Parrot 2.5.0 Released! | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GSOC Students: trac.parrot.org/parrot/wiki/GSoCersStartHere | Priorities: merge gc_massacre, remove deprecated items, add deprecation notices for 2.6, close tickets.
Set by moderator on 25 June 2010.
Coke GAHHHHHHHHH. I am going to go svn blame this code and make whoever it was fix it. 00:23
mikescott? 00:24
this pod2html stuff was added by mikescott in the 5500s.
chromatic I think it came from another project.
Don't quote me, but it has that feel. 00:25
darbelo Coke: "Look, you imported a whole lot of ugly code that got broken over the years you haven't been around. I demand you fix it." ? 00:26
GeJ It's worth trying.
darbelo I'm willing to be proven wrong, but I doubt that'll be met with a lot of exitement... 00:27
I, for one, do not look forward to the day where all of the crappy code I've left behind will catch up with me ;) 00:29
chromatic I solve that by not writing bad code! Heh. Right.
GeJ That's where PAFO is a good thing, now that you have signed a paper, they will find you. Always. 00:30
darbelo OMG. I'm screwed.
GeJ It's a good thing that the New Caledonian postal service lost my CLA. 00:31
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#34644), fulltest) at r47918 - Ubuntu 10.04 amd64 (g++ with --optimize) 00:32
whiteknight darbelo: ? 00:33
darbelo whiteknight: ?
GeJ: Heh. I emailed a pdf. Argentinian mail can't reliably convey mail over distances larger than five kilometers. 00:35
GeJ darbelo: I've heard that they're using sea turtles over here. I bet they're far less reliable than any kind of camelid the Argentinian mail service is using to move mail to Portland. 00:41
darbelo We use llamas and guanacos. Great for crossing arid land. They don't deal gracefully with oceans, however. 00:44
dalek tracwiki: v7 | mikehh++ | CFunctionDocs 00:49
tracwiki: trac.parrot.org/parrot/wiki/CFuncti...ction=diff
rrot: r47919 | tcurtis++ | branches/gsoc_past_optimization/runtime/parrot/library/Tree/Pattern.nqp:
Add support for options in Tree::Pattern.transform.
rrot: r47920 | tcurtis++ | branches/gsoc_past_optimization/runtime/parrot/library/Tree/Pattern/Transformer.nqp:
Add *%adverbs to the signature of Tree::Pattern::Transformer.new.
rrot: r47921 | darbelo++ | branches/gsoc_nfg (8 files):
Redo the packfile dance.
rrot: r47922 | darbelo++ | branches/gsoc_nfg/src/packfile/pf_items.c:
Fix the non-ICU build.
darbelo goes off to ingest some dead cow. 00:51
GeJ we tried with jellyfish(es?) first, problem is that they only go where the current sends them. Then we tried sharks, same problem only they're tricky to tame and only go where they want to go. Then tuna, and we found out that all our mail ended up in sushi bars in Japan. Now sea turtles... looks better, but not 100% sure.
00:54 bluescreen joined
GeJ leaves to eat dead bird. 00:55
00:59 abqar_ joined
dalek tracwiki: v5 | mikehh++ | MissingCFunctionDocs 01:05
tracwiki: trac.parrot.org/parrot/wiki/Missing...ction=diff
01:06 plobsing joined
dalek rrot: r47923 | tcurtis++ | branches/gsoc_past_optimization/runtime/parrot/library/Tree/Pattern/Transformer.nqp:
Add .depth;
 01:06
rrot: r47924 | tcurtis++ | branches/gsoc_past_optimization/runtime/parrot/library/Tree/Pattern/Transformer.nqp:
Add Tree::Pattern::Transformer.min_depth and thus Tree::Pattern.transform(..., min_depth(...)).
01:10 dalek left, dalek joined
Coke YAY. progress. 01:10
mikehh kid51: everything seems ok with the merge, I've modified the Trac:wiki docs, you might want to review that 01:15
dalek rrot: r47925 | coke++ | branches/html_cleanup (1 files):
Actually generate some HTML output again.

a very ugly state for now. if you run tools/docs/make_html_docs.l you can see we have several dynpmcs that have no POD documentation. This should probably be fixed in trunk.		 01:23
rrot: r47926 | Chandon++ | branches/gsoc_threads (13 files):
Progress on alarms / timers.
tracwiki: v3 | mikehh++ | BoilerplateOnlyDocs
tracwiki: trac.parrot.org/parrot/wiki/Boilerp...ction=diff
kid51 mikehh: Thanks. Will do that in next few days. 01:32
01:45 contingencyplan joined 01:49 tcurtis joined 02:04 Chandon joined 02:09 bluescreen joined
dalek rrot: r47927 | Chandon++ | failed to fetch changeset:
[gsoc_threads] Merge from trunk.
 02:29
rrot: r47928 | tcurtis++ | branches/gsoc_past_optimization/t/library/pastpattern.t:
Add some tests for :min_depth.
 02:46
rrot: r47929 | plobsing++ | branches/dynop_mapping/src (2 files):
satisfy c_function_docs.t
02:52 kid51 joined
kid51 mikehh ping 02:53
mikehh kid51: pong 02:58
kid51 Mike: I looked at those two wiki pages.
I'm puzzled about this: "Bear in mind we are dealing with POD and that it expects a maximum line length of 80."
To the best of my knowledge, POD has no expectation of its own re maximum line length. 02:59
cotto ~~
kid51 AFAICT, the '80' simply comes from a setting coded by Coke in lib/Parrot/Headerizer.pm.
An *appropriate* setting, IMO, but nonetheless not connected to anything in POD. 03:00
03:00 janus joined
mikehh kid51: I changed that - used to be 78, because that was defined, but Headerizer formats it to 80 characters, the =item line that is, and it fails if it is different 03:00
kid51 Okay. While I think it's okay to say "80 characters" for these particular POD items, I don't think it's correct to imply that this is a requirement of POD itself. 03:02
In any case, after tonight's merge we should no longer have any files needing listing here: trac.parrot.org/parrot/wiki/MissingCFunctionDocs 03:03
... even though there probably should be files listed here: trac.parrot.org/parrot/wiki/BoilerplateOnlyDocs
mikehh I got that from somewhere, I don't remember now, but I found it somewhere, maybe Pod::Simple or podchecker or something
kid51 k
chromatic pod2man seems to prefer some 78 characters.
kid51 chromatic: True, that may be its default setting. But we can adjust it however we want. 03:04
I don't see any need to change anything here.
that is, anything in the code 03:05
kid51 must sleep
purl $kid51->sleep(8 * 3600);
mikehh I'll think about re-wording that
I think I need some sleep, just had the 4am news 03:06
chromatic: btw I did find 1 error in control_flow.pod - line 427 - C(<$_>) instead of C<($_)> 03:13
I am going cross-eyed, will continue later
chromatic Thanks! 03:14
bacek_at_work ~~ 03:29
cotto, www.thinkgeek.com/homeoffice/mugs/78fb/ :)
03:42 mikehh_ joined
cotto apropos 03:53
pmichaud, ping 04:01
Chandon The headerizer is a big jerk. If I format function headers wrong it just ignores them and doesn't even tell me it's doing so. 04:06
plobsing Chandon: that's not so bad. I'm pretty sure it inserted syntax errors on me once. 04:10
Chandon That's a pretty sweet feature. 04:12
cotto one of many things that are nice when they work 04:13
chromatic Like garbage collection or register allocation! 04:15
dalek rrot: r47930 | tcurtis++ | branches/gsoc_past_optimization/runtime/parrot/library/Tree/Pattern/Transformer.nqp:
Implement :descend_until attribute and option. Not yet tested.
 04:28
04:54 LoganLK joined 05:29 fperrad joined
dalek rrot: r47931 | petdance++ | trunk (2 files):
function annotations
 05:50
rrot: r47932 | plobsing++ | trunk/src/packfile.c:
document what fixup_packed_size does
 06:07
rrot: r47933 | tcurtis++ | branches/gsoc_past_optimization (2 files):
Make :descend_until actually work and write tests for it.
06:09 uniejo joined
dalek rrot: r47934 | petdance++ | trunk/src/pmc/fixedintegerarray.pmc:
consting and annotating
 06:23
06:27 mmcleric joined 06:44 AndyA joined 07:05 TiMBuS joined
dalek rrot: r47935 | tcurtis++ | branches/gsoc_past_optimization/t/library/pastpattern.t:
Add failing test for :descend_until with :min_depth.
 07:12
rrot: r47936 | tcurtis++ | branches/gsoc_past_optimization/runtime/parrot/library/Tree/Pattern/Transformer.nqp:
Make :descend_until and :min_depth work together.
08:49 AndyA joined 08:50 mikehh_ joined 08:52 mikehh_ joined 09:32 mmcleric_ joined 09:37 lucian joined
dalek kudo: 261eb2a | moritz++ | README:
README: emphasize that installing Rakudo is really necessary
 10:03
moritz purl: msg tcurtis please checkout the 'optimizations' branch of rakudo. I currently only run the assign_type_check stage - which returns $/.orig in fold, so it should do nothing... still it blows up when trying to build Test.pm 10:19
purl Message for tcurtis stored.
moritz purl: msg tcurtis github.com/rakudo/rakudo/blob/optim...zer.pm#L69 contains the code... any idea what's going wrong? 10:20
purl Message for tcurtis stored.
10:33 mmcleric joined
bacek ~~ 10:36
dalek kudo: aa015ad | moritz++ | src/Perl6/Compiler.pir:
more awesome error message when running a non-installed Rakudo outside of the

  masak++ for complaining, jnthn++ for telling me where to steal the code
  (alpha's perl6.pirg)		 10:38
10:55 kthakore joined
kthakore bacek_at_work: I hear you had more tickets for me? 10:56
bacek kthakore, I did msg you with them.
kthakore oh ok
bacek: how do I see those again/
bacek /msg purl messages
purl, messages?
purl To access purl's messages, msg me with the word "messages".
kthakore thanks
nopaste "bacek" at 192.168.1.3 pasted "Interesting... Somehow PIRATE generate _faster_ PBC already... Absence of PackfileDebug?" (8 lines) at nopaste.snit.ch/21701 11:29
dalek r: 4d1e009 | bacek++ | src/POST/Compiler.pm:
Factor out POST::Compiler %context creation function.
 11:30
r: d5ef3ea | bacek++ | src/POST/Sub.pm:
Add POST::Sub.constant_index
r: 1c28474 | bacek++ | src/POST/Compiler.pm:
First attempt to implement constant Subs. Require major polish and proper
11:39 GodFather joined 12:04 TiMBuS|Away joined 12:17 ruoso joined 12:20 whiteknight joined 12:30 szbalint joined
whiteknight good morning, #parrot 12:33
bacek aloha, whiteknight
dalek r: 05f14b1 | bacek++ | src/POST/Sub.pm:
Add Sub modifiers.
r: 63b457e | bacek++ | src/PIR/Actions.pm:
Handle .sub modifiers. Not tested.
 12:50
13:00 leprevost joined
whiteknight hello bacek, how are you doing todaY? 13:04
13:05 mmcleric_ joined 13:18 gbacon joined
Coke do we have lib/Pod/* just to avoid people having to install something via cpan? 13:31
Coke will fix that as well. 13:33
looks like lib/Pod/* is used only for 2 codingstd tests and 'make html'. anyone see anything else? 13:34
particle sounds about right to me 13:35
Coke will kill that, make it conditional in the tests, make the html script die nicely without it. 13:36
(and add it to Task::Parrot)
... but after work. =-)
atrodo But working on stuff like this is more fun when you're at work 13:42
(sure glad my boss doesn't hang out in this channel)
Coke atrodo: this channel is logged, btw. =-) 13:43
atrodo Hmmmm, I did forget about that detail. Best never mention it again then
13:49 bubaflub joined 14:01 plobsing joined 14:12 ash_ joined
dalek kudo: 806efc8 | pmichaud++ | src/ (2 files):
Disable IO::ArgFiles for the time being, it's got bugs and is forcing people to
 14:14
14:18 mikehh joined 14:20 kthakore left 14:27 khairul joined 14:28 szbalint joined 14:37 he joined 14:38 mj41 joined 14:45 jdv79 joined
jdv79 I've forgotten. can I and how do I call perl5 DBI from rakudo? 14:45
there was a p5 pmc or something, right?
whiteknight jdv79: There is a Perl5 interop project 14:46
I can't remember the name, it's one of jnthn's on github. "zavolaj" or something
jdv79 ah, thanks
one of the first things i would need to play with rakudo at work would probably be db access for tools and such. just wondering... 14:47
whiteknight some people are working on other db tools in Perl6 or even in raw Parrot. I don't know how mature any of them are yet
there is a group embedding a Parrot interpreter into Postgres, for instance
janus there was a post about postgres use recently 14:48
jdv79 yeah, in terms of stability i thought p5 DBI is pretty stable and its probably less of an issue if p5 wasn't so stable...
don't need to be corrupting data or anything like that:)
the p5 integration, that is.
mysql here 14:49
janus jdv79: here's what i ment: blogs.perl.org/users/kaare/2010/06/...-code.html
jdv79: blogs.perl.org/users/martin_berends...bases.html might also be worth a read 14:50
jdv79 thanks 14:51
14:52 clinton joined
jnthn jdv79: The Perl 5 interop project is Blizkost. However, you may also be interested in FakeDBI. 14:52
jdv79 nice 14:54
whiteknight ah right, blizkost 14:55
I'm bad with names. Especially non-english ones
but *ESPECIALLY* english ones
14:58 JimmyZ joined
dalek tracwiki: v8 | mikehh++ | CFunctionDocs 15:00
tracwiki: trac.parrot.org/parrot/wiki/CFuncti...ction=diff
15:06 ash_ joined 15:17 ambs joined
dalek tracwiki: v6 | mikehh++ | MissingCFunctionDocs 15:34
tracwiki: trac.parrot.org/parrot/wiki/Missing...ction=diff
tracwiki: v4 | mikehh++ | BoilerplateOnlyDocs
tracwiki: trac.parrot.org/parrot/wiki/Boilerp...ction=diff
15:35 theory joined 15:43 patspam joined
cotto "The Liveness" would make a good name for a band. 15:44
ash_ i bet they would do a lot of live shows 15:47
dalek rrot: r47937 | khairul++ | branches/gsoc_instrument (5 files):
Refactored EventDispatcher + added GC events.
 16:00
16:05 patspam1 joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#34658), fulltest) at r47936 - Ubuntu 10.04 amd64 (g++ with --optimize) 16:06
pir/PITATE (63b457e ) - build ok / test PASS - parrot at r47936 - Ubuntu 10.04 amd64 (g++ with --optimize) 16:19
s/PITATE/PIRATE/
partcl-nqp (f9c3d4a) - make ok / make test PASS - on parrot at r47936 - Ubuntu 10.04 amd64 (g++ with --optimize) 16:24
16:34 whiteknight joined
dalek kudo: 8e7bf30 | pmichaud++ | src/core/IO.pm:
Add IO.open to allow attaching an existing IO object to a file. Also now
 16:39
kudo: 7df2c27 | pmichaud++ | src/ (2 files):
Refactor IO::ArgFiles.
jdv79 is make install DESTDIR=/some/local/path supposed to work? 16:57
moritz jdv79: no, perl Configure.pl --prefix=/some/local/path
jdv79 ok
Unknown option: prefix 17:00
sorry, i was talking about rakudo
moritz rakudo always installs into the parrot location 17:02
mikehh rakudo (806efc8) builds on parrot r47936 - make test PASS, spectest_smolder -> #34660 (pugs r31517) FAIL - Ubuntu 10.04 amd64 (g++ with --optimize)
t/spec/S03-metaops/reduce.rakudo - Failed tests: 89-92 - Parse errors: Bad plan. You planned 88 tests but ran 92
t/spec/S32-num/rat.rakudo - Failed test: 798
19 TODO PASSes in 5 files
jdv79 cool
17:06 patspam joined
Coke past question - I want to pastify this pirop: "exists $P1[$P2]" - any pointers on how to do the keyed syntax in a pirop? 17:25
er, PAST question.
darbelo pastify? 17:26
Coke write it using PAST instead of PIR. 17:29
currently, my code unconditionally assumes $P2 exists, which returns PMCNULL sometimes, which goes boom later. 17:30
er, assumes $P1[$P2] exists, that is.
so I'm trying to turn the keyed lookup into an Op (:type<if>, (check for exists), Var(keyed lookup), Op(throw error))' 17:31
I guess I could just do an explicit null check on the keyed lookup.
particle use pirate to generate past? 17:32
darbelo Didn't we add opcodes that did magic things on keyed lookup for rakudo some time ago? 17:33
One of those might do what you need.
particle vivify?
purl i think vivify is "get a pmc, if it doesn't exist, create one of this type and put it in the aggregate I just tried to fetch from"
particle pets purl 17:34
purl pets particle back, harder
darbelo I had fetch in mind, which was added at the same time.
=item B<fetch>(out PMC, in PMC, in PMC, in PMC) 17:35
Fetches a value from $2, keyed by $3 into $1. If the resulting PMC is PMCNULL,
uses the type in $4 to create and return a new PMC.
NotFound thinks about a DWIM op
atrodo Everything needs more DWIM 17:39
Coke particle: I don't /want/ it to vivify. i want to throw an error if I can't find the variable in the lexpad. 17:48
the <<lexpad>>, sorry, as we're faking it.
NotFound: you laugh, but we pulled that op about six years ago.
particle ah
Coke particle - but I want it to be /my/ error, not *VOMIT NULL ACCESS* 17:49
particle so you want if null $P2 goto MYERROR
atrodo Everything needs more *VOMIT* exceptions 17:51
Coke particle: yah. I was hoping for "exists", but I can get the keyed access and check that for null, I guess.
since it doesn't require keyed access, that'll do, I guess.
NotFound Coke: I think exists exists
particle INTVAL exists_keyed 17:52
it's a vtable function
Coke NotFound: not that I see in compilers/pct/src/PAST
17:52 davidfetter joined
NotFound =item B<exists>(out INT, invar PMC, in KEY) 17:52
Coke *FACEPALM*
yes. my original question is how to write a KEY in PAST. 17:53
and pass it to an Op ( :type<pirop>)
so I could use that very opcode. =-)
particle oh. 17:54
Coke tries the isnull approach. 17:57
17:58 theory joined
Coke wishes there was a more compact syntax for PAST in NQP. 17:58
(but only a little)
Coke needs to update the docs on PAST nodes. 18:04
(the difference between a call and a pirop is jarring based on an initial read of the docs.) 18:05
\\o/ 18:19
pmichaud++ 18:21
18:26 mikehh joined 18:30 clinton joined
cotto_work pmichaud: you around? 18:31
msg pmichaud How do you feel about my pull request to add ... to nqp-rx? 18:34
purl Message for pmichaud stored.
19:03 AndyA joined
dalek nxed: r526 | NotFound++ | trunk/examples/Mysql.winxed:
fix --host option in example Mysql
 19:13
19:16 LoganLK joined 19:21 joeri joined 19:39 theory joined 19:41 theory joined
cotto_work I decided to start working on the linear scan allocator in PIRATE's master but later decided that it'd be better to put it into a branch. Given that I haven't pushed anything, is there an easy way to move some commits to a branch? 20:03
20:04 Hunger joined
darbelo A merge? 20:04
purl a merge is the default in'it
atrodo cotto_work> git I assume?
cotto_work yes, git 20:05
I could get a clean clone and apply all the diffs by hand to a branch, but it strikes me that there'd be a better way.
atrodo There is 20:06
whiteknight cotto_work: I'm leaving now, but I've had that same exact question many times before
20:06 theory joined
whiteknight so if you get a good answer, msg me? 20:06
cotto_work will do 20:07
whiteknight cotto++
cotto_work I can't work on it right now, but a pointer to a solution for when I get home would be fine.
I'm sure our friendly neighborhood magical coding robot knows a thing or two.
darbelo cotto_work: How about "git branch newbranch; git reset --hard HEAD~<number of commits to bo back>; git checkout newbranch;" ? 20:08
atrodo That's what I was about to post 20:09
darbelo I haven't tried it, but I think it'll do what you want.
atrodo minus, I would do "git reset --mixed origin/master"
cotto_work I've run git pull since I started work, so some intervening commits are from bacek.
jdv79 FakeDBI segfaults for me 20:10
in Parrot_gc_mark_PMC_alive_fun 20:11
:(
cotto_work jdv79: do you have a repro?
jdv79 probabl not - i just copied from the test and ran it 20:12
i can reproduce it here...
cotto_work I mean a reproduction of the problem
my ambiguity
pmichaud phone 20:13
jdv79 its all over the place - gdb says Parrot_Exception_mark and Parrot_CallContext_mark too on subsequent runs 20:14
fun stuff
purl it has been said that fun stuff is rarely the stuff people want to pay you for.
jdv79 guess its not that stable 20:15
cotto_work I guess not.
20:15 Hunger joined, lucian joined
cotto_work If you can find a stable bug, we should be able to fix it. 20:15
20:22 ambs joined, pjcj joined
GeJ Bonjour everyone. 20:36
cotto_work hi GeJ
GeJ Servus cotto. 20:37
dalek nxed: r527 | NotFound++ | trunk/ (3 files):
experimental syntax for heredocs
 20:40
GeJ Ok, that's going to be a tricky one : under gcdebug runcore, I have a one line failing test (namely t/compilers/imcc/syn/regressions_18.pir) on FreeBSD 7. However, I seem to remember it passing on FreeBSD 8. 21:08
cotto_work you mentioned something about that 21:09
GeJ Since it seems to be a typical case of "Oh great, the GeJ box again" bug, I'd appreciate any hint on how I could narrow the problem so you brainiacs can eventually fix it. ;) 21:10
cotto_work It's more like "Excellent. GeJ has managed to find a new and creative way to expose something broken in Parrot." 21:11
GeJ It doesn't seem to coredump, I tried the step-by-step GC bug with gdb tutorial on Trac, it just fails and returns 1.
Coke how old is freebsd 7? 21:13
nopaste "GeJ" at 192.168.1.3 pasted "(gdb) t/compilers/imcc/syn/regressions_18.pir with gcdebug on FreeBSD 7 amd64" (18 lines) at nopaste.snit.ch/21722
Coke ohh, that smells. 21:14
GeJ Mine is from last august. Since it's my workstation and FreeBSD 3rd party software is mostly build-from-source, I didn't upgrade it yet. 21:15
FreeBSD kate.scsi.local 7.2-STABLE FreeBSD 7.2-STABLE #0: Tue Aug 18 10:31:03 NCT 2009 21:16
NotFound GeJ: a backtrace will be helpful
Coke if the problem really is with the max key len, that controls how big a literal key you can make.
(and I think that's currently hardcoded) 21:17
NotFound Coke: looks like this test is deliberately using a key too long.
cotto_work what NotFound said
GeJ so 'bt' ?
NotFound # more than MAX_KEY_LEN.
cotto_work It just tests that imcc doesn't segfault in that case
NotFound And is doing it well.
Coke ... so what's the problem then? =-) 21:18
GeJ I have 0 experience when it comes to gdb, I'm probably going to need some guiding. Sorry.
Coke his gdb run doesnt' show a segfault, does it?
NotFound Coke: it test tha parrot doesn't segfaults, but it segfaults.
GeJ (gdb) bt
No stack.
cotto_work I don't see it segfaulting in gej's no paste. 21:19
It's just showing imcc's error message. 21:20
NotFound Then I fail to understand what's the problem. 21:21
cotto_work Me too.
NotFound returning 1 isn't a problem, the test checks error output 21:22
cotto_work GeJ: can you nopaste the output of perl t/harness -G t/compilers/imcc/syn/regressions.t
Coke did we see the test failure?
cotto_work++
GeJ # perl t/harness -G t/compilers/imcc/syn/regressions.t 21:27
t/compilers/imcc/syn/regressions.t .. ok
All tests successful.
Files=1, Tests=21, 1 wallclock secs ( 0.02 usr 0.02 sys + 0.28 cusr 0.23 csys = 0.55 CPU)
purl Are you feeling lucky?
GeJ Result: PASS
Sorry, should have nopaste'd it. 21:28
cotto_work It appears that nothing's wrong. Is that the box where make testgcd failed?
GeJ yes 21:29
NotFound nqp isn't installed as language? 21:30
cotto_work Does the same thing happen with perl t/harness --gc-debug -G t/compilers/imcc/syn/regressions.t
nopaste "GeJ" at 192.168.1.3 pasted "Relevant excerpt from `gmake fulltest` exposing the failure of regressions.t" (28 lines) at nopaste.snit.ch/21723 21:32
cotto_work Now we're getting somewhere. That's definitely odd. 21:33
NotFound load_bytecode 'nqp-rx.pbc' .... compreg 'NQP-rx'.... Are you guys making it difficult on purpose? 21:37
cotto_work GeJ: what happens when you run t/compilers/imcc/syn/regressions_18.pir with either --runcore=gcdebug or --gc-debug ? 21:39
i.e. what's the exit status? 21:40
atrodo NotFound> Yea. I've had to manually move the nqp directory to a nqp-rx directory to make load_library work 21:41
NotFound atrodo: And the funny cases has some rason?
atrodo NotFound> No idea why it's that way, but I agree. load_library and compreg should both work and take the same string 21:44
NotFound Don't even talk about load_language
atrodo Oh, I did 21:45
21:45 tcurtis joined
nopaste "GeJ" at 192.168.1.3 pasted "Command line invokation of regressions_18.pir and Exit status (per cotto's request)" (12 lines) at nopaste.snit.ch/21724 21:48
cotto_work If it's easier, you can just tell me the exit code. 21:49
21:49 hercynium joined
GeJ NotFound: no nqp here except for the one bundled with Parrot. 21:49
cotto_work What about when you use both options, and both plus -D40 ? 21:50
GeJ same output, same exit code. 21:51
darbelo cotto_work: Looks to me like the harness isn't correctly matching his output.
cotto_work The test harness complains that the exit code is 0.
GeJ: thanks.
21:51 kid51 joined
NotFound $status? What shell is that? 21:51
kid51 Live .. from Bowling Green ... where there's pretty decent wireless in the park ... 21:52
GeJ NotFound: "$?" under C-shells... don't ask ;) 21:53
ash_ anyway here know libffi at all? 21:57
or want to see a fun segfault
darbelo ash_: nopaste it and see if anyone laughs?
That's my usual metric for fun ;) 21:58
kid51 is posting Smolders of trunk, gsoc_past_optimization and gsoc_nfg (Linux/i386).
ash_ darbelo: the segmentation fault is just Segmentation fault (when used with an NCI callback) 21:59
darbelo kid51++ # Even if I doubt gsoc_nfg isn't in a very smolderable state :) 22:00
kid51 darbelo: It passed 'make test' ... but you'll need to run 'make codetest' and fix several files. 22:02
darbelo Yeah, I know about those. I'll tackle them once the API is stable enough to be worth documenting. 22:03
I also haven't annotated most function params for the same reason. I change my mind about what shoud and shouldn't be NULL quite often in some places. 22:04
ash_ i don't like when gdb backtraces say: #0 0x000000010161fa20 in ?? () 22:06
kid51 gsoc_past_optimization passes 'make test' and 'make codetest'
22:09 whiteknight joined
bacek Good morning, humans 22:09
cotto_work hey bacek 22:10
bacek aloha, cotto
cotto_work I have a git question that you may be able to answer. 22:11
I decided to start working on the linear scan allocator in PIRATE's master but later decided that it'd be better to put it into a branch. Given that I haven't pushed anything, is there an easy way to move some commits to a branch?
bacek git rebase origin master 22:13
giti checkout -b the_branch
git checkout master
git reset --hard HEAD~<number of your commits>
something like this
cotto_work Will that work if I've run git pull since I started work on the branch? 22:14
darbelo That's what the rebase is for.
bacek hm. Are you on branch now?
cotto_work no 22:15
bacek then what darbelo said :)
cotto_work Great. I'll take a shot at that when I get home.
bacek++
bacek git++ :)
cotto_work and darbelo++ just because
yeah. git++
and svn--
darbelo svn-- #out of habit
dalek rrot: r47938 | jkeenan++ | branches/gsoc_nfg/src/packfile/pf_items.c:
[codingstd] No trailing whitespace.
 22:19
GeJ G'Day bacek 22:22
bacek G'Day GeJ 22:23
darbelo Ugh. I hate whitespace conflicts. 22:24
cotto_work lame they are 22:30
22:33 hercynium joined 22:37 hercynium_ joined
ash_ yes, i found my bug, now i can get callbacks working and all of the nci old stuff should be working fine 22:55
23:22 plobsing_work joined
whiteknight ash_: nice! 23:28
plobsing_work ping ash_ 23:32
darbelo Boy am I going to meet my 'hate packfiles' goal this week. 23:41
cotto_work Just think how much easier it'll be to refactor them when PIRATE takes over the world. 23:42
"somewhat" 23:43
darbelo To be fair. Any code that tries to replace them, will be of similar complexity, having amd64 execute vax bytecode is just messy. 23:44
cotto_work The PIRATE pbc code is fairly self-contained though. You just have to mess with one file. 23:45
darbelo Heh. I've stayed in one file. src/packfile/pf_items.c 23:46
I had to change the way we store strings in pbc. 23:47
23:48 jsut joined
cotto_work That code makes my eyes glaze over after about 3 seconds. 23:48
darbelo You'll want to skip review on my next few commits then :) 23:50
cotto_work Are you using git-svn? 23:52
darbelo Yep. 23:53
cotto_work How is it?
darbelo Not as good as git.
23:54 Psyche^ joined
darbelo But offline operation is a big win for me. 23:54
sorear darbelo: What's the most hateful thing about pack files?
darbelo The api. Hands down. 23:55
Use the PMCs if you can.
tcurtis How fun. --trace=1 segfaults on the code I'm debugging. 23:57
darbelo sorear: I also have some gripes about the code and the conventions it uses, but those are linked to the API. 23:58
OTOH, as I mentioned to cotto, any code that tries to do everything that our packfiles try to do is going to be of that same complexity, so there isn't too much that can be done on that area. 23:59
Unless you want to move to a less featureful packfile format, that is.

Powered by Cro and the Raku® Programming Language.

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

Thank you!