Parrot 2.11.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Keep up with GCI
Set by moderator on 5 January 2011.
cotto_work kid51: also testing on windows, just because 00:01
00:05 plobsing joined
dalek rrot: b4b6eb4 | jkeenan++ | / (44 files):
Merge branch 'master' into remove_auto_jit
 00:20
rrot: aed05ed | jkeenan++ | / (15 files):
Merge branch 'remove_auto_jit'
kid51 You know, I think that merge is very screwed up. 00:27
I still have config/auto/jit.pm and t/steps/auto/01-jit.t 00:29
whiteknight kid51: looks like you merged the wrong way 00:30
you merged master into remove_auto_jit, not the other way around
kid51 But I was in branch master, and, following the git_workflow.
pod, I said "git merge --no-ff remove_auto_jit" 00:31
cotto_work paging dr dukeleto 00:32
plobsing kid51: it looks like you never git-rm'd config/auto/jit 00:37
even before the merge it was a tracked file
kid51 K 00:38
cotto_work looks like the merge was otherwise successful 00:39
kid51 re-running make test 00:42
cotto_work kid51: after deleting those files?
kid51 Yes
cotto_work ok 00:43
thanks for deleting that step
kid51 Hmm, I didn't see t/op/jit.t and t/op/jitn.t. 00:44
Wonder if they should be removed?
But they still PASS entirely? 00:45
plobsing they were conceived as basic execution tests for the JIT, but they are applicable to anything that executes ops
they are the first thing you'd want to get working if you wrote a new runcor
e
kid51 running 'make moretests'; should all be fixed in 10 min 00:47
cotto_work great
dalek rrot: aa66af3 | jkeenan++ | / (2 files):
Really delete config/auto/jit.pm and t/steps/auto/01-jit.t.
 00:55
kid51 Ran make tests and make moretests
00:58 kid51 is now known as kid51_at_dinner 01:00 plobsing left
dalek TT #1914 closed by cotto++: Dead code in src/pmc/Capture.pmc 01:01
TT #1914: trac.parrot.org/parrot/ticket/1914
bacek_at_work cotto_work, just my $0.02. We should not remove get_foo_keyed_str vtables. It's a bug in imcc which create PMC keys instead of str keys for $P0["foo"]. Which I'm going to fix in pirate. 01:02
cotto_work, it's about #1914
01:04 whiteknight_ joined 01:06 whiteknight left, plobsing joined, whiteknight_ is now known as whiteknight
cotto_work bacek_at_work: thanks for mentioning that, but doesn't that mean that PIRATE will need to reproduce that bug? 01:14
I'd think that a lot of code would (unwittingly) depend on that behavior. 01:15
01:16 Kapace left 01:17 plobsing left
dukeleto ~~ 01:20
cotto_work imcc-- 01:35
whiteknight imcc-- 01:39
cotto_work I love that song. 01:42
bacek_at_work: we need to consider how to deal with that misbehavior of imcc-- . PIRATE will definitely need to reproduce the brokenness, at least while we're learning to trust it. 01:44
I'm reasonably sure that that x_x_keyed_str removal wasn't the only one.
I'll revert the capture changes, but we probably have others. 01:45
whiteknight misbehavior of imcc--? How is imcc-- misbehaving now? I thought imcc-- was doing alright for now 01:46
cotto_work it doesn't ever call get_x_keyed_str 01:47
and by "it", I mean imcc--
01:47 plobsing joined
dalek TT #1914 reopened by cotto++: Dead code in src/pmc/Capture.pmc 01:51
TT #1914: trac.parrot.org/parrot/ticket/1914
01:52 kid51_at_dinner is now known as kid51
cotto_work whiteknight: ping 01:53
whiteknight pong
cotto_work whiteknight: what do you think about #1271 (spawing Parrot processes from PIR in tests)? Some tests already do that. 01:54
plobsing cotto_work: why revert the changes right now? if it is desired, it can be resurected or reimplemented.
cotto_work bacek_at_work: how easy is it to make PIRATE behave the same as imcc-- ? 01:55
whiteknight cotto_work: close that ticket. I don't think it's really meaningfull
kid51 Uh-oh: Andy Dougherty has opened 4 tickets in the last 5 hours! 01:56
whiteknight yeah, I think it would be easier if Andy just committed the fixes directly 01:58
we all trust that he's not going to break things! 01:59
dalek rrot: be8f1a6 | cotto++ | t/src/misc.t:
[t] fix src tests for C89 compatibility; doughera++ for the patch
02:00 Kristaba left
bacek_at_work cotto_work, way too hard. I'm not going to implement bug-to-bug compatibility 02:00
dukeleto (bug-for-bug compatibility)-- 02:01
cotto_work bacek_at_work: if you think that this is one of the things we can clean up during the imcc->PIRATE transition, that's fine. 02:02
I just don't want to see mysterious bugs happen because of the difference. 02:03
bacek_at_work I'm trying to keep compatibility as much as possible.
Look at t/03-pbc/99-imcc.t
kid51 whiteknight: I see you did the TT #1952 patch. Are you going to do the other 2? Should I? 02:04
whiteknight #1952 patch? I didn't touch that ticket 02:07
kid51 Oh, cotto did
cotto_work kid51: go for it
thanks
kid51 testing the 1950 and 1953 patches 02:08
That leaves his test report in TT #1951
dalek TT #1271 closed by cotto++: Ability to spawn Parrot processes from PIR in tests
TT #1271: trac.parrot.org/parrot/ticket/1271
rrot: b51c019 | jkeenan++ | / (2 files):
Apply patches submitte by doughera++ in TT #1950 and TT #1953.
 02:20
rrot: 901764e | jkeenan++ | / (2 files):
Merge branch 'doughera_3patches'. Applies patches submitted by doughera++ in
cotto_work bacek_at_work: I very much appreciate the effort you're expending to preserve backwards-compatibility in PIRATE. How would you suggest we deal with that inconsistency between imcc-- and PIRATE? 02:23
bacek_at_work Best case scenario - use imcc for bootstrapping only.
No 02:24
Best case - emit C from PIRATE for bootstrapping.
02:24 Yuki`N left
bacek_at_work Good case - use imcc for bootstrapping only. 02:24
cotto_work That's my long-term hope.
bacek_at_work Worth case - we all die in 2012
plobsing separate IMCC out to separate executable (already planned), add PIRATE fakecutable to parrot. deprecate imcc executable after a few cycles. 02:25
bacek_at_work plobsing, +1 02:27
plobsing we'll need a long transition period 02:28
cotto_work +1 02:29
02:31 gbacon joined 02:39 whiteknight left
dalek TT #1954 created by jkeenan++: 'make docs' target: Why are these files not installed? 02:41
TT #1954: trac.parrot.org/parrot/ticket/1954
02:43 preflex left 02:46 preflex joined 03:06 kurahaupo1 left 03:25 kid51 left
dalek rrot: 72addf8 | plobsing++ | / (2 files):
add serialization for Exception

see Ī©Ī·;)XD for an example of why this might be considered useful		 03:32
rrot/nwellnhof/platform_src: 8bd0d93 | nwellnhof++ | / (11 files):
[io] Refactor platform socket API
 03:55
04:08 Yuki`N joined 04:41 gbacon left 04:47 kurahaupo left 04:48 kurahaupo joined 04:57 nwellnhof left 05:07 cotto joined
cotto ~~ 05:07
05:12 Yuki`N left
dalek rrot/leto/embed_grant: fd066d2 | dukeleto++ | t/src/embed.t:
[t] Various tests for Parrot_PMC_* functions
 05:18
05:19 davidfetter joined
cotto seen nwellnhof 05:23
aloha nwellnhof was last seen in #parrot 6 hours 38 mins ago saying "but we could really need some win32 hackers...".
05:39 ppant joined 05:40 estrabd left
dukeleto what is our new goal for this week? 05:40
plobsing we had 5 IIRC 05:41
dukeleto plobsing: yes, but our /topic didn't change
how would I test Parrot_PMC_destroy() from C ? 05:42
cotto If only some slacker hadn't forgotten to update it.
dukeleto cotto: slacker!
plobsing cotto: you've saved yourself some trouble, one task is already down
dukeleto late binding to the rescue! 05:43
cotto plobsing, I saw. Thanks!
dukeleto, kinky
;)
dukeleto cotto: whoa nellie! 05:44
05:45 rurban_ joined
dukeleto plobsing++ # thaw+freeze for Exception 05:46
05:48 rurban left, rurban_ is now known as rurban
dalek rrot/leto/embed_grant: 26b09c1 | dukeleto++ | t/src/embed.t:
[t] More Parrot_PMC_* tests
 05:48
plobsing dukeleto: my motivations for that were purely selfish 05:49
although I can imagine that would be an efficient way to build exceptions in many situations
dukeleto plobsing: i am interested in freeze/thawing exceptions in PL/Parrot 05:50
plobsing dukeleto: I'm curious to the applicability there. care to elaborate? 05:51
05:52 kurahaupo left
moderator Parrot 2.11.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Goals: write GettingStartedWithPlumage, review html documentation, test HLLs, review deprecations | no major changes after Tuesday, only critical fixes and documentation updates after Friday 05:53
dukeleto plobsing: i just want to toy around with freezing the state of the parrot interpreter that is embedded in the database 06:02
plobsing it doesn't store ephemeral stuff like that just yet. that's a much harder problem. 06:04
dukeleto plobsing: i understand, but it is the first step in serializing certain things
plobsing ok, so long as you aren't expecting miracles 06:05
dukeleto plobsing: does freezing an interp object work?
plobsing: no, i expect nothing and I am pleased with whatever works
plobsing dukeleto: no. it did something you wouldn't expect and was abused by IMCC. now it explicitly dissallows freezing.
if we get serializable execution, that can change. 06:06
dukeleto plobsing: what needs to happen for serializable execution? 06:07
plobsing I haven't really thought about it much (beyond "zomg smalltalk is nifty") 06:09
06:10 fbrito joined
plobsing we have much more pressing matters when it comes to serialization 06:10
dukeleto plobsing: i recently got the smalltalk "blue book", still need to read more of it
plobsing jnthn very much wants delimited serialization
dukeleto plobsing: what does that mean?
plobsing serialization that stops on certain conditions in stead of traversing the whole object graph
dukeleto sounds useful 06:11
such as "serialize all the Foo PMCs" ?
incoming
plobsing more like, serialize this PMC, but don't bother storing the entire MOP 06:12
dalek rrot: fd066d2 | dukeleto++ | t/src/embed.t:
[t] Various tests for Parrot_PMC_* functions
rrot: 26b09c1 | dukeleto++ | t/src/embed.t:
[t] More Parrot_PMC_* tests
rrot: f90b118 | dukeleto++ | / (361 files):
Merge branch 'master' into leto/embed_grant
rrot: 2d7d641 | dukeleto++ | / (5 files):
Merge remote branch 'origin/master'
plobsing that's jnthn's usecase. but from the serializer's perspective it should be more or less arbitrary 06:13
dukeleto sure. a callback or something that causes serialization to stop when it is false, or something like that 06:14
plobsing yes. something equivalent. it's an easy enough problem if you're just thinking about 1 constant in 1 PBC 06:15
but then you have to think about pbc_merge. is there a way to make it smart about the inter-pbc references? can it fix them up somehow? 06:16
cotto Cool. We have a random pull request. 06:26
plobsing the pull request seems to have picked up a hitchhiker from src/pmc/sub.pmc. fairly sure that part is unecessary 06:30
cotto looks mostly harmless 06:32
dukeleto cotto: cool! 06:35
cotto Does Parrot have any current sponsors? 06:56
dukeleto cotto: no and I would like to fix that 06:57
cotto: who would want to sponsor us? 06:58
cotto dukeleto, I was just asking because our release announcement template mentions some. 07:03
parrot.org/sponsors <-what about them? 07:04
dukeleto cotto: very old info 07:05
cotto I figured.
dukeleto cotto: they all donated in the past
cotto: in that sense, they are sponsors
cotto: prepare for a torrent of smoke reports 07:06
cotto prepares
dukeleto cotto: i am running a script in a while loop that smokes 11 differently compiled flavors of parrot, on two machines 07:08
cotto dukeleto, on the gcc compile farm?
dukeleto cotto: one machine is in the farm, the other is my laptop
cotto: i already found one test failure in t/src/atomic.t when all external libs are turned off 07:09
cotto nice
dukeleto cotto: i am not testing every combination, just the ones that seem more important than others 07:14
cotto: i am testing with/without optimized with/without gmp/icu/threads/libffi 07:15
cotto: as well as one with no extra args and another that turns off all external libs 07:16
cotto: this is running on a 64bit linux machine 07:17
cotto dukeleto, is that going to be a regular deal? It sounds like a good thing to run regularly overnight. 07:19
dukeleto cotto: here is that atomic.t failure: smolder.parrot.org/app/projects/rep...tails/2215 07:20
cotto: yes, i am going to try to make it a regular thing. It is running in a screen session now, and it runs any script that matches a pattern in a directory, so I can add new types of builds without restarting the script
cotto: github.com/leto/Util/blob/master/b...parrot_all 07:21
cotto: we have a parrot smoker!
07:23 hatseflats joined
cotto dukeleto, I love your laziness 07:23
07:23 sjn joined
dukeleto cotto: yes, it is a virtue 07:24
cotto: those scripts are wheels within wheels :)
we just need better reporting tools for smolder 07:25
i feel that once info goes in, it is hard to get out
our smolder rss feeds are broken (because smolder has a bug where the URLs in the feeds are incorrect) and we don't get emails about smoke failures 07:26
cotto yeah. Does Smolder expose any kind of api? 07:29
07:29 fbrito left
dukeleto cotto: not really 07:34
cotto: i am seeing some colorful smoke
bacek_at_work dukeleto, is it legal to smoke such a colourful stuff??? 07:36
:)
dukeleto bacek_at_work: sure is fun, regardless ;) 07:37
07:38 fperrad joined
dukeleto smolder.parrot.org/app/projects/tap...m/2228/305 07:38
that looks bad
from smolder.parrot.org/app/projects/rep...tails/2228
odd stuff
oh, i understand 07:39
i put in hard ulimits
cotto That's what happens when you smoke the hard stuff.
bad things
dukeleto indeed. It all goes up in smoke.
dukeleto reset his virtual memory ulimit and tries again 07:46
07:47 contingencyplan left
dukeleto cotto: that did the trick 07:49
cotto: now i have 16 permutations of build flags :) 07:50
GONNA FIND SOME BUGS
cotto delicious bus
bugs
dukeleto cotto: don't get hit by the delicious bus 07:51
dukeleto wonders if a company has ever taken a team out to eat bugs as a "team building" exercise
cotto the tastiest of all injuries
dukeleto "make test" takes 1.5 minutes on the compile farm machine 07:54
moritz dukeleto: is that parallelized?
dukeleto moritz: i am using TEST_JOBS=3 07:55
it is a 24 cpu machine 07:56
moritz dukeleto: how many cores?
wow
moritz wants
dukeleto moritz: i gonna hook up rakudo spectests too
moritz: you can request access by following the instructions on gcc.gnu.org/wiki/CompileFarm 07:57
moritz: but i plan to start smoking HLLs on parrot master and doing spectesting
moritz: it would be nice to have bus number >1 and have a person in both rakudo and parrot that can admin the smokers 07:58
moritz dukeleto: aye... unfortunately my time is currently more limited than usually
$wife being pregnant and having health problems :( 07:59
I'll put it on my TODO list
dukeleto moritz: no worries. I am slowly but surely doing it. As well as 1/0 other things
moritz: i hope your wife feels better and good luck with Baby.new 08:00
smolder graphs are broken too 08:02
dukeleto pouts
moritz thanks 08:03
cotto lta 08:06
dukeleto cotto: smolder.parrot.org/app/projects/rep...tails/2245 <-- --without-threads on 64bit fails "make test" 08:39
mikehh: ^^^ 08:40
08:44 theory left 08:55 TiMBuS left 08:56 TiMBuS joined 08:57 ppant left
bacek aloha, humans 09:29
10:54 dip joined 11:11 Kristaba joined 11:51 mj41 left 11:53 mj41 joined 12:37 he_ joined 12:55 gbacon joined 13:01 extreme left 13:06 estrabd joined 13:09 extreme joined 13:45 rurban_ joined 13:48 rurban left, rurban_ is now known as rurban 14:11 mj41 left 14:13 plobsing left 14:16 whiteknight joined 14:24 mj41 joined
whiteknight good morning, #parrot 14:26
14:29 kthakore is now known as Texas, Texas is now known as kthakore 14:32 bluescreen joined 14:48 mtk joined 14:58 kapace_ left 15:07 fbrito joined 15:12 Andy joined 15:16 contingencyplan joined 15:25 dmalcolm joined
whiteknight I'm going to install an OpenIndiana image now, to start getting some tests on that platform 15:32
mikehh t/src/embed.t does not like g++ (make corevm/makje coretest) 15:40
15:41 hercynium joined
mikehh following that make world fails to build with g++ -> src/pbc_disassemble.c:80:1: error: invalid conversion from ā€˜intā€™ to ā€˜OPTION_flagsā€™ 15:41
15:45 plobsing joined 15:50 Patterner left 15:51 Psyche^ joined, Psyche^ is now known as Patterner 15:56 extreme left
whiteknight msg plobsing when you get a chance can you take a look at TT #387? I don't know if it's necessary, and if not we can reject it. Thanks 15:57
aloha OK. I'll deliver the message.
dukeleto ~~ 16:02
whiteknight msg cotto can you take a look at TT #379? I'm thinking we can deprecate the err opcodes and no longer exposing PIR users to raw errno constants, but I want a second opinion on the sanity of that idea
aloha OK. I'll deliver the message.
mikehh dukeleto: your changes to t/src/embed.t are not liked by g++ 16:04
plobsing whiteknight: (re: 387) I don't really care one way or another. I think the coupling between packfiles of different systems caused by the expectation of "read anywhere (but be more efficient locally)" are wrong. 16:05
whiteknight plobsing: Okay, that's fine too. What do we replace it with?
plobsing Every parrot should be able to read 2 and only 2 PBC formats: local, decode-efficient format, and universal, space-efficient format. 16:06
whiteknight okay. So we need two PBC reader/writer systems for those
plobsing The difference is that universal format is *not* the default, you have to ask. Also, everybody pays equally for universal.
I'm thinking universal would be some kind of variable-width format like that employed by protobufs 16:07
16:07 theory joined, Kapace joined
whiteknight plobsing: Jot those ideas down in the PackfileTasklist wiki page so we keep track, and close that #387 ticket since it's not heading in that direction 16:07
plobsing because if you are going to transfer files between systems, and those systems are going to copy the PBC (as opposed to mmap), you care very much about the space
16:07 davidfetter left
whiteknight or, I can close the ticket 16:08
plobsing under that framework, there is nothing stopping someone from configuring with single-precision float
because nobody else has to care
16:10 extreme joined
plobsing whiteknight: we already have a reader systems that is sufficiently general to support this. all we need is indirection in the writter system (and I think there were patches some time ago to do just that). 16:10
whiteknight okay
how do we make a universal format equally painful on all systems? Integers are big- or little-endian, not both 16:11
dalek TT #387 closed by whiteknight++: single float 16:12
TT #387: trac.parrot.org/parrot/ticket/387
plobsing integers are variable width, and must always be decoded byte-by-byte. space efficient, codec-inefficient. see protobufs as an example.
endianness doesn't come in to play 16:13
dukeleto mikehh: what kind of errors do you get?
whiteknight damnit. sockets failing on Windows 16:15
16:15 bluescreen left
mikehh dukeleto: loads - t/src/embed.t - Failed tests: 8-41 16:19
dukeleto: 16:20
# Failed test 'Parrot_PMC_i_add_float'
# at t/src/embed.t line 75.
# 'g++ -D_GNU_SOURCE -DHASATTRIBUTE_CONST -DHASATTRIBUTE_DEPRECATED -DHASATTRIBUTE_MALLOC -DHASATTRIBUTE_NONNULL -DHASATTRIBUTE_NORETURN -DHASATTRIBUTE_PURE -DHASATTRIBUTE_UNUSED -DHASATTRIBUTE_WARN_UNUSED_RESULT -DHASATTRIBUTE_HOT -DHASATTRIBUTE_COLD -DHAS_GETTEXT -g -I./include -c -o t/src/embed_9.o t/src/embed_9.c' failed with exit code 1
# Failed to build 't/src/embed_9.o': t/src/embed.t: In function ā€˜int main()ā€™:
# t/src/embed.t:91:72: error: cannot convert ā€˜PMC*ā€™ to ā€˜parrot_string_t*ā€™ in assignment
# t/src/embed.t:98:44: error: ā€˜Parrot_io_printfā€™ was not declared in this scope
dukeleto: lots of similar failures 16:22
dukeleto: of course it passes when built with gcc 16:23
dukeleto mikehh: interesting! 16:24
mikehh or should I say run with a gcc build
dukeleto cotto, whiteknight : i just sent another email about deprecation to parrot-dev
mikehh: i think i was being lazy with some typecasts, but i don't get the error about it not knowing about Parrot_io_printf. Is that error spurious? 16:25
mikehh g++ requires includes that gcc seems to accept 16:26
or I think it might have different scoping rules 16:27
dalek TT #1955 created by fperrad++: socket broken on Windows 16:28
TT #1955: trac.parrot.org/parrot/ticket/1955
dukeleto mikehh: looks like i need to setup a c++ smoker 16:33
mikehh: can you nopaste the commands that you use to run your test suites? 16:34
whiteknight plobsing: What is a PackFile_ByteCode structure, and what's the difference between that and a PackFile*? 16:36
dukeleto mikehh: also, if you can give me a link to a smoke report for your g++ failures, i can try to sort them out
whiteknight plobsing: IMCC compiles files and stores the results directly in interp->code (a PackFile_ByteCode*). Obviously this is really bad for coupling and I might like to do a little untangling in the kill_packfile_new_dummy branch 16:37
instead of writing directly to interp->code, I would like those functions to return the necessary pointers, and let libparrot decide where to put them 16:38
and all the functions in compilers/imcc/pbc.c which use interp->code can instead take an extra parameter argument to a PackFile_ByteCode* structure and use that 16:42
oooh. I think this cleanup is going to be fun
plobsing whiteknight: PackFile_ByteCode is *only* the Ops segment. It does not contain constants, annotations, debugs, etc... 16:43
perhaps we should rename to PackFile_OpCodes. 16:44
whiteknight okay, so it's part of the PackFile. interp->code is just a cache for interp->initial_pf->(path/to/Ops segment)?
plobsing considering our "bytecode" has little if nothing to do with bytes (octets for the pedants among us) at the current time
Coke is doing a g++ build now, will report a smoke link for you.
whiteknight okay. I am going to try and decouple IMCC from interp->code and interp->initial_pf right now 16:45
it actually doesn't look so hard to do
plobsing whiteknight: really? IMCC uses the interp as its state storage in many places. I'd be surprised if it were easy.
whiteknight at the very least, it looks relatively straight-forward to make IMCC stop using interp->code 16:46
plobsing do we support the distinction between utf-16le and utf-16be somehow? if so how? 16:47
likewise for ucs 16:48
very probably we'll need to encode all unicode strings utf-8 for network pbc 16:49
dukeleto Coke++
nopaste "mikehh" at 192.168.1.3 pasted "test run options for dukeleto" (26 lines) at nopaste.snit.ch/27688 16:52
dukeleto mikehh++ 16:55
mikehh dukeleto: used to also run with --maintainer but dropped that recently 16:57
git gives me a hard time if I don't add the changes 16:58
dukeleto mikehh: what problem is git giving you? 16:59
mikehh if I used --maintainer with cConfigure and it regenerated some imcc files 17:00
plobsing mikehh: git checkout compilers/imcc will reset it.
mikehh plobsing: ok, maybe I will readd it 17:01
Coke I can't even build with g++
src/pbc_disassemble.c:80: error: invalid conversion from Ć¢intĆ¢ to Ć¢OPTION_flagsĆ¢
dalek tracwiki: v8 | plobsing++ | PackfileTasklist
tracwiki: add preliminary ideas about PBC portability (avoiding n-to-n translation as we have now)
tracwiki: trac.parrot.org/parrot/wiki/Packfil...ction=diff
Coke I'm at 2d7d641 17:02
mikehh Coke: I also got that but I could build with make corevm/make coretest
Coke realizes his putty settings on this box are wrong. 17:04
src/pbc_disassemble.c:80: error: invalid conversion from ā€˜intā€™ to ā€˜OPTION_flagsā€™ 17:05
(there, that's better)
pretty sure I have a fix for that. 17:07
dalek rrot: 44e5a0a | Coke++ | src/pbc_disassemble.c:
Fix g++ build.
 17:09
17:10 bluescreen joined
plobsing does host-to-host PBC-compatibility modification require a deprecation notice? 17:10
Coke doing a 'make smoke'
has it ever ever worked?
and/or is it covered by existing PBC notices?
plobsing I think everyone can read x86_32-generated packfiles ATM
I *will* be breaking that, and replacing it with a more suitable network format. 17:11
Coke s/replacing/providing a new thing/ (so folks can actually switch before the old stuff is removed.) 17:12
plobsing the new will be added before the old removed, certainly 17:14
17:24 plobsing left
Coke smolder.parrot.org/app/projects/rep...tails/2341 17:26
(g++ test failures)
17:29 plobsing joined 17:34 theory left 17:39 fbrito left 17:41 fbrito joined 17:46 fbrito left
dukeleto Coke++ # thanks. It looks like a casting issue 17:50
17:54 KatrinaTheLamia left
cotto_work ~~ 17:55
dukeleto cotto_work: mornin' 17:58
mj41 Perl 6 and Parrot links stats: year 2010 309 links, year 2009 367 links, year 2008 202 links ... tapir2.ro.vutbr.cz/P6aP-links/stats.html
dukeleto mj41++ # glad to see you around
mj41: clicking on the links there starts a download of index.php, is that supposed to happen? 17:59
plobsing mj41: are those auto-added somehow, or is that index manually maintained? 18:00
mj41 dukeleto: manually maintained, index.php is MediaWiki way to download paragraph 18:01
dukeleto: moving home from work. I'm hungry. 18:03
18:08 bluescreen left 18:14 theory joined
dukeleto nom++ 18:22
18:23 plobsing left
cotto_work nom++ indeed 18:23
or were you referring to food?
18:29 plobsing joined
dukeleto cotto_work: both 18:33
18:35 gbacon left
cotto_work aloha: clock? 18:40
aloha cotto_work: LAX: Wed, 10:40 PST / CHI: Wed, 12:40 CST / NYC: Wed, 13:40 EST / UTC: Wed, 18:40 UTC / LON: Wed, 18:40 GMT / BER: Wed, 19:40 CET / TOK: Thu, 03:40 JST / SYD: Thu, 05:40 EST
18:52 gbacon joined 18:59 bluescreen joined 19:29 plobsing left, plobsing joined
cotto_work msg bacek If we were to make the Packfile PMCs the official (and only) way for users to interact with packfiles starting at 3.0, how would the increased stability requirements affect PIRATE? 19:32
aloha OK. I'll deliver the message.
whiteknight cotto_work: I sincerely hope that's not an "if"
cotto_work whiteknight: the question is 3.0 or 3.3. That's all. 19:33
whiteknight a much better question
cotto_work I just don't want to enforce stability on the PMCs before they're ready.
er, force
Coke seems like if there's another way to do it now, you can't say that /in/ 3.0 they can't. 19:51
or do you mean immediately after 3.0?
19:51 plobsing left
cotto_work Coke: immediately after. 19:56
outlook-- 19:58
TimToady phone 19:59
dukeleto which versions of gcc do we want to smoke parrot on? 20:02
dukeleto is starting with just 4.3.1 20:05
4.3.2 rather
the compile farm has pre-compiled binaries of every version along with pre-releases and trunk and whatnot
dalek TT #1956 created by doughera++: Confused test 2 in t/src/misc.t 20:10
TT #1956: trac.parrot.org/parrot/ticket/1956
tadzik ...
oops, ww
20:11 plobsing joined
dukeleto cotto_work: now we are smoking gcc and g++ on each flavor of parrot 20:12
tadzik hmm, my blagpost tagged perl 6 was published to planet parrot, is that intentional? 20:13
dukeleto tadzik: your entire blog was probably added to planet parrot
tadzik yeah, hence my wondering, I remember someone (whiteknight?) asking me specifically about the parrot category feed 20:14
20:27 theory left
dalek TT #1957 created by doughera++: Configure.pl ignores argument given to --optimize 20:27
TT #1957: trac.parrot.org/parrot/ticket/1957
20:28 mikehh left
Coke [ttjjss.wordpress.com/category/parrot/feed/] 20:32
name = tadzik
tadzik weird then
well, doesn't matter as long as you don't mind seeing my perl 6 stuff on the Parrot planet 20:33
dalek rrot: 4740da3 | NotFound++ | t/src/misc.t:
fix test Parrot_vsnprintf with len 0, TT #1956
tadzik the first thing in the feed seems to still be the "Presenting Parrot" post
Coke might be worth asking the perl.org admins (since they admin the planet servers, including parrot's) 20:37
but I don't care enough to worry about it, no. 20:38
20:39 Yuki`N joined
dalek TT #1956 closed by NotFound++: Confused test 2 in t/src/misc.t 20:44
TT #1956: trac.parrot.org/parrot/ticket/1956
20:49 shortcircuit left
dalek rrot/exception_backtraces: 06edc1c | plobsing++ | / (2 files):
only be locale-specific about character type (LC_CTYPE, eg: UTF8)

respecting locale globally breaks tests in many non-english locales by changin printf, errno messages, etc		 20:50
rrot/exception_backtraces: 06b1fad | Whiteknight++ | DEPRECATED.pod:
Deprecate functions in src/embed.c. TT #1910
20:50 shortcircuit joined
rrot/exception_backtraces: ae83646 | plobsing++ | / (3 files):
reset locale after grabbing character encoding - fixes tests on internationalized systems
rrot/exception_backtraces: d539bfb | NotFound++ | t/dynpmc/os.t:
untodo OS test, TT #1836
rrot/exception_backtraces: c638c31 | bacek++ | tools/dev/pbc_to_exe.pir:
Remove c++-style commented out line. Closes #1949. doughera++
rrot/exception_backtraces: b4b6eb4 | jkeenan++ | / (44 files):
Merge branch 'master' into remove_auto_jit
rrot/exception_backtraces: aed05ed | jkeenan++ | / (15 files):
Merge branch 'remove_auto_jit'
rrot/exception_backtraces: aa66af3 | jkeenan++ | / (2 files):
Really delete config/auto/jit.pm and t/steps/auto/01-jit.t.
rrot/exception_backtraces: be8f1a6 | cotto++ | t/src/misc.t:
[t] fix src tests for C89 compatibility; doughera++ for the patch
rrot/exception_backtraces: b51c019 | jkeenan++ | / (2 files):
Apply patches submitte by doughera++ in TT #1950 and TT #1953.
rrot/exception_backtraces: 901764e | jkeenan++ | / (2 files):
Merge branch 'doughera_3patches'. Applies patches submitted by doughera++ in
rrot/exception_backtraces: 72addf8 | plobsing++ | / (2 files):
add serialization for Exception

see Ī©Ī·;)XD for an example of why this might be considered useful
rrot/exception_backtraces: 8c2f5cc | plobsing++ | / (27 files):
Merge branch 'master' into exception_backtraces
plobsing heuristic branch merge failure 20:51
NotFound Suggestion: don't merge master into branches, better branch again from master and merge the old branch
plobsing NotFound: does the heuristic treat master as special somehow? 20:52
NotFound plobsing: this way it doesn't repeat master commits
And probably it gives less and cleaner conflicts. 20:54
A lot of functions in src/string/encoding.c doesn't follow the Parrot_subsystem naming convention. 20:55
dalek rrot/imcc_interp_code: 620bd8d | Whiteknight++ | / (30 files):
Merge branch 'master' into kill_packfile_new_dummy
 21:01
rrot/imcc_interp_code: 74c1720 | Whiteknight++ | / (2 files):
PackFile_append_pbc is renamed PackFile_append, and takes a PackFile argument instead of a char* filename
rrot/imcc_interp_code: 1d371d6 | Whiteknight++ | / (21 files):
start decoupling IMCC from libparrot. First step is to change IMCC so it does not directly write to interp->code. This commit goes most of the way there, but the build segfaults when compiling HLLCompiler.
Heuristic branch merge: pushed 17 commits to parrot/kill_packfile_new_dummy by Whiteknight
Coke (re-branching) ew.
whiteknight Coke: was that aimed at me?
Coke pita if other people are working on the same branch... 21:02
NotFound.
NotFound Coke: you can merge again 21:03
dalek rrot/smoke_fulltest: 4fca70c | dukeleto++ | config/gen/makefiles/root.in:
Attempt to add smoke_fulltest make target

This doesn't work properly because our harness seems to only respond to a single
  --foo-tests argument, so it only runs --code-tests.
dukeleto Coke: feel free to add some tuits to the smoke_fulltest branch. I have run out for the time being. 21:04
whiteknight plobsing: take a look at imcc_interp_code. I'm about 90% of the way to making IMCC not write to interp->code directly
dukeleto that goes for anybody else, as well
whiteknight actually, I've already removed all instances of interp->code from there, but the build doesn't complete
dukeleto whiteknight++ # making imcc-- less hateful
whiteknight I suspect I'm screwing up some of the Packfile/segment-related crap. The API for working with those things is not exactly friendly 21:05
keeping track of the current ByteCode segment at all times is tricky, since IMCC is always calling Parrot_pbc_load to change the currently-loaded one 21:07
Coke dukeleto: me or cotto? 21:11
whiteknight What I really want is to be able to hand IMCC a .pir file and get back a PackFile structure, or PackFile PMC in return 21:16
21:17 plobsing left
whiteknight then libparrot can decide whether to append that new PackFile for immediate execution, or write it to a file, or whatever it wants to do 21:17
21:17 shortcircuit left
whiteknight and the only way that works cleanly is if IMCC is not writing data directly to the interp structure 21:18
21:18 shortcircuit joined
whiteknight if IMCC assumes it's the only compiler in town, it's going to create a situation that makes other compilers impossible to implement. Self-fullfilling prophecy 21:18
actually, I was wrong. the imcc_interp_code branch builds and is running tests right now 21:19
NotFound whiteknight: maybe a better start will be to change all imcc functions that take a PARROT_INTERP parameter and get imcc_info from it to get an imcc data argument an get the interpreter from it. 21:20
That is, get rid of the imc_info from the interpreter, maybe replacing it with a new structure with default values for the flags and the like. 21:21
21:25 plobsing joined 21:26 jan left
dukeleto Coke: all of the above 21:26
whiteknight NotFound: Yes, that's probably a good idea too 21:27
I'm not doing that right now though
NotFound Doesn't need to be the first step, it was a suggestion just in case the current work was being too difficult to fix. 21:28
21:28 whiteknight left 21:29 jan joined 21:31 nwellnhof joined
dalek rrot/kill_packfile_new_dummy: fa6d538 | plobsing++ | src/packfile/api.c:
make sure initial_pf is in place whenever we build default segs
 21:35
nwellnhof working on parrot's io subsystem turns out to be giant yak shaving exercise.
i just realized that sockets on Win32 must have never really worked.
except cygwin 21:36
plobsing not entirely surprising
dalek rrot: 513d109 | nwellnhof++ | t/pmc/ (2 files):
[t] Add a real test for the Socket PMC

It turned out that Win32 is broken.		 21:38
plobsing as a heavy weight, highly abstracted VM, things like cross-platform io, unicode, etc should be our strong points, yet they are our weakest.
nwellnhof this commit should bring the socket code to some reasonable test coverage 21:39
plobsing msg whiteknight kill_packfile_new_dummy passes up to codetests now
aloha OK. I'll deliver the message.
nwellnhof it has been somewhere at 25% up to now
i forgot to update MANIFEST 21:41
21:42 theory joined
dalek rrot: 237613e | nwellnhof++ | MANIFEST:
Update MANIFEST
 21:43
21:46 rurban_ joined 21:47 bluescreen left 21:48 rurban left, rurban_ is now known as rurban 21:55 Patterner left 21:56 Psyche^ joined, Psyche^ is now known as Patterner 21:57 plobsing left 21:59 plobsing joined, fperrad left
tadzik github.com/parrot/plumage/issues#issue/3 -- could someone apply this? 22:01
dalek TT #1958 created by nwellnhof++: Byte order of string encodings 22:07
TT #1958: trac.parrot.org/parrot/ticket/1958
cotto_work seen chromatic 22:10
aloha chromatic was last seen in #parrot 3 days 3 hours ago joining the channel.
22:10 vts joined
dalek rrot: 113df7b | NotFound++ | runtime/parrot/library/Test/More.pir:
allows to use a string with PIR source as invokable in throws_type
 22:11
nopaste "tadzik" at 192.168.1.3 pasted "Using a module from lua in Perl 6" (85 lines) at nopaste.snit.ch/27689 22:14
tadzik could someone take a look? 22:15
cotto_work bacek: ping 22:16
bacek_at_work cotto_work, pong (barely here)
cotto_work bacek_at_work: how are you building PIRATE? tree-optimization doesn't build with the latest parrot.
dalek rrot/nwellnhof/platform_src: 97a37c9 | nwellnhof++ | / (7 files):
Move process related code from io.c to exec.c
rrot/nwellnhof/platform_src: 3446a1b | nwellnhof++ | src/io/socket_api.c:
[io] Fix bug in recv
rrot/nwellnhof/platform_src: 81fc286 | nwellnhof++ | / (4 files):
[t] Fix TT #1944

We don't generate platform-specific code now, so this problem is solved.
bacek_at_work cotto_work, hm. It was buildable few days ago.
cotto_work actually, it doesn't pass its tests
I guess that doesn't keep it from installing. 22:18
dalek umage: e7c5d3b | NotFound++ | plumage/metadata/cardinal.json:
update Cardinal metadata to its new repository
 22:20
NotFound tadzik: done
tadzik++ 22:21
22:21 bluescreen joined 22:24 vts left
bacek_at_work cotto_work, ah. yes. Just install it. 22:39
cotto_work, and if can catch tcurtis and move tree-optimizations under parrot's corporation it will be good. 22:40
dalek rrot: 025ff86 | NotFound++ | src/packfile/api.c:
rename local variable 'main' to avoid the warning: 'main' is usually a function
 22:41
cotto_work dukeleto: ping
bacek_at_work: ok. It looks much better if I just install. 22:42
22:46 whiteknight joined
whiteknight hello #parrot 22:48
plobsing: ping 22:49
tadzik hello whiteknight
whiteknight hello tadzik
tadzik: submit a CLA so you can become a Parrot committer
tadzik tadzik: it's on the go to the USA
whiteknight: I haven't commited much yet :) 22:50
ow, I must stop talking to myself
nwellnhof i submitted my CLA by fax. much easier...
tadzik bah, #perl6 wasn't sure if that was (still?) possible 22:51
whiteknight I tried twice to fax my CLA before I finally printed it out and mailed it
tadzik whiteknight: how for do you have to CA?
whiteknight msg plobsing I would really like to re-do PF_create_default_segs. First off, I would like it to take a PackFile* reference, and not act directly on interp->initial_pf or interp->code 22:52
aloha OK. I'll deliver the message.
whiteknight that would help clean up IMCC code a hell of a lot
tadzik: I do not understand what you just said
tadzik whiteknight: what distance exists between you, and the CLA delivery location?
whiteknight tadzik: I am in US east coast. I think the CLAs go to the west coast 22:53
22:53 Andy left
whiteknight so ~3000 miles 22:53
~4830km
tadzik hah, I have a few countries and one ocean :)
I hope it will get there in a week 22:54
whiteknight i hope so too. I've only looked at one of your patches and I'm already tired of doing it
tadzik so bad? :) 22:55
bacek_at_work cotto_work, (packfiles) Yes, we should deprecate Parrot_pf_* functions. Current Packfile PMCs should be declared experimental until further notice :)
cotto_work bacek_at_work: how long do you think they'll be experimental? 22:57
we eventually need to make the official 22:58
tadzik whiteknight: were they so bad? 23:00
NotFound lua doesn't install from plumage for me, it fails two tests
tadzik same here, but it fails assign.t non-deterministically 23:01
dukeleto cotto_work: pong 23:08
bacek_at_work cotto_work, till 3.3 I hope
dukeleto tadzik: you can email in a scanned version of the CLA, i am pretty sure 23:09
bacek_at_work I just don't want to finalize APIs right now.
OTOH, I don't expect many changes.
dukeleto particle: ping! Is it ok to email in a scanned version of the CLA?
tadzik: i think you send it to legal@parrot.org, but i would verify that with particle
tadzik dukeleto: no worries, I'm not in a hurry (that much) 23:10
dukeleto tadzik: we are :)
tadzik oh :)
dukeleto tadzik: feel free to fork parrot on github and send pull requests in the meantime
tadzik come on, GCI students deserve more commit bits :) 23:11
I'm only commiting small things from time to time
cotto_work dukeleto: can you move tree-optimizations into the parrot org on github? 23:13
dukeleto: a scanned cla is fine 23:14
23:15 KatrinaTheLamia joined
tadzik I can reprint it and scan and mail you, if you wish 23:16
dukeleto tadzik: sure. you can mail a snail mail copy if you want, but emailing is fine. 23:18
23:18 bluescreen left
dukeleto cotto_work: sure 23:18
cotto_work: i assume you want that to get tested along with HLLs
cotto_work yes
tadzik dukeleto: the snailmail is alredy on the go
dukeleto tadzik: gotcha. well, if you in an email, we can get you a bit much sooner :) 23:19
cotto_work: you want it to live at parrot/tree-optimization ?
cotto_work yes
tadzik dukeleto: I'll maybe do it tomorrow
dukeleto cotto_work: ask and ye shall receive : github.com/parrot/tree-optimization 23:24
dukeleto is adding the proper github hooks to it now 23:25
cotto_work: ok, commits to the new tree-optimization repo should be reported in here 23:26
cotto_work dukeleto++
dukeleto: if I've cloned from another repo, how do I tell git to use parrot/tree-optimizations?
tadzik cotto_work: change the url in .git/config 23:27
cotto_work tadzik: that's all there is? Nice. 23:28
dukeleto Coke: url = git@github.com:parrot/tree-optimization.git
cotto: url = git@github.com:parrot/tree-optimization.git
Coke: oops, sorry :)
cotto: just change the appropriate line in .git/config to look like that
cotto: in the [remote "origin"] section
tadzik cotto_work: I think so, works for me
dukeleto cotto: there are commands to do it, but just editing the string is easier 23:29
23:29 hercynium left
cotto_work wfm 23:29
dukeleto cotto_work: i guess we can just consider tree-optimization as an HLL for testing purposes. We just want to run it against parrot master and see what smoke comes out 23:30
cotto_work yup 23:31
currently at least one test fails 23:32
plobsing whiteknight: I understand you want to eliminate initial_pf, but I don't want to follow the "extract and pick up the pieces" development model. eliminate all places that depend on it first. 23:33
dukeleto cotto_work: shall i setup a project on smolder for it?
tadzik merging it will make it available to use in Rakudo, NQP, LOLCODE and anything else?
cotto_work dukeleto: sure
though we really need a way to address the write-onliness of smolder 23:34
dukeleto cotto_work: smolder.parrot.org/app/projects/smoke_reports/9
cotto_work: yes, i am thinking hard about how to address that 23:35
who wants to add a "make smoke" target to tree-optimization ?
tadzik o/
dalek rrot/nwellnhof/platform_src: 025ff86 | NotFound++ | src/packfile/api.c:
rename local variable 'main' to avoid the warning: 'main' is usually a function
 23:36
rrot/nwellnhof/platform_src: 83cc02b | nwellnhof++ | / (29 files):
Merge branch 'master' into nwellnhof/platform_src
rrot/nwellnhof/platform_src: d2bde39 | nwellnhof++ | src/pmc/socket.pmc:
Really fix Socket PMC
tadzik or take someone with a commit bit :)
23:36 mikehh joined
cotto_work tadzik: you can always send a pull request 23:37
dukeleto tadzik: it is all you
tadzik: i await your pull request :)
tadzik or I can gist a patch and you will just curl <it> | git am :) 23:38
23:40 dmalcolm left
dukeleto tadzik: pull requests are very convenient 23:40
tadzik: because i can do a wget $pull_request_url.patch and then do "git am --signoff foo.patch" and I am done :)
tadzik: but whatever works 23:41
tadzik dukeleto: you can also wget $gist_url.patch and then git -am and you're done, it's just as nice, tried today :) 23:42
where is distutils documented?
dukeleto tadzik: perldoc distutils.pir ? 23:43
tadzik tree-optimization is using it, so I got confused. It's not like adding a target to a makefile
dukeleto tadzik: nope, it should be easier :)
tadzik hmm
dukeleto tadzik: just find another distutils project that already uses smolder
tadzik: copy and paste 23:44
tadzik: such as lua
tadzik dukeleto: that's what I'm doing ;)
oh, that'll save ack some work
23:45 kid51 joined
dalek tracwiki: v19 | nwellnhof++ | IOTasklist 23:46
tracwiki: trac.parrot.org/parrot/wiki/IOTaskl...ction=diff
tracwiki: v20 | nwellnhof++ | IOTasklist
tracwiki: trac.parrot.org/parrot/wiki/IOTaskl...ction=diff
23:46 plobsing left
whiteknight nwellnhof: ping 23:50
msg nwellnhof We were thinking about a separate Pipe PMC type to handle bidirectional communications with other programs. The ability to capture stdin and stdout, etc 23:55
aloha OK. I'll deliver the message.
tadzik ** POST smolder.parrot.org/app/projects/pro...d_report/9 ==> 200 OK (2s)
but there's nothing on smolder.parrot.org/app/projects/smoke_reports/9 23:56
dukeleto tadzik: that is happening to me too, in my branch of nqp-rx that adds a smoke target 23:57
tadzik: it usually means that smolder is dumb and returning a 200 even though something is wrong
tadzik is "Project does not allow anonymous reports" related?
kid51 cotto: ping
dukeleto tadzik: !
cotto_work kid51: pong 23:58
dukeleto tadzik: maybe
kid51 cotto_work: 2 TTs filed by Andy D
tadzik dukeleto: don't think so, Parrot project page on smolder says the same thing
kid51 TT #1950: I am currently testing his revised patch posted today.
cotto_work kid51: what about them?
kid51 If it PASSes, should it be applied -- or are we in code freeze? 23:59
trac.parrot.org/parrot/ticket/1950
patch is contained in post
dukeleto tadzik: did you give the proper project # in your distutils config ?
tadzik dukeleto: I've set all the options that lua sets

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!