
shorten dalek's url is at xrl.us/bftnzd
dalek p-rx: 4ed10a8 | pmichaud++ | (5 files):
Add goal-matching syntax to P6Regex.
shorten dalek's url is at xrl.us/bftnzf
dalek TT #495 closed by darbelo++: [TODO] remove hardcoded path from installable pbc_to_exe
18:28 Util joined 18:33 jonathan joined
dalek rrot: r41978 | chromatic++ | trunk/src/runcore/profiling.c:
[runcore] Fixed a compiler warning in runops_profiling_core(). Casts, bah.
 18:36
Coke the profiler does seem much faster now.
cotto_work I didn't know about that function.
Coke (it's a shame it's not working for tcl.) 18:37
cotto_work Coke, glad to hear it.
Coke you had to wait for the backhand! =-)
could it be related to the optimized build? 18:38
cotto_work possibly, but I also added line number caching to hit Parrot_Sub_get_line_from_pc less often. 18:39
--optimize isn't likely to hurt
Coke would that cause wrong line #'s ?
er..
chromatic I'm sure that helps, but replacing an O(n) operation for every op, where n is the size of the current PackFile segment, with an O(1) won't hurt.
Coke wrong file names, I mean.
cotto_work Coke, not likely. The profiling runcore just calls a function to get the current filename, so there's not too much it could do to mess that up 18:41
.
whiteknight ETOOMUCHDALEK
cotto_work Coke, does the tcl profiling bug still work after the merge or should I test with a pre-merge parrot? 18:42
(whenever I find the time, that is)
Coke pre-merge.
partcl is broken against trunk now.
whiteknight it's not "broken", it's a feature 18:43
partcl is feature against trunk now
18:43 Zak joined
Coke it's broken. 18:43
cotto_work ok. I'll try to pay enough attention so that I know when it works again.
bacek Good morning
allison Coke: can you point me at the C in Tcl that segfaults?
Coke allison: it's in the ticket. 18:44
cotto_work hi bacek
Coke moment.
allison Coke: ticket #?
Coke moment.
#1129
allison: (reminder, partcl is now on github) 18:45
ttbot Parrot trunk/ r41978 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117121.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
dalek rrot: r41979 | bacek++ | trunk (3 files):
[cage] Fix typo in Parrot_pcc_set_handLers
 18:46
ttbot Parrot trunk/ r41979 i386-linux-thread-multi make error tt.ro.vutbr.cz/file/cmdout/117134.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
japhb Tene++ #lolcode work 18:47
allison Coke: if the 'method' variable is the method object, and the routine takes no arguments, what is 'get_bool'? 18:48
Coke the name of the method. 18:49
which is required by the method sig.
(and then ignored, IIRC. double checking.)
yup. 18:51
allison Coke: you know what, we're missing an option in the invocation functions
Coke: one for invoking methods by method object instead of name
Coke: I was going to suggest Parrot_pcc_invoke_sub_from_c_args, but it doesn't have a place for the invocant 18:52
Coke: give me a second...
dalek rrot: r41980 | bacek++ | trunk/src/call/context.c:
[cage] Fix one more typo
rrot: r41981 | NotFound++ | trunk/t/op/annotate-old.t:
[t] add a test for the annotations problem fixed in r41976
18:53 desertm4x joined
Coke NotFound: the file name in your new test doesn't match the filenam.e 18:54
allison Coke: how does Parrot_invoke strike you as the name for the general function for invoking any Parrot sub or method? 18:55
Coke NotFound: (I'd like to just remove the filename from the POD everywhere, here's a reason to. =-)
allison Coke: is 'invoke' too obscure?
NotFound Just a typo :-X
Coke I would probably bikeshed more like Parrot_call_* so it matched the other call_s
Parrot_call_general? common? 18:56
but I don't care too much. =-)
(even just Parrot_call())
whiteknight Parrot_pcc_invoke_from_sig_object lets you invoke from a sub object and a signature 18:58
and Parrot_pcc_build_sig_object_from_varargs does the rest
ttbot Parrot trunk/ r41980 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117191.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
whiteknight a function to wrap those two would be fine also
allison whiteknight: yes, but that's a terrible interface
whiteknight: exactly 18:59
dalek rrot: r41982 | NotFound++ | trunk/t/op/annotate-old.t:
[cage] fix a typo, Coke++
allison Coke: Parrot_call it is
Coke: Parrot_ext_call, to include the subsystem extension
whiteknight Parrot_call_invoke?
Coke seems redundant. ext_call works. 19:00
whiteknight why the "ext"?
Coke src/extend.c 19:01
whiteknight oh, okay. I didn't realize this was an extension interface
chromatic Don't forget that the API has to take an (optional) invocant. 19:02
whiteknight that could be included with the rest of the args with a "Pi" signature
chromatic That idea doesn't thrill me. 19:04
I don't like it in Perl 5, I really don't like it in Perl 6, and I think it's confusing in Parrot.
However, my best alternative is a C parameter which may or may not be NULL, which isn't much better. 19:05
whiteknight I'm just trying to avoid an explosive increase in the number of external-facing APIs for every little corner case
ttbot Parrot trunk/ r41981 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117219.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
chromatic Yeah, I've rejected that combinatorial explosion already too.
whiteknight We have a lot of method-call APIs in there now that are redundant at best and we're stuck with until a good deprecation point 19:06
chromatic I can talk myself into accepting it, provided that it always must be the first argument, if provided, and that I keep telling myself "You're mapping invocation semantics onto C, which is barely a step above PDP-11 assembly langauge."
whiteknight Two external-facing APIs would be fine: one for subs and one for methods
ttbot Parrot trunk/ r41982 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117227.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
whiteknight one would be better, but I won't push it
chromatic Two isn't bad. 19:07
allison is spending 10 times as long on documenting the function as coding it 19:08
whiteknight we need to start deleting now-obsoleted functions so we can get a clearer picture of what we do have
allison whiteknight: yes, and enter the deprecation notices for the ones to be removed after 2.0 19:09
whiteknight but that's neither here nor there
chromatic We can't forget those, at least without me distracting allison and releasing 2.0.1 with the deprecation notices. 19:11
whiteknight I'll try to do some triage tonight to see what we can cut 19:16
...if tests are passing, of course
whiteknight is trying to get git for windows. This promises to be a learning experience 19:21
Coke I'm running tortoisegit now. 19:22
darbelo pain is a great techer, I'm told.
19:24 davidfetter joined
dalek TT #1130 created by doughera++: dependency problem for libnci_test 19:32
jonathan whiteknight: the msys git works very well for me on Windows, fwiw. 19:34
whiteknight: even git gui works nicely.
whiteknight that's what I'm trying now 19:35
jonathan OK. Hope you have decent experience with it too.
whiteknight I can't seem to find which windows folder corresponds to my bash homedir though
jonathan C:\\Documents and settings\\youruser\\ perhaps
whiteknight tried it, not thre
I created an ssh key somewhere, but can't find it 19:36
jonathan Start...Search... ;-)
whiteknight doing that now
jonathan I'm sure it put mine there.
whiteknight that's what the docs say too
but I look...not there
jonathan Yeah, I have a .ssh directory in there. 19:37
whiteknight ah, nevermind. Found it
ssh-keygen helpfully put it on a network drive
jonathan oh, how nice of it 19:38
:-S
"hi, plz I'd like to share with you my private key"
Coke that's probably where %HOMEDIR% was. 19:39
moritz "my %HOMEDIR%'s not quite my castle" 19:40
dalek kudo: 0117dd3 | pmichaud++ | docs/spectest-progress.csv:
spectest-progress.csv update: 452 files, 27582 (72.3% of 38174) pass, 265 fail

S02-builtin_data_types/capture.rakudo 16 - unflattened captures can be passed to subs S02-lexical-conventions/unicode.rakudo aborted 5 test(s) S03-junctions/autothreading.rakudo aborted 78 test(s) S06-multi/proto.rakudo passed 2 unplanned test(s) S06-signature/defaults.t aborted 2 test(s) S06-signature/optional.rakudo aborted 3 test(s) S10-packages/basic.rakudo aborted 41 test(s) S12-attributes/delegation.rakudo aborted 16 test(s) S12-attributes/instance.rakudo aborted 88 test(s) S12-methods/parallel-dispatch.t aborted 21 test(s) integration/man-or-boy.t aborted 10 test(s)		 19:41
shorten dalek's url is at xrl.us/bftohf
jonathan whoa 19:44
bacek yak... 19:45
cotto_work that's with the new nqp-rx shininess?
moritz no
but with jonathan's new shiny signature binder
cotto_work It's going to be a bumpy month. 19:46
moritz aye 19:49
whiteknight yay! I have git cloning a repo on win32 19:52
bacek purl: (8505330719-7943471284)/8505330719*100 19:53
purl 6.6059681106211
bacek looks good.
darbelo 6% of what?
cotto_work bacek, what's 6.6%?
bacek speedup
darbelo bacek++ 19:54
cotto_work looks like a long-running program 19:55
dalek rrot: r41983 | bacek++ | trunk (5 files):
[cage] Rename Parrot_pcc_constants to Parrot_pcc_get_constants for consistency with other methods
 19:56
rrot: r41984 | bacek++ | trunk (6 files):
[core] Split trivial Context accessor functions into separate file
rrot: r41985 | bacek++ | trunk (4 files):
Move more simple context accessors into context_accessors.c
rrot: r41986 | bacek++ | trunk/include/parrot/context.h:
Move Parrot_Context definition into beginning of context.h
bacek cotto_work: fib.pir
Coke bacek: is NDEBUG defined if you have --optimize but --ccflags=-g ?
bacek: also, shouldn't macros be ALL_CAPS? 19:58
bacek Coke: hmm... I'm not sure about NDEBUG.
Coke allison, any luck with the new invoking function?
bacek Coke: it's very special macros. I'm shortcircuiting sub calls. 19:59
allison Coke: just about to commit it
your code will be:
darbelo bacek: I think --optimize disables NDEBUG.
allison INTVAL result;
Parrot_ext_call(interp, method, "Pi->I", &result);
return result;
dalek rrot: r41987 | bacek++ | trunk (2 files):
Provide macros for directly poking into Context on optimised builds.
 20:00
allison Coke: sorry, that middle line is:
Parrot_ext_call(interp, method, "Pi->I", SELF, &result);
whiteknight bacek+ # r41987 20:02
chromatic --optimize should *enable* NDEBUG.
whiteknight can I ask what NDEBUG does, anyway?
bacek whiteknight: nothing. It's just #define 20:03
whiteknight bacek: right, I'm trying to figure why it is defined
chromatic To disable debugging code.
ttbot Parrot trunk/ r41985 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117355.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 20:04
bacek make benchmark_tests now 37 vs 39 seconds on my box. 20:05
20:06 ash_ joined
dalek rrot: r41988 | allison++ | trunk (3 files):
[pcc] Add a generic extend/embed interface function for invoking Parrot
 20:07
cotto_work Usage in lib/Parrot/Harness/Options.pm could use an update if someone's bored. 20:08
s/Usage/Usage()/
whiteknight leaves for home. Later
ash_ i have a platform question, maybe someone here can answer, has anyone tried to build parrot for an arm based processor yet? 20:09
cotto_work ash_, we seem to have netbsd4.0-arm-gcc-4.1.2 listed in PLATFORMS 20:10
ttbot Parrot trunk/ r41987 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117390.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
darbelo cotto_work: (Usage) Ticket it! 20:12
cotto_work I'll do that if nobody gets it soon. It's a pretty trivial fix. 20:13
20:15 joeri joined, joeri left
darbelo cotto_work: What's missing from the usage? 20:16
ttbot Parrot trunk/ r41988 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117422.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 20:17
cotto_work it mentions jit
and omits -r
darbelo We left the 'JIT' interface in place agter the removal.
you can still tun make testj even. 20:18
cotto_work right, but that doesn't mean we need to leave the documentation as-is
darbelo Leave the option there, but don't mention it usage?
cotto_work Documenting it as a synonym for fast would probably be better. 20:19
darbelo ok, going with that. 20:20
20:22 nbrown joined
dalek rrot: r41989 | bacek++ | trunk/include/parrot/context.h:
[cage][codetest] Wrap MACRO arguments, indenting, trainling whitespaces.
 20:23
rrot: r41990 | bacek++ | trunk/src/call (2 files):
[codetest] Fix POD syntax
rtcl: 7945915 | coke++ | docs/spectest- (2 files):
record another spectest run
rtcl: 326fb65 | coke++ | (2 files):
Fix build against post-1.7.0 parrot; allison++ for the new function.
shorten dalek's url is at xrl.us/bftozr
shorten dalek's url is at xrl.us/bftozt
nopaste "darbelo" at 200.49.154.173 pasted "Updated usage for cotto_work" (19 lines) at nopaste.snit.ch/18405 20:24
darbelo cotto_work: nopaste.snit.ch/18405 ?
cotto_work +1 20:26
purl 1
darbelo <purl--> 20:27
20:28 theory joined
darbelo cotto_work: commited as r41991 20:29
cotto_work darbelo++
darbelo cotto++ # I am merely a vessel. 20:30
dalek rrot: r41991 | darbelo++ | trunk/lib/Parrot/Harness/Options.pm:
Update the test harness Usage(). cotto++ for pointing it out.
ttbot Parrot trunk/ r41990 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117509.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 20:33
bacek ok. Looks like it's good time to merge CallSignature and Context. 20:34
cotto_work bacek, why now? just to shake things up more?
;)
bacek cotto_work: indeed :)
darbelo I'd wait for somebody to fix w32 first. 20:35
But I'm just cautious like that ;)
ash_ cotto_work: cool, just curious if anyone had done it successfully, i am trying to build it, but i keep getting some errors, it might be because I am cross complining it though, rather than building on
oops
meant to delete that
bacek darbelo: w32? I have no idea what you are talking about :) 20:36
ash_ does the parrot config script support cross compiling?
darbelo ash_: unlikely.
It relies too much on the data it gets out of perl.
ash_ so not without modifing the make script? or not without modifing lots? 20:37
i guess, how integrated is the build process with perl? 20:38
darbelo ash_: You would need to, at least, give the target platform a hints file.
Coke allison, chromatic : there's your new segfault. (#1131) 20:39
allison Coke: thanks!
dalek TT #1131 created by coke++: segfault in Parrot_assign_p_p after merge of pcc branch.
ttbot Parrot trunk/ r41991 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117537.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 20:40
20:40 dukeleto joined
Coke the failure is in assign, but it's right next to something that passes in a sub for something else to invoke, so I'm guessing that's the real culprit. 20:40
(this is the same bit of code that helped trigger the inferior runloop issue earlier) 20:41
pmichaud++
allison++ 20:42
pmichaud: let me know if there's any prep work I can do.
chromatic: in my defense, I stripped that down to 2 lines of tcl from several hundred. =-) 20:43
chromatic That's still several hundred too many lines of Tcl for me before lunch.
Coke I cannot generate errors in negative lines of tcl. sorry.
I think the PIR snippet is instructive, though. 20:44
jonathan Ah, pcc_reapply merged! :-)
jonathan hasn't really been watching today ;-)
Nice!
cotto_work jonathan, apparently win32/msvc has a sad. 20:45
jonathan Aww. Poor win32/msvc. 20:46
I'm currently busy slide-hacking.
I'll try and steal a moment at a conference tomorrow to look at that though.
Coke I tucked my daughter in last night; the cat snuck into her room, and I promised to come back in 5m to let him out. she came out 20m later, letting the cat out, came over, gave me a kiss on the cheek and said, "it's ok. you can has a forget."
Took me another 30s to figure out she meant and then what I had forgotten. 20:47
jonathan kids++
The generation that will speak fluent lolspeak. :-) 20:48
Coke (this is the kid who I was horrified to realize I had given the initials IRC.) 20:50
though I know a guy who's daughter's middle name is perl.
cotto_work the tla space is pretty crowded. I had a roommate who associated me with crude palm oil because of my initials. 20:51
dalek rrot-plumage: ac7b845 | japhb++ | :
[TOOLS] import_proto.p6: Check out repos for further analysis; skip epox...
 20:52
shorten dalek's url is at xrl.us/bftpey
dalek rrot: r41992 | chromatic++ | trunk (4 files):
[IMCC] Enabled escape processing in double-quoted PIR subroutine names.
 21:00
TT #1125 closed by chromatic++: [BUG] imcc handles backslashes improperly in sub names 21:04
nopaste "NotFound" at 213.96.228.50 pasted "A preview of my new toy" (383 lines) at nopaste.snit.ch/18406 21:08
Tene NotFound: "winxed"? 21:09
NotFound Tene: yes
If you like an intelectual reference, is an invented word from Finnegan's Wake X-) 21:10
chromatic I found a bug in the grammar already, NotFound.
NotFound chromatic: What bug? 21:11
purl bug is www.cbttape.org/funny/bug3.jpg or img227.imageshack.us/img227/2596/featureiu1.jpg
chromatic The grammar allows zero whitespace characters between the identifier and the assignment operator. Clearly a bug! 21:12
ttbot Parrot trunk/ r41992 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117635.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ )
NotFound That's just my personal preference. 21:13
cotto_work maybe it's just a postfix sigil
darbelo NotFound: Looks cool. What's the target niche? 21:14
NotFound People used to javascript syntax. 21:15
darbelo Have you posted it somewhere? 21:16
NotFound Not yet.
The copmpiler is written in C++, with the idea of rewriting in itself some day. 21:17
21:18 bacek joined
cotto_work Oh good. It'll be nice to see an external tool working with Parrot. 21:19
21:19 Whiteknight joined
NotFound cotto_work: yes, one of the ideas is to test a non-PCT approach. 21:21
cotto_work NotFound, are you generating pir or PAST or what? 21:22
NotFound pir
Without runtime lib, the few predefined functions are inserted. I ommited them in the nopaste for brevity. 21:23
Whiteknight Matrixy built just fine with newparrot 21:30
so that makes me happy
dukeleto Whiteknight++ 21:31
dalek kudo: c4e6d0f | (Geoffrey Broadwell)++ | src/builtins/io.pir:
Implement mkdir
 21:32
shorten dalek's url is at xrl.us/bftpmi 21:33
Whiteknight Matrixy also passes all tests 21:34
so, my work for the evening is basically done
:)
darbelo Whiteknight: Unless you want to get started on parrot-linear-algebra.
Whiteknight darbelo: I just might :) 21:35
actually, I have a ticket to close first, then I'll do that
darbelo is worse than a conscience.
Whiteknight much worse :) 21:36
dalek rrot: r41993 | mikehh++ | trunk/src/call/context_accessors.c:
set svn properties
 21:37
darbelo Whiteknight: I'm seeing failures on matrixy's "make test" 21:40
Whiteknight darbelo: some of those are expected. I haven't found a good way to mark the whole file TODO
nopaste "darbelo" at 200.49.154.173 pasted "Matrixy failures." (10 lines) at nopaste.snit.ch/18407 21:41
Whiteknight the fortran_helpers.t and loadlibrary-examples.t are dependent on CBLAS, so they are supposed to fail
I've never seen computer.t fail (though it's probably because you're on OpenBSD)
darbelo Oh, right. 21:42
Whiteknight we should add OpenBSD support now 21:43
dalek rrot: r41994 | mikehh++ | trunk/MANIFEST:
regenerate MANIFEST
ttbot Parrot trunk/ r41993 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117700.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 21:44
dalek rrot: r41995 | whiteknight++ | trunk/compilers/imcc/parser_util.c:
[imcc] initialize two strings to shut up some warnings about using uninitialized values. TT #1128 (waiting for help from moritz++ to verify)
 21:46
moritz Whiteknight: getting verification will have to wait until after the rakudo release, when it builds again with parrot trunk 21:47
Whiteknight moritz: no worries
I updated the ticket, we can come back to it 21:48
dalek rrot: r41996 | mikehh++ | trunk (2 files):
fix codetest failure - replace hard tabs with spaces
 21:50
ttbot Parrot trunk/ r41994 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117737.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 21:51
Parrot trunk/ r41995 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117759.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 21:56
mikehh I am getting a pre-confiug test failure in trunk 21:59
t/steps/auto/frames-01.t - Failed tests: 12, 15 22:00
ttbot Parrot trunk/ r41996 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117785.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 22:02
darbelo mikehh: The frame builder is hosed. Detection was disabled, so failures there are to be expected. 22:05
dukeleto sees that everyone has made good their promise to nuke trunk. GREAT JOB! 22:06
dalek rrot: r41997 | mikehh++ | trunk/t/op/annotate-old.t:
set svn properties
 22:09
22:11 fperrad_ joined
darbelo dukeleto: we try :) 22:14
ttbot Parrot trunk/ r41997 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117840.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 22:18
dukeleto maybe we need to start calling darbelo "The Wrecking Ball" 22:19
darbelo is the magical breaking robot. 22:20
cotto_work he does seem to like ripping stuff out 22:24
darbelo Yeah, but I usually keep the working bits ;)
cotto_work darbelo?
purl i guess darbelo is Daniel Arbelo Arrocha <mailto:arbelo@gmail.com>
Util docs/embed.pod says that Parrot_PMC_invoke is deprecated. What took its place? 22:26
chromatic One of the PCC functions, I suspect. 22:30
dalek p-rx: 31f8ae7 | pmichaud++ | src/NQP/ (2 files):
Add some (non-working) block rules to NQP.
 22:37
p-rx: 9104507 | pmichaud++ | src/ (4 files):
Add some primitive regex debugging capabilities.
shorten dalek's url is at xrl.us/bftqpg
dalek p-rx: 9681684 | pmichaud++ | src/NQP/ (2 files):
[nqp]: Add statement_control:sym<if> .
shorten dalek's url is at xrl.us/bftqpk
dalek's url is at xrl.us/bftqpn
dalek TT #1132 created by jkeenan++: t/steps/auto/frames-01.t: Failures following pcc_reapply merge 22:39
22:43 Zak joined 22:44 hercynium_ joined
dalek rrot: r41998 | jkeenan++ | trunk (2 files):
TODO two tests due to 'temporary' disabling of build frames functionality at time of pcc_reapply branch merge. Cf.: trac.parrot.org/parrot/ticket/1132.
 22:46
22:47 NathanJY joined, Aisling joined 22:52 patspam joined
ttbot Parrot trunk/ r41998 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117898.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 22:52
chromatic Hm, we're not using the fixed size allocator in Contexts. Let's time that. 22:56
darbelo guesses ~7%
bacek_at_work chromatic, we are using fixed size allocator in Contexts. 22:57
chromatic Nope; the symbol defined in the private GC header isn't visible there. 22:58
It *looks* like we are, but the #ifdef instead of #if hides the fact that we aren't.
bacek_at_work chromatic, oh shi...
my fault.. 22:59
chromatic My guess is closer to ~6%. Benchmarking now.
22:59 kid51 joined
chromatic Good news: it's an improvement. Bad news, only 2.339%. 22:59
Are we going to take it? YES WE CAN!
cotto_work silly question 23:00
of course
chromatic Let me run the test suite and then we will.
23:02 jsut_ joined
cotto_work that parallel build bug is getting old 23:03
dalek rrot: r41999 | chromatic++ | trunk/src (2 files):
[PCC] Made the Context PMC used the fixed-sized allocator for its contained

GC_USE_FIXED_SIZE_ALLOCATOR definition is in the private GC header file, it was never visible here. This gives a 2.339% performance improvement in the fib.pir benchmark.		 23:05
purl It seems faster
dukeleto i really like the detail that went into this page: wiki.postgresql.org/wiki/Submitting_a_Patch . we might want to emulate it 23:06
japhb I think I'm going to need to teach Plumage to patch source trees after fetch. Lots of proto-created projects assume the old build directory layouts, and that's hardcoded in their Configure.pm modules. :-(
davidfetter dukeleto :)
dukeleto wiki.postgresql.org/wiki/Reviewing_a_Patch is pretty nice as well
davidfetter <-- mostly wasting time on the pg project lately
i'm afraid we're a heresy against the church of xp ;) 23:08
dalek rrot-plumage: 1786f1d | japhb++ | :
[TOOLS] import_proto.p6: Begin repo analysis; currently stuck on project...
shorten dalek's url is at xrl.us/bftq6f
dukeleto davidfetter: i just like how detailed it is
davidfetter that's how we roll :)
dukeleto is at a FLOSS manual sprint for GSoC, so i will be randomly mentioning things like that for the next few days 23:09
icanhaz.com/gsocbook
japhb dukeleto: as long as you collect all that coolness into a single place. Like a blog post or wiki page .... 23:10
dukeleto japhb: yes, I will try 23:11
japhb excellent
ttbot Parrot trunk/ r41999 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117943.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 23:12
cotto_work ttbot, we already know about that. 23:13
Whiteknight It always makes me happy to see that fixed-size allocator actually have a positive impact on performance 23:16
even if it's a meezly 2.34%
chromatic I'm hunting 30% and 40% improvements, but if we get a series of 2 and 3 and 4s, that's fine today. 23:17
It's probably time to pull mark_special into Parrot_gc_mark_PMC_fun or whatever its name is now. 23:19
Parrot_gc_mark_PMC_alive_fun(). 23:20
Whiteknight I'm in favor of that
I still want to try out my idea of the context thing in the runloop 23:21
so I may work on that now
chromatic We're up to almost an 11% trunk improvement today so far.
darbelo Whiteknight: Do you want me to convert matrixy to use a nqp Configure like I did for parrot-linear-algebra?
Whiteknight darbelo: That would be awesome if you have the tuits 23:22
23:22 dalek joined
darbelo Doing it. Sould be a simple matter of copy-pasting. 23:22
I'll have to look at dukeleto's nqp test harness later as well. 23:23
dukeleto darbelo: what do you want to do with it? have a similar one in matrixy?
darbelo steal it :)
dukeleto: yes. 23:24
But I don't know how tied to plumage it is.
dukeleto darbelo: i am flattered you want to steal, but now I worry about it being copy-pasted into a few projects and there being no canonical place for it. it might need to become it's own project that is shipped with some other projects (like plumage)
kid51 also like the level of detail in those two pages about patches on the postgresql.org wiki
dukeleto darbelo: not tied at all to plumage 23:25
kid51: cool!
ttbot Parrot trunk/ r42000 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/117982.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 23:26
23:26 jonathan joined
darbelo Hmm. Too much forking during the infancy can be a bad thing. 23:26
23:26 Coke joined
darbelo I think you are right there. 23:26
23:27 pmichaud_ joined
darbelo I would go for a 'blessed' copy somewhere that people should be able to 'vendor drop' into their projects. 23:27
Separate distribution of the test harness seems silly. 23:28
dukeleto darbelo: i can agree and disagree on that on some points 23:30
darbelo: but I agree with the 'blessed' copy for nowish
darbelo: the harness it totally self contained. just copy t/harness and you are good 23:31
japhb If you make it a separate project, other projects can just depend on it for testing phase, and Plumage will make sure it's installed.
Dog Food. It's What's For Dinner.
23:31 Util joined
dukeleto the harness will go through a large refactor when nqp-rx comes out, which sounds soon-ish 23:31
darbelo dukeleto: It's your code, your policy. I'm just some random dude on IRC talking about stuff he didn't do.
:) 23:32
23:32 PerlJam joined
ttbot Parrot trunk/ r42001 MSWin32-x86-multi-thread make error tt.ro.vutbr.cz/file/cmdout/118011.txt ( tt.ro.vutbr.cz//buildstatus/pr-Parrot/rp-trunk/ ) 23:32
darbelo japhb: true, but doesn't help out-of-plumage distribution. 23:33
Whiteknight this Ops2c script is rediculously convoluted 23:35
japhb darbelo, There are other forms of distribution?!? Perish the thought!
darbelo, seriously, though. Anyone not using plumage can be forced to read the README.
darbelo japhb: only until plumage self-distributes ;) 23:36
japhb: Ture.
s/Ture/True/
cotto_work Whiteknight, yes. I wouldn't at all mind getting back into opsc/pmcc to get rid of those messes.
23:37 dukeleto_ joined
darbelo cotto_work: Do you remember what pmcc was blocking on? 23:37
japhb Since I've had to do it a couple times now, I've actually been considering giving Plumage an "upgrade Parrot" command. Nukes the installed parrot, pulls a more recent checkout, builds, install-dev, and rebuild all previously installed projects on top of it, in dep order.
darbelo japhb++ # +1 23:38
japhb Of course, there are a few details to work out, like picking a parrot rev that doesn't break the HLLs ....
dukeleto darbelo: any help that you can give with the harness is greatly appreciated. The harness itself doesn't have any tests yet.... 23:39
darbelo: you run into a recursive yak hole when you want to test the testers
23:39 plobsing joined
japhb "yak hole" ... oh my, what a turn of phrase 23:40
darbelo japhb: Hmm. It should be user-overridable. I *want* to grab trunk and break all alnguages.
japhb darbelo, sure.
darbelo plumage show-me-the-broken 23:41
japhb I'm thinking of having metadata for each HLL that says where to find their current desired parrot-rev, and then allowing users to say 'upgrade to tip' or 'upgrade to version blessed by project foo'
23:44 jsut joined 23:49 payload joined 23:50 jsut_ joined 23:51 cotto_work joined
darbelo purl: msg Whiteknight Say 'hi' to Configure.nqp next time you drop by the matrixy repo. 23:51
purl Message for whiteknight stored.
cotto_work darbelo, sorry. my connection dropped.
Whiteknight darbelo++
cotto_work I think bacek and I decided that it'd be better to work on the ops first. 23:52
darbelo What stalled the ops then?
dalek rrot-plumage: ac0643f | darbelo++ | :
Update matrixy configure type to NQP.
 23:53
shorten dalek's url is at xrl.us/bftrsk
dalek rrot-plumage: d2daca1 | darbelo++ | :
Merge branch 'master' of gitorious.org:parrot-plumage/parrot-plumage
shorten dalek's url is at xrl.us/bftrsr
dalek rrot-plumage: 968202b | darbelo++ | :
Merge branch 'master' of gitorious.org:parrot-plumage/parrot-plumage
shorten dalek's url is at xrl.us/bftrs5
darbelo wtf? I'm pretty sure didn't merge anything.
dalek trixy: c9616b2 | darbelo++ | Configure.pl:
Get Configure.pl out of the way.
shorten dalek's url is at xrl.us/bftrt2
dalek trixy: ab4de52 | darbelo++ | (2 files):
Add Configure.nqp and helper library from parrot-linear-algebra.
shorten dalek's url is at xrl.us/bftrt4
dalek trixy: 3fd9ba3 | darbelo++ | config/makefiles/root.in:
The new Configure doesnt support #IF() in makefile templates, remove them.
shorten dalek's url is at xrl.us/bftrt6
cotto_work I can't remember why we put opsc on hold. 23:55
Whiteknight I wish you hadn't :) 23:58
seriously, the more I see of Ops2c, the less I like it
chromatic Hah, we're 12.01% faster on this benchmark for the day. 23:59
Don't ask about how we are for the week though.

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!