Parrot 2.8.0 released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | close 25 tickets; remove deprecated items (especially CodeString);
Set by moderator on 24 September 2010.
cotto bacek_at_work, he got killed. There was some kind of hardware problem with feather involving smoke. 00:06
bacek_at_work cotto, oookey. 00:07
GeJ is thinking about writing a non-C<print>ing-by-default Data::Dumper 00:15
unless capturing STDOUT to a String is doable. 00:16
bacek_at_work Data::Dumper doesn't print by default. 00:17
or you mean parrot's D::D?
GeJ parrot's D::D 00:18
Data::Dumper::Default prints everywhere. 00:20
00:29 whiteknight left
bacek_at_work yeah... I think we should change it to produce string. 00:34
But it will require touching a lot of things. Including PCT
GeJ how so? 00:44
bacek_at_work PCT tests are using D::D afair 00:45
GeJ Well, I'd say that most of the tests are Perl tests. Rewriting D::D is some yak::shaving in my grand plan of nuking every instance of "#! perl" from t/ 00:53
00:54 tcurtis left 01:12 dngor left, dngor joined 01:17 tcurtis joined 01:24 dngor_ joined 01:26 dngor left 01:59 davidfetter left
mikehh Gej: you still around 02:23
GeJ Oh I'm still here for sure. 02:25
GeJ just discover D::D's __dump callback. 02:27
mikehh Gej: in looking at the ASSERT_ARGS test there are a couple of things you need to look at 02:32
GeJ: if the test - t/codingstd/c_arg_assert.t FAILs it usually lists the functions that cause the problem 02:34
Gej: if they are listed in .h files only you need to check if the function exists in .c files use -> ack function_name 02:35
02:35 janus left, janus joined
mikehh Gej: if not you probably need to run make headerizer 02:36
Gej: ASSERT_ARGS needs to be the first thing in the c function straignt after the opening { - ASSERT_ARGS(function_name) [no ;] 02:37
GeJ what's this headerizer. I've heard the name several times, but I haven't figured out what it does yet. 02:38
mikehh Gej: it generates header files (.h) from .c files with appropiate assertions/defines etc. 02:39
02:56 theory left
GeJ Okay. Thanks for the explanations. At least I'll have a clearer idea of what to do the next time this test breaks. 02:59
mikehh++
mikehh Gej: it took me quite a while to work it out 03:03
03:43 dngor_ is now known as dngor 03:47 dukeleto joined
dukeleto hmm, we never updated the /topic for this weeks priorities 03:48
04:05 mikehh left 04:06 davidfetter joined
cotto dukeleto, ping 04:27
dukeleto, unping 04:45
dukeleto cotto: unpong ? 05:16
moderator Parrot 2.8.0 released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | smoke GC-related branches and attack GC tickets 05:17
05:44 ash_ left 05:45 theory joined 05:47 theory left, tcurtis left
cotto dukeleto, I was asking about something I shortly thereafter realized that you'd put on the GitMigration page 05:54
dukeleto cotto: how is stuff coming along? 05:55
cotto barring further suggestions, I'm calling the github plugin "ready". 05:56
the test cases exposed some bugs that I'm glad people won't have to run into post-migration
Actually, I do need to put together a recommended configuration that includes the navadd plugin so trac.parrot.org can have the browser nav item link to github. 06:00
the code's there though
(or so I'm trying to convince myself)
dukeleto i need to sit down and port mk_lang_shell/create_language, but I haven't found strong enough whisky yet 06:06
cotto What's the hard part? 06:08
06:10 dukeleto_ joined
cotto dukeleto_, what's the hard part of converting the language shell builder scripts? 06:15
cotto is not sure why those scripts feel the need to be self-contained. 06:17
though I can see the need for whiskey 06:18
dukeleto cotto: the hard part is not breaking every existing parrot tool that depends on svn revs 06:29
cotto: and a way to depend on versions of parrot and being able to tell if $version_string_1 is older/newer than $version_string_2 without a git repo 06:30
cotto: i think i understand what needs to happen, but I haven't gotten my half-baked ideas into a plan or usuable code, yet 06:31
cotto: "git describe" and some heuristics are involved 06:32
cotto Is that something that you can solve between 2.9 and 2.10? 06:33
I recall something about using an augmented commit id in the form of yyyy-mm-dd-hhhh-hash. 06:35
to make it easy to compare versions 06:36
06:38 fperrad joined
dukeleto cotto: git describe is kind of like that, but with nukes attached 06:39
cotto: if you make the version string yyyy-mm-dd-hhhh-hash, how can a tool compare version strings using the hash? wouldn't it have to just base it on yyyy-mm-dd-hhhh ? 06:40
cotto: it is pretty sad that doing date arithmetic is much easier than version string arithmetic. It is a sign that something is wrong. 06:41
cotto: which means we should do date arithmetic, if we can get away with it
cotto The hash would be in there in case someone wanted to look it up in the repo. 06:42
06:42 kurahaupo joined
dukeleto cotto: just using the timestamp of the parrot release would seemingly work, but the case of those minor versions throws a wrench in there 06:42
cotto: in theory x.2.1 could come out after x.3.0, if "supported releases" had a security fix applied 06:43
cxreg volunteers as an extra "training and questions" git person
cotto cxreg, great. We can never have too many of those.
dukeleto cotto: that hasn't ever happened. and if we think that we won't want that, then our version arithmetic gets a lot less hairy
cxreg++
cxreg: good to hear from you
cxreg hey. been stuck in divorce land :( 06:44
dukeleto cxreg: you can add yourself to the GitMigration wiki page if you want
cxreg: sorry to hear that. We have lots of things to hack on if you need to get your mind off of it ;)
cxreg: trac.parrot.org/parrot/wiki/GitMigration
cxreg heh, hopefully soon. i'm also coming out of the water on a big project at work 06:45
might need some open source goodness to clear my head
cotto knows the feeling
dukeleto cxreg: sounds good. PL/Parrot still needs plenty of love, too
cxreg: i have PL/Perl6 working reasonably, so you can use Perl 6 grammars in procedures 06:46
cxreg sweet!
anyone ever get spi working proper?
i could dive back in on that if not
dukeleto cxreg: not any more than your branch. the parrot nci stuff seemingly hasn't landed
cxreg: that would be awesome
cxreg crap, that sucks. i wonder what happened with ash_ 06:47
's gsoc
dukeleto cxreg: the branch is still there, i think it hasn't been merged yet, tho, and I am not entirely clear why
cotto dukeleto, so that problem could introduce itself if someone has 3.6.1 (released after 3.7.0) installed and tries to build a language which depends on a 3.7 feature.
dukeleto cotto: indeedy. i have been thinking about it for a while. that edge case really makes life annoying. 06:48
06:48 jan left
cxreg dare i ask if lorito is out of the imaginary phase at all yet? 06:48
dukeleto cxreg: yes, there are actually a bunch of github implementations that do various things
cotto there's an unblessed implementation by atrodo++ on github
cxreg oh cool 06:49
cotto I don't know how much scrutiny it's gotten though.
dukeleto cxreg: also github.com/ekiru/yalp-asm
cxreg: github.com/atrodo/lorito
cxreg: the spec of what lorito is a lot more solidified these days, but implementation details still lurk 06:50
cotto I'm planning on getting out the Lorito whip once we're all migratorated to git.
dukeleto cotto: yes, indeed. 06:51
cxreg postgres just finished their git migration
im sure dukeleto knows :)
dukeleto cxreg: github.com/parrot/parrot/compare/ma.....gsoc_nci <-- looks like ash just merged trunk in 2 days ago 06:52
cxreg: perhaps that branch needs some smoking
cxreg oh thats good 06:53
dukeleto cxreg: yes, i heard they found bugs in the git tools and their authors promptly fixed them and then postgres could actually convert to git
lollerskates
moritz ash said he needed to write more tests
cxreg dukeleto: yeah it took a while, but it's Officially Official now
the day following 9.0, to be exact 06:54
dukeleto moritz: i am going to run "make smoke" on it now and see what happens
moritz: thanks for the info 06:55
cxreg: i was going to try PL/Parrot on the release, hopefully it works :)
cxreg: do you know anything about making debian packages? I am trying to make a debian package of PL/Parrot
cotto Ouch. They moved from cvs. 06:57
I'd gladly fix bugs on someone else's code to get away from that.
kurahaupo waves hand for building Debian packages 07:08
cxreg dukeleto: only enough to be dangerous :)
kurahaupo dukeleto, cxreg: you want help? 07:09
cotto night
dukeleto kurahaupo: oh really? 07:14
kurahaupo dukeleto: Yes, I've been building Debian (well really Ubuntu) packages for work. (Sorry, I've been lurking away from Parrot for many months.) 07:15
dukeleto kurahaupo: ubuntu package would be awesome.
odd thing about the gsoc_nci branch, is that when I build it, the src/pmc/nci.pmc file is modified, with a bunch of stuff deleted 07:33
seems like something in the build is changing the file
nopaste "bacek" at 192.168.1.3 pasted "make coretest results on gen_gc branch." (10 lines) at nopaste.snit.ch/23890 07:36
bacek_at_work ho-ho-ho
It's almost here
GeJ dukeleto:I'm no expert, but it looks like src/pmc/nci.pmc is replaced at config time depending on HAS_LIBFFI 07:41
see config/gen/libffi.pm 07:42
dukeleto GeJ: you are getting to be an expert. I just figured that out :) 07:45
GeJ: i didn't have libffi installed
bacek_at_work: can we put milk and cookies out for you, somewhere?
bacek_at_work nope
only beer and steaks
kurahaupo (Bacek lives in Oz?) 07:46
GeJ is diving into Data::Dumper::Default... scratches head.
bacek_at_work kurahaupo, yes. Sydney.
GeJ Aaaah, Sydney.
dukeleto smokes gsoc_nci with libffi 07:51
cxreg libfwiffo 07:53
dukeleto smolder.parrot.org/app/projects/rep...etails/368 07:54
thar be smoke in them reports
looks like only 1 failed test and a bunch of incorrect test counts 07:55
cxreg heh, got a nice segfault in nci.t
GeJ could anyone could give me the dumbed down version of the benefits of having libffi over the current nci ?
dukeleto ah, but the test counts are wrong because of segfaults. nice.
GeJ or at least point me to some document explaining it. 07:56
dukeleto GeJ: our current nci doesn't support a lot of data types that people want
07:56 Kulag left, Kulag joined
cxreg in addition, you can't do anything with complex types, such as structs 07:57
not sure what else we get
i was wondering about read/write parameters 07:58
(eg, passing &foo)
dukeleto GeJ: there may be some emails in the parrot-dev archives 08:00
GeJ: www.parrot.org/content/gsoc-project...ovements-0 08:01
GeJ ll 08:05
oops 08:06
dukeleto github.com/parrot/parrot/graphs/impact looks pretty crazy if you scroll the right a bit 08:08
08:09 Kulag left, Kulag joined 08:10 dukeleto_ left 08:12 dukeleto_ joined 08:18 Kulag left 08:21 dukeleto_ left, Kulag joined 08:22 dukeleto_ joined 08:27 dukeleto left 08:38 dukeleto_ left 08:42 Kulag left 08:43 esskar__ joined, Kulag joined 08:45 ppant joined 08:46 contingencyplan left 08:47 esskar left, esskar__ is now known as esskar 08:48 Drossel joined 08:50 Kulag left 08:54 Kulag joined 08:55 Drossel left 09:04 Kulag left 09:13 Kulag joined 09:20 Kulag left 09:22 mikehh joined 09:34 Kulag joined 09:41 Kulag left 10:10 ruoso left 10:15 ppant left 10:18 tadzik joined 10:29 kurahaupo left 10:54 jsut_ joined 10:58 allison joined 10:59 jsut left 11:09 allison_ joined 11:11 allison left, allison_ is now known as allison 11:38 tadzik1 joined 11:42 tadzik left 11:53 esskar__ joined 11:57 tadzik1 is now known as tadzik, esskar left, esskar__ is now known as esskar 12:05 whiteknight joined
whiteknight good morning, #parrot 12:07
12:27 cognominal left 12:43 Coke left 12:50 bluescreen joined 12:53 tcurtis joined, tcurtis left 13:02 Kulag joined 13:14 perlpilot joined 13:15 cognominal joined 13:20 Patterner left 13:21 ash__ joined 13:29 Psyche^ joined, Psyche^ is now known as Patterner 13:32 perlpilot left, PerlJam joined, ruoso joined 13:46 whiteknight left
cotto ~~ 13:47
atrodo Morning cotto 13:51
awfully early for you, isn't it?
cotto yes
my brain wouldn't go to sleep
atrodo Bad brain 13:52
13:52 davidfetter left, whiteknight joined
whiteknight blah. The power went out 14:01
14:02 esskar__ joined
cotto plobsing, ping 14:04
aloha, seen plobsing 14:05
aloha cotto: plobsing was last seen in #parrot 2 days 7 hours ago saying "should work. last reported success was 20090720 though (according to PLATFORMS).".
14:05 esskar left, esskar__ is now known as esskar 14:18 mikehh left 14:21 whiteknight left 14:29 contingencyplan joined 14:36 GodFather joined
cotto aloha msg plobsing Can you take a look at trac.parrot.org/parrot/ticket/1745 ? The bug appeared when dynop_mapping was merged. 15:00
aloha cotto: OK. I'll deliver the message.
15:15 cotto left 15:21 Hunger left 15:28 Hunger joined 15:40 cotto joined 15:54 sjn left, esskar__ joined 15:55 ruoso left 15:58 theory joined, esskar left, esskar__ is now known as esskar 15:59 M_o_C joined 16:18 davidfetter joined 16:24 dukeleto joined
dukeleto waves|particles hello 16:25
davidfetter diffracts, quantizedly 16:26
16:26 allison left
davidfetter what's shakin', your grace? 16:27
dukeleto is barely awake and becoming friends with his coffee cup 16:30
davidfetter: what have you been up to lately? 16:31
ash__: ping
ash__ pong
16:32 jan joined
dukeleto ash__: let me know how I can help with getting the gsoc_nci tests passing, or adding more 16:32
ash__: can you reproduce those failures on your linux box? 16:33
ash__: what does "Unknown param Signature I" mean?
ash__ I haven't had a chance to test them on my linux box, but i should have some free time this afternoon or tomorrow, an unknown signature param is an unknown identifer passed to delfunc 16:35
i haven't implemented I it would seem, thats odd, it doesn't have any tests in t/pmc/nci.t 16:37
hm, no, i do implement it at: trac.parrot.org/parrot/browser/bran...mc.in#L361 so, something else must be causing that error 16:38
davidfetter dukeleto, enjoying vmware, wondering whattf i should be doing about pl/parrot and whentf i'll actually have time to do it 16:43
davidfetter embraces the tf suffix
davidfetter contemplates running parrot directly atop esx 16:45
16:48 sjn joined 16:50 whiteknight joined
davidfetter dukeleto, yourself? 16:51
16:54 GodFather left 16:58 M_o_C left 17:02 allison joined
dukeleto davidfetter: knee-deep in the parrot git conversion 17:14
davidfetter w00t!
go, dukeleto, go!
dukeleto davidfetter: cotto and whiteknight are flying the git conversion flag as well. we are getting close 17:20
davidfetter w00t!!!!
did you hear about the pg conversion?
(to git)
PerlJam Does PaFo have a github account?
dukeleto PerlJam: github.com/parrot 17:21
davidfetter: yep, i was watching the fireworks
davidfetter as fireworks go, they were unspectacular 17:22
davidfetter considers writing up what happens when 2 FLOSS projects that are reasonably functional come together
davidfetter just wishes the pointy-haired of the world hadn't so abused the word, "synergy"
17:23 fperrad left 17:26 ruoso joined
dukeleto davidfetter: parrot synergistically uses distributed version control to enable web 3.0 technologies 17:36
davidfetter heh 17:39
17:43 bluescreen left 17:44 bluescreen joined
dukeleto davidfetter: i am still recruiting help to get debian/ubuntu packages for PL/Parrot, and we got an rpm spec recently 17:45
davidfetter: i want to have some actual end-users so they can give us some feedback
davidfetter lemme pester some debian people over in #postgresql
17:51 bluescreen left, ruoso left, bluescreen joined 17:53 mikehh joined
ash__ is there an easy way to set a break point in a pmc? in the gdb 18:02
whiteknight probably not by line number 18:07
what are you trying to monitor?
ash__ i am getting an error freeing something 18:12
18:13 ruoso joined
ash__ and i am trying to figure out why, but its hard to inspect values if i can't break before the error, but i figured out a way, i just said to break at the full name of the C function the pmc vtable method gets turned into (Parrot_NCI_invoke) and it lets me break then 18:13
weird, i have a test that if i run it like: ./parrot t/pmc/nci_55.pasm it outputs the correct thing, but if i do: perl t/pmc/nci.t it says the test fails 18:23
whiteknight there are some tests like that, I don't know what the difference is in test invocation 18:24
18:25 Coke joined
ash__ commenting out all of my memory free's lets it work, it doesn't give me an error on os x, i wonder why it is on linux... 18:26
18:33 dalek joined
ash__ still i had to remove all of the free's to get it to work without a segfault, so thats not really a solution 18:34
18:36 dalek left 18:37 dalek joined, dalek left 18:38 dalek joined, dalek left, dalek joined, tcurtis joined
dukeleto ash__: have you tried valgrind ? 18:39
ash__ i don't really know how to use valgrind, but i can try to see if it tells me something new
dukeleto ash__: it can tell you about leaked memory, use of uninitialized memory, etc 18:40
ash__ i'll see if i can figure it out 18:41
19:16 ruoso left 19:22 ilbot2 joined
moderator Parrot 2.8.0 released | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | smoke GC-related branches and attack GC tickets
19:25 ingy left, ingy joined 19:32 tcurtis left 19:37 ruoso joined 20:19 ellefu joined 20:24 nwellnhof joined
whiteknight chromatic++ 20:35
21:05 perlite left 21:08 perlite joined
ellefu Is it just me, or does mk_language_shell.pl generate bogus C code for the get() vtable method in the .pmc file? 21:12
There's an undeclared variable called "shape", and the "property" variable is used uninitialized. 21:13
21:14 ingy left, ingy joined 21:17 dalek left
ellefu Hmm. I did an svn up to see if that would fix it, but now the build seems to be broken; the Makefile's CLASS_PMC_FILES includes "src/pmc/imageio.pmc", which doesn't exist. 21:19
21:19 bluescreen left
ellefu (never mind that last bit; just my environment being retarded.) 21:24
21:27 dalek joined, dalek left
dukeleto ellefu: it is possible that mk_language_shell has bugs in the C code for .pmc files 21:31
ellefu: i am not sure many people have used that command line option
ellefu: or pmc vtables could have changed since it was written
davidfetter mmm...cookiez 21:33
22:04 bacek left 22:08 dalek joined, dalek left 22:09 nwellnhof left 22:16 ruoso left 22:17 dalek joined, dalek left 22:18 dalek joined, dalek left, dalek joined 22:27 whiteknight left 22:36 tcurtis joined
dalek rrot: r49402 | plobsing++ | trunk (6 files):
detect if a platform has the BSD stat extensions
 22:37
rrot: r49403 | plobsing++ | trunk/src/gc (4 files):
malloc can return null
dukeleto plobsing++ # the NULL defender 22:54
dukeleto imagines a mutated NULL pointer holding a mop
22:54 tcurtis left
plobsing those commits fall out of porting to minix. 22:54
its malloc/calloc return NULL on empty allocations (which is legal per my linux manpages) 22:55
dukeleto plobsing: i gathered that. thanks for working on it
sorear hey, NULL points to at least 0 addressable bytes, none of which alias any other allocated block 22:58
plobsing yes. it's a nifty trick for systems with small memory resources or with simple implementations 22:59
23:00 tadzik left
plobsing src/io/socket_unix.c:get_sockaddr_in calls inet_{a,p}ton falling back on gethostbyname. any reason we shouldn't just be calling getaddrinfo which does the same? 23:01
dalek rrot: r49404 | plobsing++ | trunk/MANIFEST:
mk_manifest_and_skip
 23:07
cotto plobsing, ping 23:19
plobsing cotto: pong
I will look into the profiling runcore failure
cotto thanks
23:31 dngor left, dngor joined 23:36 whiteknight joined 23:49 d4l3k_ joined, d4l3k_ left 23:50 d4l3k_ joined, d4l3k_ left 23:58 ruoso joined, Patterner left

Powered by Cro and the Raku® Programming Language.

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

Thank you!