#parrot Parrot 2.1.1 Released! | parrot.org/ | Channel log: irclog.perlgeek.de/parrot/today | Tasks: PCC deprecations hackathon on Saturday, TT #389 branch
Set by moderator on 5 March 2010.
00:07 zpmorgan joined
theory dukeleto: ping 00:19
00:24 kid51 joined
plobsing aside from the fact that it's not causing any bugs ATM, is there a reason Complex should have an invoke vtable? 00:41
00:48 slavorg joined 01:01 abqar joined 01:02 Whiteknight joined
Whiteknight oi, the infinite exception loop is back 01:13
damn fart crap
plobsing Whiteknight: working on TT #1499, I got an idea to solve TT #1015 using the cleaned up visit system. 01:14
do you have any plans in the works for that one?
Whiteknight let me look those up
ah, ok 01:16
I tried a solution a while ago that used a cache to avoid cyclic data structures in clone 01:17
a similar mechaism would be nice for freeze/thaw too 01:18
plobsing freeze/thaw already does that
* and has for a while
Whiteknight maybe I'm mis-remembering the details
plobsing and now that it has been cleaned up, you can plug other PMCs into the same code to execute different actions (eg: deep clone) 01:19
Whiteknight nice
plobsing at least that's the theory. It only has freeze/thaw and "get the size of freezing without actually doing it" 01:20
for now 01:21
I'm thinking a cloner would be a good test of that infrastructure
Whiteknight give it a try 01:22
I would love to see that
01:26 snarkyboojum joined
Whiteknight let me know if you need any help 01:29
dalek rrot: r44751 | petdance++ | trunk/config/gen/makefiles/root.in:
Change "ex:" to "example:" because vim was getting confused
 01:34
purl dalek: that doesn't look right
rrot: r44752 | plobsing++ | branches/tt1015:
creating branch to attempt to solve TT #1015 using recently cleaned visit system
 01:50
rrot: r44753 | plobsing++ | trunk (2 files):
freeze/thaw for Complex
dukeleto theory: pong 01:56
plobsing++ # freeze/thaw for complex 02:00
02:16 hudnix joined 03:09 adu joined
kid51 It appears that the position of two config steps was changed when the rm_cflags branch was merged in. 03:24
I'm getting a warning during Configure.pl on my Darwin/PPC -- but not on Linux/i386. 03:25
auto::arch - Determine CPU architecture and OS...Use of uninitialized value in pattern match (m//) at config/auto/arch.pm line 63.
...............done.
Will look into this.
Coke ping 03:27
It's late; I'll open a ticket. 03:28
theory dukeleto: back 03:41
kid51 Anyone building Parrot on Darwin/i386, please have a look at this new ticket: trac.parrot.org/parrot/ticket/1504 03:44
kid51 must sleep
purl $kid51->sleep(8 * 3600);
dalek TT #1504 created by jkeenan++: config/auto/arch.pm: Uninitialized value warning on Darwin due to change ... 03:46
03:52 snarkyboojum left 03:54 janus joined 04:02 snarkyboojum joined 04:45 chromatic joined 04:57 tuxdna joined 04:58 parthm joined 04:59 Austin joined
Austin Buenos tardes, #parrot. 05:00
05:22 parthm left
dukeleto theory: repong 06:28
theory heh
dukeleto: What is it you're trying to do, exactly? 06:29
dukeleto theory: when i run the tests, i get lots of noise in my log about "creating implicit index". i don't want that noise in my logs
06:30 payload joined
theory dukeleto: I know. 06:30
You shouldn't get that in your logs.
oh, wait
hang on a sec
dukeleto theory: lets move this to #plparrot
that is the 4th time someone has told me that today. I want to not implicitly create indices. How do I *not* implicity create indices? 06:33
oops
dukeleto facepalms
chromatic /mesg dukeleto Create them explicitly?
heh
dukeleto hmmm. trunk is a bit wonky on darwin x86 06:43
looks like my 'parrot --version' says 'ppc-darwin' now 06:44
dukeleto is running a "make smoke" now
07:03 uniejo joined 07:19 JimmyZ joined 07:31 bacek joined
dukeleto yep, parrot --version is wrong on darwin-x86 now . see trac.parrot.org/parrot/ticket/1504 for details 07:31
bacek: o hai
dalek parrot: 602f52e | dukeleto++ | pgtap.sql:
Get rid of log noise about creating implicit sequences/indices
 07:45
07:57 eiro joined
dukeleto theory++ # help with implicit indices/sequences 08:09
08:19 payload joined 08:26 fperrad joined 08:30 bacek joined 08:45 iblechbot joined
bacek ~~ 08:56
09:24 AndyA joined 09:25 payload joined 09:32 aukjan joined 09:53 TonyC joined
dalek kudo: f56934d | moritz++ | t/spectest.data:
we now pass S09-typed-arrays/hashes.t again, arnsholt++
 10:00
10:02 barney joined
dalek kudo: b1fd5d8 | moritz++ | src/builtins/Parcel.pir:
Fix RT #73128, stringifications of Parcel

the one defined in Mu. Another vtable override fixes that.		 10:06
kudo: df743fe | moritz++ | t/spectest.data:
run the new parcel.t file
 10:12
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32549), fulltest) at r44753 - Ubuntu 9.10 amd64 (g++ with --optimize)All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32549), fulltest) at r44753 - Ubuntu 9.10 amd64 (g++ with --optimize) 10:15
bah - try again 10:16
All tests PASS (pre/post-config, make corevm/make coretest, smoke (#32549), fulltest) at r44753 - Ubuntu 9.10 amd64 (g++ with --optimize)
10:29 gaz joined
dalek rrot: r44754 | mikehh++ | branches/ops_pct/compilers (11 files):
add svn properties
 10:46
rrot: r44755 | mikehh++ | branches/ops_pct (25 files):
add svn properties
 11:19
11:25 TiMBuS joined 11:47 ruoso joined 11:50 allison joined
dalek kudo: 393fcd4 | moritz++ | src/ (4 files):
fix stringification of Any, and thus RT #73410
 12:06
kudo: bd7c6f5 | moritz++ | src/core/Hash.pm:
implement method Bool in Hash
12:15 eiro joined 12:27 tuxdna joined 12:33 Austin joined 13:01 tetragon joined 13:05 aukjan joined
Coke skips review 13:10
msg kid51 - the problem with the configure steps is that they have steps that are prereqs, but since this thisn't a makefile, we can't easily track what those deps are. I'll see if I can come up with a way to fix it. 13:11
purl Message for kid51 stored.
Coke msg kid51 ah, just saw your response on the ticket... commenting there. 13:12
purl Message for kid51 stored.
13:43 bluescreen joined 14:02 atrodo joined 14:03 allison joined 14:08 GeJ joined, magnachef joined 14:16 whiteknight joined
whiteknight hello #parrot! 14:30
Coke ~~ 14:35
14:36 smash joined
smash hello everyone 14:36
whiteknight hello smash, Coke 14:45
14:47 jsut joined
Coke blogs.perl.org/users/robert_phaylon...recap.html //parrot 14:48
14:57 plobsing joined 15:10 Andy joined
dalek kudo: d183ebe | moritz++ | t/spectest.data:
enable context-forcers.t
 15:15
15:21 zostay joined
dalek rrot: r44756 | plobsing++ | branches/tt1015/src (3 files):
first cut of cloning visitor

functionality through clone than through freeze, or flat out don't provide freeze. Bad PMCs!		 15:23
15:24 PacoLinux joined 15:25 iblechbot joined
dalek kudo: 54c983e | jonathan++ | src/ (3 files):
Start to be a tiny bit smarter about :auth and :ver - now we treat them as not just part of the name, but also something you can introspect on the meta-class.
 15:26
kudo: 8f90546 | jonathan++ | (2 files):
Translate much of moritz++'s prototype implementation of module location in NQP. Untested and unused so far.
15:31 payload joined 15:34 Psyche^ joined 15:40 theory joined
Coke Andy: ping. 15:40
Andy ? 15:41
Coke hio. you have branches/depends that has a first pass on doing makedepends - I think we're better off dumping that and using the checkdepend.pl tool for that (it's a lot less naive, though it's not perfect.) 15:43
wondering if you would be sad if I deleted it.
Andy no, toss it. 15:44
Coke danke. 15:46
dalek tracwiki: v30 | coke++ | BranchDescriptions
tracwiki: trac.parrot.org/parrot/wiki/BranchD...ction=diff
Coke Andy++ 15:49
Andy Right now I'm working on tidyp a lot, and I'm stealing a lot of the splint stuff I did on Parrot. 15:50
Coke Andy: that reminds me, I have a question for you. =-) 15:51
Andy yes?
Coke I just redid the warnings probes as part of a bid to eliminate the perl script that was doing our compiling for us. Was looking at moving some warnings flags out of cage into the standard set. 15:52
one of them seems reasonable but fails on every ASSERT_ARGS.
-Wunreachable-code
dalek rrot: r44757 | coke++ | branches/depends:
Removing old branch -

instead of just complaining about missing ones.		 15:56
Coke er, so the question is - any idea how to make those assertions work on that warning flag? 16:00
dalek tracwiki: v31 | coke++ | BranchDescriptions 16:03
tracwiki: trac.parrot.org/parrot/wiki/BranchD...ction=diff
Coke msg Austin your yahoo address is generating temp bounce messages to the tickets list. 16:10
purl Message for austin stored.
Coke delivery temporarily suspended... delivery temporarily suspended due to user complaints 16:11
er, temporarily deferred due ...
(cut and pasto)
16:14 patspam joined 16:18 davidfetter joined
Coke is the only way to create a new wiki page in trac to link to it? 16:21
ISTR that I used to be able to do a search for FooBar and have one of the results on "no matches found" be "create this page". 16:23
particle coke: there's likely a config setting for that behavior 16:30
Coke particle: global or personal. (and I checked in both places already, saw nothing likely)
trac.edgewall.org/wiki/WikiNewPage seems to indicate it should JFW. 16:33
trac-- # yet another case where I have to fight the tool to get what I want.
particle trac.parrot.org/parrot/wiki/foobar 16:34
Coke yes, you can manually jump to the page by hand editing the url. 16:35
particle it seems you can type in the url you want
Coke I also just found that if you search for [wiki:SomeNewPage] that also works.
(gives you the quick jump link)
dalek kudo: e533f86 | moritz++ | t/spectest.data:
three more test files that we pass now that parcels stringify correctly
 16:41
16:45 atrodo_ joined
Coke particle: wonder if it the "ignore_missing_pages" option in trac.ini is related. 16:46
particle ah, yes, that might be it
i'm no wiki expert, for sure, but i do recall setting something in trac for some site in the past... so vague it's not useful. :( 16:47
Coke looks like it just affects markup on links pointing to missing pages. 16:48
16:49 Psyche^ joined
Coke now I'm back to the trac-- behavior of not saving my changes when I click on the save button, but having to click the login link again. 16:52
dalek tracwiki: v1 | coke++ | tasksCoke
Coke that then saves my changes and brings me back to the start page of the wiki.
dalek tracwiki: use tasks first to get auto-indexing of page in tasks category
tracwiki: trac.parrot.org/parrot/wiki/tasksCo...ction=diff
Coke fun.
Are we using mod_python, particle, do you know? 16:54
particle i do not know 16:58
dalek kudo: 49749c0 | masak++ | src/core/Signature.pm:
[Signature.pm] removed extra '&' for params in .perl
 17:04
tracwiki: v2 | coke++ | tasksCoke 17:09
tracwiki: use trac's wiki formatting.
tracwiki: trac.parrot.org/parrot/wiki/tasksCo...ction=diff
Coke opened a ticket with OSU on my trac login issue. mebbe they can help. 17:10
17:11 bubaflub joined 17:17 atrodo joined 17:22 atrodo_ joined 17:23 payload joined 17:25 atrodo joined 17:42 atrodo joined
dalek kudo: 9978b34 | jonathan++ | src/ (4 files):
Need to set up @*INC by BEGIN time, othrewise we'll not be able to use it in module loading. Kinda switch on new module locator code; gets us looking through @*INC again and should handle loading Foo::Bar style things.
 17:44
tracwiki: v2 | basic++ | OSCON2008%20Talk%20Proposals 17:45
tracwiki: trac.parrot.org/parrot/wiki/OSCON20...ction=diff
tracwiki: v3 | basic++ | OSCON2008%20Talk%20Proposals
tracwiki: trac.parrot.org/parrot/wiki/OSCON20...ction=diff
moritz I have a .sub '' :vtable('get_string') :method and when I add an :anon to it, it doesn't work anymore 17:47
does anyone want to help me to write a pir-only test for that? I only experienced it in rakudo
already managed it... 17:56
17:58 lucian joined
dalek TT #1505 created by moritz++: :anon vtable overrides don't work in PIR 17:59
18:01 kurahaupo joined 18:05 payload joined 18:15 TiMBuS joined
dalek rrot: r44758 | plobsing++ | branches/tt1015/src/pmc/visitclone.pmc:
slight reordering of operations
 18:32
18:43 bubaflub_ joined
dalek rrot: r44759 | plobsing++ | branches/tt1015 (2 files):
add visit,freeze,thaw,thawfinish to ArrayIterator and modify a test to keep failing
 18:49
rrot: r44760 | jonathan++ | trunk/src/pmc/os.pmc:
[os.pmc] Add Win32 implementation for readdir (was POSIX only up to now).
kudo: e4e1a0d | (Jaume Martí)++ | src/Perl6/Grammar.pm:
Rewrite of parse_name in NQP
 18:52
kudo: e66c48d | moritz++ | src/ (4 files):
Merge branch 'master' of github.com:rakudo/rakudo
kudo: 086bed5 | moritz++ | src/Perl6/Compiler.pir:
fix using Test.pm by putting the . as the first entry in @*INC

than not running any tests when there's a Test.pm installed
18:58 jsut_ joined 18:59 AndyA joined, darbelo joined 19:02 chromatic joined 19:03 AndyA_ joined
dalek tracwiki: v3 | coke++ | tasksCoke 19:08
tracwiki: Testing login issue...
tracwiki: trac.parrot.org/parrot/wiki/tasksCo...ction=diff
tracwiki: v4 | coke++ | tasksCoke
tracwiki: trac.parrot.org/parrot/wiki/tasksCo...ction=diff
tracwiki: v5 | coke++ | tasksCoke
tracwiki: trac.parrot.org/parrot/wiki/tasksCo...ction=diff
tracwiki: v6 | coke++ | tasksCoke
tracwiki: trac.parrot.org/parrot/wiki/tasksCo...ction=diff
19:10 hercynium joined
Coke hey, trac-- is still borked. =-) 19:10
dalek rrot: r44761 | plobsing++ | branches/tt1015/src/pmc/fixedfloatarray.pmc:
freeze/thaw for FixedFloatArray (propagates to RFA)
 19:22
rrot: r44762 | plobsing++ | branches/tt1015/src/pmc/fixedintegerarray.pmc:
remove dead code and doc
rrot: r44763 | plobsing++ | branches/tt1015/src/pmc/fixedstringarray.pmc:
handle thawing of uninitialized FSAs (propagates to RSA)
19:22 bacek joined 19:32 joeri joined 19:37 whiteknight joined 19:41 ash_ joined 19:45 bubaflub joined
Austin_away Hello, # parrot. 19:47
Austin Hello, #parrot.
19:47 iblechbot joined
darbelo Hi, Austin. 19:47
Austin Man, this mocking business is taking way longer than planned. 19:48
19:49 kjeldahl_ joined
Austin message Coke I have no idea why. I'm not subscribed, so I assume trac is sending to me because they're my tickets, but what does the list have to do with that? 19:49
purl Message for coke stored.
19:50 bubaflub joined
whiteknight hello Austin 19:50
19:50 allison joined
Austin howdy, whiteknight, darbelo. 19:51
darbelo Austin: Mocking objects? Lot's of material there...
Austin whiteknight: www.tk421.net/lotr/film/fotr/img/fotr0165.jpg
darbelo: Yeah. But it's always a little bit harder than it should be. 19:52
whiteknight: How'd your weekend finally go? Get the chimney sorted out?
whiteknight Austin: yeah. we got the chimney out, and not an hour later it was picked up by people searching the neighborhood for scrap metal 19:53
Austin Yeah. I've witnessed the amazing efficiency of the scrap collectors here in NJ.
whiteknight we also got most of the cabinets installed, without damaging them too much and without inventing too many cursewords 19:54
Austin I put some very, very, VERY old computer equipment out one trash day, and before I got to my front door -- all of about 35 feet from the curb -- there was a pickup truck parked on the curb.
whiteknight and we've discovered that my son really really hates the screetching sound of driving a 3" drywall screw into a 2x4
Austin Heh
dalek rrot: r44764 | plobsing++ | branches/tt1015/src/pmc/nci.pmc:
add freeze/thaw for NCI so it can be cloned by VisitClone
 19:55
whiteknight plobsing++ is kicking ass and taking names
Austin Well, good for him. What's he doing?
whiteknight he was fixing freeze/thaw, and along the way found a great way to fix the problem with cyclic references in deep clones 19:56
Austin Rock on.
purl The rock is jammed at the 'puree' setting.
whiteknight so he started that tt1015 branch to fix that very issue
bacek morning... good morning
dalek kudo: f0a5552 | (Solomon Foster)++ | src/ (2 files):
Remove obsolete Str.z and Str.filebytes functions. Add Str.d, Str.f, and Str.s functions.
Austin That bug's been there for too long.
kudo: d928a86 | (Solomon Foster)++ | src/Perl6/ (2 files):
Merge branch 'master' of git@github.com:rakudo/rakudo
kudo: 45b806a | (Solomon Foster)++ | src/builtins/Str.pir:
Port :l from alpha.
Austin Good morning, Bacek
Coke huh. gmail thought that last commit message from bacek was malay and offered to translate it for me. 20:02
dalek kudo: ec9b0df | (Solomon Foster)++ | src/builtins/Str.pir:
Tweaks to :l to make it work (more or less).
Austin Maybe Bacek's working on his 5th language?
bacek Coke, it was actually in klingon
Coke yIDoghQo' 20:03
20:05 davidfetter joined
Coke who is doing the aix smokes. 20:09
?
whiteknight no idea
darbelo Do we even build on aix? 20:10
whiteknight NotFound: ping
Coke darbelo: 12114 test cases: 11972 ok, 142 failed, 266 todo, 657 skipped and 0 unexpectedly succeeded, tags: "2.1.0", "Perl 5.10.0 ppc-aix-thread-multi", "aix", "cc_r (gcc 4.2)", "ppc32" 20:11
Add starLikeShareShare with noteEmailKeep unreadEdit tags: Programming Alerts
ignore that last line.
70.42.42.150/app/public_projects/sm...port/32551
dalek rrot: r44765 | bacek++ | branches/ops_pct/compilers/opsc/t/05-opsfile.t:
Update test.
rrot: r44766 | bacek++ | branches/ops_pct/compilers/opsc/t/04-oplib_BUILD.t:
Remove outdated test
rrot: r44767 | bacek++ | branches/ops_pct/compilers/opsc (3 files):
Readd Ops::OpLib
rrot: r44768 | bacek++ | branches/ops_pct/compilers/opsc/t (0 files):
Rename test files.
rrot: r44769 | bacek++ | branches/ops_pct/compilers/opsc/t/05-oplib.t:
Update tests.
rrot: r44770 | bacek++ | branches/ops_pct/compilers/opsc/ops/oplib.pm:
Remove old oplib.
rrot: r44771 | NotFound++ | trunk/src/pmc/parrotthread.pmc:
set manual_attrs flag in ParrotThread PMC, TT #1501
rrot: r44772 | bacek++ | branches/ops_pct/compilers/opsc/src/Ops/Compiler (2 files):
Don't put {} into op body.
rrot: r44773 | bacek++ | branches/ops_pct/compilers/opsc/t/03-past.t:
Update test
rrot: r44774 | bacek++ | branches/ops_pct/compilers/opsc/runcore (6 files):
Remove old stub code
TT #1501 closed by NotFound++: parrotthread.pmc - PMC has attributes but no auto_attrs or manual_attrs
20:14 ash_ joined
NotFound whiteknight: pong 20:17
whiteknight NotFound: (Re TT #1501) can we write a test somewhere that checks all pmclass declarations for a proper *_attrs flag? 20:18
NotFound whiteknight: I think is better to wait until we document as mandatory to have one or the other. 20:20
I must create a ticket to trac the evolution of this thing, now is disperserd in #ps logs 20:22
whiteknight ok
dalek rrot: r44775 | bacek++ | branches/ops_pct/compilers/opsc (2 files):
Don't parse ops in OpLib
 20:28
rrot: r44776 | bacek++ | branches/ops_pct/compilers/opsc (3 files):
Skip ops in compiler
Coke where is PARROT_ASSERT_ARG defined? 20:40
dammit. nevermind. =-)
ok. I see no way to make ASSERT_ARGS compatible with -Wunreachable-code 20:42
dalek TT #1506 created by NotFound++: PMC attribute allocation flags design and evolution 20:45
20:49 rurban joined
rurban So I have now my jitted (char*)code, which can be called via (*((int (*)())code))(), but I cannot disassemble it in gdb... "No function contains specified address" 20:50
chromatic You probably need to emit Dwarf information or convince gdb to load debug information from a separate file. If you write it out as a .o file, you can use add-symbol-file, I hear. 20:52
rurban That's parrot. I just wrote a perl5 jitter :) without any dwarf info yet 20:53
I want to fool gdb that (char*)code is a (void)code() function. Looks hard 20:54
20:56 ash_ joined
chromatic I don't know of any other possibilities to give gdb the right information. 20:57
rurban Thanks. 20:58
20:58 rurban left
redbrain thats pretty cool did you have to make your own runtime assembler? 21:16
whiteknight Coke: I wouldn't be unhappy if ASSERT_ARGS disappeared 21:18
tewk ruban disasemble code code+200 21:27
Austin rurban ^^
21:28 Essobi joined
tewk whois ruban 21:29
ruban?
purl ruban are you attempting to use ncigen or just build it?
tewk fonseka@gmail.com
?
emailing
Austin ru - R - ban
tewk duh
chromatic rurban@x-ray.at
Coke rurban? 21:31
purl rurban is on Win XP sp2
Coke no, rurban is rurban@x-ray.at, Reini Urban.
purl okay, Coke.
Austin ruban was a dutch pornographer in the 1600s who specialized in softcore chubbys.
Or something.
chromatic Rubens? 21:32
Austin Your way is good, too.
Coke I think I saw some of those pictures while playing Assassin's Creed 2. 21:33
chromatic Like all good hackers, I know a bit about painting.
Austin or pr0n
Coke: I guess they don't have to pay licensing fees, with him being dead and all, but is AC2 set in a museum or something? 21:34
21:34 hercynium joined
chromatic Only if that museum has an always-on Internet connection. 21:34
Coke Austin: set in the 1600s.
Austin Ah
chromatic: Don't most museums do the internet cafe thing, with $9.00 lattes? 21:35
Coke wonders if chromatic would like AC21. 21:36
er, AC2.
chromatic I enjoyed parts of AC, at least watching someone else play the fun bits. Then again, I enjoyed watching someone else play the fun bits of Mirror's Edge. 21:37
I liked System Shock and Thief.
Austin sings, "When I sing this tune, it don't upset her. She's half your age, and twice as hot!" 21:38
Coke ^_O
any win32 that can explain manifests to me? 22:04
win32 *devs*
lucian i'm curious if there's libffi support exposed to language developers 22:05
22:05 cotto_work joined
chromatic lucian, I believe not. 22:06
dalek rrot: r44777 | allison++ | branches/pcc_hackathon_6Mar10 (5 files):
[pcc] Correctly pass return results from generated NCI methods in PMCs. Add

argument.
purl argument is that open SMTP ports end up getting blocked, no matter what port number they're on. Do you think ISPs have this thing about the number 25? If SMTP becomes available on port N, then spammers will use port N, and port N will end up blocked.
lucian chromatic: so the only way to call C code is to write wrapper C code?
chromatic You can use the NCI system, but it doesn't use libffi. Maybe I'm misunderstanding your question. 22:07
Coke purl, no, argument is <reply>This is abuse.
purl okay, Coke.
lucian chromatic: can you call an arbitrary C library from a language implemented in parrot? 22:08
chromatic: i.e. python's ctypes
Tene lucian: I'm not familiar with ctypes. You can load a dynamic library and call functions from it, yes. 22:09
chromatic Yes, similar to ctypes. 22:10
lucian right, ok
allison lucian: aye, when the Python implementation on Parrot supports ctypes, it'll do it via NCI 22:11
lucian allison: let's have it first support classes :)
lucian is sad that pynie has a long way to go. and has so little free time to help ... 22:12
cotto_work ops_pct needs a manifest update if anyone wants easy karma 22:13
allison lucian: they're nearly done, parsing is there, just need to generate the right AST
lucian allison: oh, then I'm behind on the updates. are you a core dev?
allison lucian: but, yes, wish I could work on it full-time for a month or so
lucian: aye
lucian: I was actually quite pleased when I did a feature review for my PyCon talk last month 22:14
lucian allison: where's the repo these days actually?
Tene I very much wish I could work on anything parrot-related full-time for a month.
allison lucian: we're only about 5 core features shy of full pure-Python support
lucian searches that pycon talk
allison Tene: that too :)
lucian: and I added dictionaries during the week of PyCon 22:15
Tene I plan to sit down and have a very stern talk with the HLL interop API tonight.
lucian allison: this? pycon.blip.tv/file/3263792/
chromatic What about the subroutine leave semantics, Tene? 22:16
allison lucian: that's it
Tene chromatic: I haven't looked at it at all yet.
If I can get HLL interop working sanely in Rakudo, I'll probably start on pynie next. 22:17
allison: where is the pynie repo these days? 22:18
code.google.com/p/pynie/ ?
I saw that there was an hg repo for it, but it hasn't seen updates in months.
allison Tene: yes 22:20
the hg repo is a rewrite of the grammar in nqp-rx
which will be merged back in
lucian what's the status of the jit plan? 22:23
allison: i may have some time to contribute, what's the repo i should look at?
allison lucian: look at the googlecode repository 22:24
lucian allison: svn one? ok 22:25
allison lucian: aye
lucian sneers at svn
allison lucian: I should have create the hg repo as a clone, but created it as a raw prototype instead
lucian: the repo will be moving to hg at some point
lucian: at least by the time py3k does :) 22:26
lucian allison: good :) it's not a big deal, i use hg or git with svn anyway
allison isn't really a SCM purist, more of a pragmatist 22:28
japhb mightily fights the urge to tease allison at this point. ;-)
allison fights the urge to sing Kumbaya 22:29
allison also wonders exactly when it was that SCM wars overtook editor wars 22:30
lucian allison: i'm not a purist either, any DVCS is fine with me
allison it was sometime in the past 2 years
lucian hg and git are so extremely similar, the whole debate is moot 22:31
bzr is a bit different, but only because it strives to be lossless 22:32
allison I will profess to a preference for hg over git, but am perfectly happy with svn
lucian svn annoys me because it's not local
allison cvs was a pain, perforce was okay
japhb Interesting that at the same time people began to switch from central to distributes VCS systems, they also switched from arguing about things that affect individuals to things that affect teams. Growing network/global awareness, maybe?
lucian i'm too young to have used CVS, but when i read that it didn't have atomic commits i was horrified 22:33
Tene We had another SVN issue recently. coke spent several hours trying to merge rm_cflags with svn, and failed. I had to merge it for him with git.
allison <shrug> my brain doesn't work when I'm not connected to a network, doesn't help me much for my source control to work when my brain doesn't
allison thankful for airplanes with network access
lucian allison: it's not just that, your server can fail. you can also not push all your local commits 22:34
and it's often faster
Tene lucian: She's heard it all before, trust me. :)
allison aye :)
lucian Tene: ok, i'll shut up
allison Tene: good that you were able to recover the branch with git 22:35
japhb Tene: I just saw in the git 1.7.0 changelogs that git svn's awareness of SVN merge tickets is improved (though I don't know the details). So anyone using git svn might want to get current.
lucian shakes fist at ISP and blip.tv 22:36
japhb 's children are watching classic tunes: "Mary had a little lamb ... BUT I ATE IT!" 22:37
er, toons 22:38
plobsing allison: I recently noticed Complex.invoke, which has a comment indicating it is for python. My python interpreter seems dissagree. Should Complex have an invoke vtable? 22:39
chromatic Hm. How did that get in there?
allison plobsing: urg, no 22:40
plobsing deprecation cycle?
purl i heard deprecation cycle was fine.
allison plobsing: probably a hang-over from the pie-contest implementation of Python
chromatic Cut it and see if any tests fail. 22:42
cotto_work That function needs to go away. 22:44
22:48 Whiteknight joined
chromatic allison, any small tasks for the interested on the hackathon branch? 22:51
allison chromatic: sure
chromatic: search for any calls to *_fill_returns_* or fill_results 22:52
chromatic Okay.
22:54 joeri left
chromatic Lots of them in the NCI thunks. 22:54
Whiteknight probably corresponds to a single line in the appropriate .pl file 22:55
plobsing those are generated in tools/dev/nci_thunk_gen.pir
chromatic What do they change to?
allison chromatic: I just updated PCCMETHOD, haven't updated the thunk generator 22:56
chromatic: should be able to copy the change from PCCMETHOD
chromatic Will do.
Whiteknight allison: two questions: (1) did that infinite exception loop get fixed, and (2) is "PCCMETHOD" still the preferred way to call a method in a PMC? 22:57
allison the infinite exception loop is fixed
Whiteknight w00t
allison have a different error in PGE now, but at least it's not as frantic
Whiteknight: what do you mean by "PCCMETHOD" 22:58
suggesting renaming the perl module that generates those?
the source now just says "METHOD" 22:59
23:01 rurban joined
rurban tewk++ disassemble code code+200 works perfectly, the jitting error was immediately visible 23:01
tewk rurban, great 23:02
Whiteknight allison: I'm talkig about the (rare) use of "PCCMETHOD" in the *.pmc files to call a method
23:02 rurban left 23:03 atrodo joined
allison do we have any l33t DVCS users who'd like to take a stab at integrating Pynie's parrot SVN history with its googlecode svn history into one hg/git repo? 23:03
chromatic Hm, we can't rebuild the thunk generator without a working PGE. 23:04
Whiteknight ah, nevermind. It's "PCCINVOKE"
great, chicken and egg
allison Whiteknight: ah, I just wondered that (PCCINVOKE) 23:05
Austin allison: code.google.com/p/support/wiki/ConvertingSvnToHg
allison austin: yah, got the conversion, but the googlecode repo started raw, chopping off all revision history from the parrot repo 23:06
Austin: not that the revision history before is particularly critical, but seems worth preserving if possible 23:07
Austin How much google-svn history is there?
Whiteknight why the pynie move? 23:10
allison Whiteknight: Parrot_PCCINVOKE should be replaced by Parrot_pcc_invoke_method_from_c_args
Whiteknight allison: In the *.pmc files, it's just the bare word "PCCINVOKE"
pmc2c translates it to Parrot_pcc_invoke_* 23:11
allison Whiteknight: that's substituted inside Pmc2c
aye
Whiteknight so that keyword is still good practce?
dalek rrot: r44778 | plobsing++ | trunk/src/pmc/os.pmc:
fix trailing whitespace
 23:12
allison Whiteknight: I think we can pick a better name
rrot: r44779 | plobsing++ | trunk/src/pmc/complex.pmc:
remove Complex.invoke
rrot: r44780 | chromatic++ | branches/pcc_hackathon_6Mar10 (3 files):
[NCI] Updated NCI to use Parrot_pcc_build_call_from_c_args() instead of
allison Austin: not sure, probably not more than 100 revisions or so
Austin: it's all interleaved with other parrot commits, so some work to filter out numbers 23:13
23:14 ash_ left
allison Tene/lucian: bitbucket.org/allison/pynie/ is now a clone of the googlecode repo, feel free to work from there instead of the SVN repo 23:18
lucian allison: great, i'll have a look around. i doubt i can actually help for now, though 23:19
allison lucian: even the occasional bug report is useful 23:20
lucian: I'm strongly motivated by failing tests :)
lucian allison: heh. what issue tracker should i use?
allison launchpad works well 23:21
23:22 payload joined
allison lucian: bugs.launchpad.net/pynie 23:23
lucian allison: ok. i've requested to join the lp team
allison lucian: approved, and welcome 23:25
chromatic/Whiteknight: my current bug in PCC refactor is a NULL call signature in a call to get_results deep inside PGE 23:29
chromatic: the irony is, it's very possible that the problem is caused by the NCI signatures, which can't be regenerated because PGE isn't working yet, (which isn't working yet because the NCI signatures haven't been regenerated...) 23:30
lucian what do people use to edit the various parrot mini-languages?
allison lucian: do you mean IDEs?
chromatic I regenerated the NCI signatures by hand. 23:31
lucian allison: more like editors. i see all sorts of weird languages, not all supported by my vim
allison chromatic: cool, will see if that fixes...
23:31 kid51 joined
allison lucian: we mostly use vim or emacs (I'm vim) 23:31
lucian: there are syntax highlighting files for these languages included in the parrot repo 23:32
lucian allison: ok, i'll have a look
allison lucian: at least for PIR, not sure about NQP
Whiteknight this is the downside of a bootstrapped buil
build
allison or bull, either works :)
Austin lucian: Check whiteknight's blog for some editor musings 23:33
Tene You can get hilighting for Perl 6 (including NQP) from github.com/petdance/vim-perl
lucian thanks all, i feel all warm and fuzzy inside :) 23:34
chromatic This is suspicious. ./parrot -t runtime/parrot/library/config.pir
Tene 'make vim-install' in the editor/ subdir of the parrot repo
chromatic 32 callmethodcc P0, "readall" P0=FileHandle=PMC(0x91d0d6c)
35 get_results PC4 (1), S4 PC4=FixedIntegerArray=PMC(0x9201a18) S4="(null)"
56 get_results PC6 (1), P3 PC6=FixedIntegerArray=PMC(0x9201a2c) P3=PMCNULL
Tene to get syntax for PIR, etc.
There's also some for emacs in there, looks like. 23:35
allison chromatic: ah well, it wasn't NCI... 23:36
Whiteknight oi. messy backtrace 23:39
allison chromatic: suspicious because it calls get_results twice?
chromatic Yes, but the second is in the exception handler, so it's not a problem here. 23:40
Whiteknight a load_bytecode, which loads_bytecode 23:41
chromatic The problem is that the get_results op isn't getting a CallSignature from the current context.
I don't know if it's material, but it's trying to get the STRING returned from FileHandle.readall(). 23:42
Whiteknight are the ncimethods updated to dtrt for returns? 23:43
allison specifically, the current context doesn't have a current_sig 23:44
though it does seem to fetch the current context fine 23:45
chromatic Righty-o.
allison Whiteknight: chromatic updated the NCI generator
chromatic I don't guarantee that I updated their returns. 23:46
allison Whiteknight: looks like that's every use of Parrot_pcc_fill_returns_from_c_args in the repo 23:47
Whiteknight ok
Austin :-)
One more passing testcase
allison Whiteknight: and the only call to fill_results is from Parrot_pcc_fill_returns_from_c_args 23:49
chromatic/Whiteknight: I'll delete those two functions now
Whiteknight nice 23:50
chromatic Parrot_pcc_set_signature(interp, _ctx, NULL); ? 23:58
Whiteknight where is that gem?
lucian Austin: Whiteknight: padre looks interesting 23:59
chromatic Every PCCMETHOD.
Whiteknight urg
allison chromatic: that's okay if it's actually done with the signature
chromatic Fortunately, it's generated code.
23:59 AndyA joined
allison (that is, the sig is supposed to be null if it's not in an active call) 23:59
Austin lucian: Yes, it does. They went backwards on multiple windows (which is critical for me, considering my display environment) but otherwise they've got a solid core.

Powered by Cro and the Raku® Programming Language.

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

Thank you!