Parrot 2.11.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | More gci tasks | Merge html_cleanup | Get Plumage working | Review current embedding api | Study EventHandler | Run examples_tests, test HLLs before merges
Set by moderator on 31 December 2010.
00:00 bluescreen joined, Andy left 00:02 aws joined, smash left
bluescreen msg tadzik it looks like $lex is returning null 00:06
aloha OK. I'll deliver the message.
cotto_work how nice. melange gives me a 500 when I try to accept a claim request. 00:09
00:10 aws left 00:11 bacek__ joined
dukeleto ~~ 00:12
cotto_work !! 00:13
sorear ~~
cotto_work Hmmm. That makes me look like I'm excited about something.
sorear bacek__: bacek_: bacek_at_work: no
00:16 bacek_ left 00:17 kid51 joined
bacek__ sorear, "no"? 00:24
00:24 bacek__ is now known as bacek
sorear bacek: utf8 is known to be problematic 00:27
bacek sorear, ok. But I think main problem with dalek was fixed by Infinoid. (Wrong url for CREDITS) 00:28
sorear, can you update dalek?
sorear dalek committed to a repository which is no longer used
Infinoid
I changed the CREDITS url as part of the last overhaul 00:29
bluescreen msg tadzik try with "if defined($lex) && ~lookup_class($lex) "
aloha OK. I'll deliver the message.
sorear 132 aliases in last CREDITS fetch - is it working?
bacek sorear, I dunno 00:30
00:33 whiteknight joined, bacek_ joined
cotto_work helo whiteknight 00:33
00:33 kid51 is now known as kid51_at_dinner
whiteknight hello cotto_work 00:36
00:38 bacek left
sorear bacek_: what was the last problematic report? 00:40
bacek_ sorear, karma for tadzik
sorear I need to test my changes
bacek_ sorear, last commit into cardinal repo 00:41
sorear bacek_: got it 00:46
bacek_ seen plobsing 00:52
aloha plobsing was last seen in #perl6 1 hours 5 mins ago joining the channel.
bacek_ plobsing, when PackFile_ByteCode.const_table was introduced? In dynop_mapping or earlier?
sorear bah. I can't figure out how to post stuff with curl anymore
bacek_ sorear, --data? 00:53
sorear bacek_: push.psgi thinks payload="" if I do that 00:54
bacek_ sorear, hmm. 00:55
cotto_work sorear: there's an example of how to use curl for mocking github's post-rx callbacks on trac.parrot.org/parrot/wiki/GitHubT...luginTests 00:57
01:00 kid51_at_dinner is now known as kid51
chromatic bacek_, I think earlier. 01:02
bacek_ chromatic, ok
trying to figure out when.
chromatic That seems to have been around the time we optimized hashes.
I was profiling Rakudo startup.
sorear I used --data-urlencode payload@tadzik but curl sends "payload=7" 01:04
:/
sorear checks the link 01:05
dukeleto sorear: check this out : sgn-build-64.sgn.cornell.edu:8090/p...ado-schema
bacek_ chromatic, hmm... Is this pointer actually used for anything?
dukeleto sorear: it is a jitterbug instance, which uses github post-receive hooks to send build jobs and iterates over every perl that is in perlbrew
sorear tadzik is a 1221 byte file 01:06
dukeleto aloha, jitterbug ?
aloha dukeleto: I give up.
sorear dukeleto: why should I check it out
dukeleto sorear: figured it might interest you
aloha, jitterbug is github.com/franckcuny/jitterbug
aloha dukeleto: Okay.
dukeleto sorear: it makes it easy to run test suites for perl-based projects on github 01:07
Kristaba bacek_: I will try to complete as much as possible my task tomorrow, but test I think it's impossible to fix some tests, as example some using the tailcalls (which do a strange outcome), or which testing a PIR error :( 01:08
dukeleto sorear: i am going to use jitterbug to run the Parrot test suite, and jitterbug takes care of using perlbrew to run the test suite with each version of perl that perlbrew know about
sorear "payload=`cat tadzik`" does exactly the same thing
curl--
dukeleto Kristaba: if you find that some of your task is impossible, just describe that when you submit the task
bacek_ Kristaba, just ask questions. I'll try to help.
Kristaba bacek_: And seriously, I think it would be really more understandable to add a sort of modifier to PIR subroutines to indicate that this sub is using get_params instead of the .param system
moderator Parrot 2.11.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Keep up with GCI | Get Rakudo working again | Get Plumage working 01:08
sorear it seems to think a 1221 byte blob urlencodes as "7" 01:08
bacek_ Kristaba, .param is syntax sugar for get_param 01:09
sorear ok, --data-binary worked.
dukeleto i though Plumage was working? why is "Get Plumage working" in the /topic ?
bacek_ Kristaba, docs/pdds/pdd19_pir.pod
dalek lek-poller: b72de5a | sorear++ | push.psgi:
Improve message after heuristic merge detection
 01:12
lek-poller: 39e7ab0 | sorear++ | push.psgi:
Fix UTF8 handling in push mode.
sorear I got it to generate "tadzik++" locally 01:13
chromatic bacek_, I haven't looked at it lately.
Kristaba bacek_: Oh, I didn't know that, sorry 01:15
bacek_ Kristaba, "difficult" task :)
cotto_work msg fbrito melange-- won't let me accept your claim because melange-- is returning 500 errors when I try to accept a claim through melange--. I'll take care of it when melange-- is working again. 01:18
aloha OK. I'll deliver the message.
cotto_work karma melange 01:19
aloha melange has karma of -5.
Kristaba bacek_: Yeah, I saw, but indeed I need to read more about some details, you're probably right 01:20
sorear wonders if frw et al will stay after gci
bacek_ Kristaba, welcome to parrot :)
cotto_work is hopeful 01:24
01:26 davidfetter left
Kristaba bacek_: oh, and about that I said (the "modifer" to indicate that the sub use get_params instead .param), I wanted to say that it's really strange to being obliged to add a "fake" .param when the sub use get_params explicitly 01:30
Unless get_params is really deprecated now?
So, sorry to have bothered, goodbye #parrot 01:35
01:35 Kristaba left 01:41 Matt221 joined
kid51 Has anyone experienced "500 Server Error" when hitting the Submit button for a GCI task? 01:47
Matt221 kid51: yup.
kid51 So I'm not alone ;-)
Yuki`N dukeleto, so about adding a CPAN module to Parrot. 01:48
I can't quite figure out how to rewrite the module into Parrot's existing Perl library. 01:49
kid51 Yuki`N: Generally speaking, we want to avoid adding new dependencies to Parrot.
You would have to file a Trac ticket or a post to parrot-dev to begin that process. 01:50
cotto_work kid51: yes. It accomplishes the difficult task of making me think less of Melange.
kid51 In other words, you have to have a *very* strong case for doing so.
Yuki`N kid51, well maybe I can just add this functionality into the perllib, cpansearch.perl.org/src/BTROTT/Filt.../Handle.pm
cotto_work Yuki`N: indeed. What's the benefit?
kapace__ I wonder how much better a system GCI students could create instead of melange... 01:51
01:51 Matt221 left
cotto_work I wonder if it's worthwhile to try to argue sanity to the Melange dev team. 01:52
kid51 GSOC project, anyone?
cotto_work I lose a non-trivial chunk of time between its braindead ui decisions and its downtime. 01:53
kapace__ cotto_work++ well said 01:54
kid51 Yuki`N: What Perl 5 are you writing for GCI that would entail use of that module?
01:59 bacek__ joined
dukeleto Yuki`N: why do we need the module? 02:00
cotto_work: that is why i write everything down in a text file and just paste into melange. It has lost too much of my work to do anything else. 02:01
cotto_work: and It's All Text makes that seamless 02:02
cotto_work code.google.com/p/soc/issues/list 02:03
kid51 Yuki`N: I just tried to install that module from CPAN.
When I ran 'make test', I got: make: *** [test_dynamic] Segmentation fault
So the answer is: No.
02:04 bacek_ left 02:06 Matt221 joined 02:08 bacek_ joined 02:14 bacek__ left
plobsing ~~ 02:15
Yuki`N dukeleto, well it allows filtering STDOUT/ERR, so that any writes to that filehandle can be redirected to a file. 02:16
Best choice as I see it; replicating tee-like behavior.
kid51 Yuki`N: As noted above, the module segfaults. 02:18
It has not been actively maintained in > 10 years.
It does not meet contemporary standards for quality of CPAN distributions. 02:19
dukeleto Yuki`N: we don't need to filter. Just dupe the filehandle and redirect the output to config.log
02:19 bacek__ joined
kid51 We may already have some code that does stuff like that under lib/Parrot/Configure/*. 02:19
But, Yuki`N, you still haven't told us *where* you want to use it.
dukeleto cotto_work: be sure to turn off the wysiwyg editor in your melange prefs. I find that is a lot of the problem. 02:20
cotto_work dukeleto: excellent idea
kid51 must sleep
02:20 kid51 left 02:24 bacek_ left
cotto_work goes home 02:34
dalek TT #1700 closed by bacek++: OpLib caching doesn't survive after initializing of parrot trace. 02:36
TT #1700: trac.parrot.org/parrot/ticket/1700
bacek__ cotto_work, are you still there?
dalek rrot/opmap_aware_pmcs: 7a05f9c | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Remove utterly wrong code from PackfileBytecode.get_pointer.
bacek__ cotto, you'll love my latest commit into opmap_aware_pmcs :) 02:37
02:37 bacek__ is now known as bacek
whiteknight bacek++ 02:39
02:40 Yuki`N left, fbrito joined 02:41 mtk left 02:43 whiteknight left 02:47 mtk joined 02:58 Yuki`N joined
Yuki`N dukeleto, I understand how you can replace STDOUT but how will duplicating it allow you to write to 2 filehandles with one? 03:01
cotto ~~ 03:09
bacek, I saw you took out a bit of code there 03:10
bacek cotto, indeed 03:11
cotto just received a shiny new ssd hdd for his laptop. Now if I could only find tuits to install it... 03:14
plobsing Yuki`N: I am also curious as to how that can be accomplished with just a dup. tee from gnu coreutils has to manage the file descriptors manually.
03:19 bluescreen left 03:28 bacek_ joined 03:31 bacek__ joined 03:33 bacek left 03:36 bacek_ left
atrodo cotto> That's the kind of shiny I find tuits for easy 03:40
03:40 bacek_ joined 03:46 bacek__ left 03:47 bacek_ left, aantn joined
Yuki`N Well I got something similar to redirection for stdout working but somehow it makes gcc fail. 03:47
I think forked processes don't like my tied filehandles or something. :/ 03:48
plobsing Yuki`N: you'll have to do a dance with dups so child processes see an appropriate filedes 03:50
s/you'll have/you might have/
Yuki`N Oh that's wonderful. 03:51
Untying and re-tying stdout/err before/after calling a child process?
Sounds like fun.
._.
I don't know. 03:52
It seems like an awful amount of trouble.
plobsing I figured you'd set up a pipe to a tee, and then move that over the stdout
that's how I'd do it. but that might be a bit of a hack. 03:53
Yuki`N I was going for the portable implementation, though.
We might be able to give it a try with a CPAN module, but that seems unnecessary for just a config script. 03:54
plobsing Yuki`N: you can probably make a tee emulator with perl's fork() (or fork emulation)
Yuki`N Hmm.
plobsing again, I might be feeding you horrible ideas. 03:55
Yuki`N I don't know, I will sleep on it. 03:56
03:56 Yuki`N left
cotto atrodo, in that case do you mind installing my new hdd? 03:56
atrodo I would, but it's an awful long drive 03:57
04:03 aantn left
particle cotto; can you clone your hdd to the ssd via usb? 04:04
cotto not directly 04:05
pmichaud I think I've found why Rakudo and NQP are failing with Parrot trunk 04:06
cotto it'll be a bit of a pain
pmichaud Do you (1) want me to file a new ticket with the bug, (2) respond to TT #1704, or (3) add the information to TT #1932 (which notes that Rakudo/NQP no longer run with parrot head)
plobsing pmichaud: lets track this @ TT #1932 04:07
pmichaud okay, posting in a moment
here's the short explanation: gist.github.com/765914 04:09
the :load sub should not be invoked here
plobsing :load subs should be invoked whenever pbc is loaded 04:10
pmichaud no, they should be invoked when a .pbc is loaded via load_bytecode 04:11
That's the way it's been defined like.... forever
see PDD 19
cotto pdd19 concurs
plobsing huh. that's the one doc that actually spelled out what it did like that. 04:12
the comments/docs I read just mumbled stuff like "pbc load" 04:13
pmichaud a fuller explanation is now at trac.parrot.org/parrot/ticket/1932#comment:3 04:14
plobsing which I took to mean "whenever someone puts new bytecode into parrot"
pmichaud: if those subs are to make sure that the same sub gets run both when :main is run and when not, then shouldn't they just be marked :init :load? 04:16
I will of course, work to make it the same as it was, but it seems to me to be solving the problem backwards. 04:17
pmichaud plobsing: that's not what those subs are for
if I create (pre-compile) a module in Perl 6 or NQP, then it later gets loaded via the "load_bytecode" op. 04:18
if a module is loaded via the load_bytecode, I have to have some mechanism to cause its mainline to be executed -- currently the :load flag is what does that 04:19
plobsing ok what is "mainline"? is that not just :main by another name?
pmichaud :main is the sub that is executed when invoked from the command line
here, I'm already inside of a Parrot program (e.g., NQP or Rakudo itself) that is running
unless you feel that load_bytecode should automatically invoke :main 04:20
plobsing of course not. it should provide access to the :main sub, with which the loading program can do whatever it likes. 04:21
04:21 aantn joined
plobsing including ignoring, invoking once, and invoking multiple times. 04:21
pmichaud I don't know if it does that yet, but historically load_bytecode has not provided any way to do that
(just checked -- it doesn't appear to do that yet) 04:22
plobsing pmichaud: this is true.
pmichaud so, the mechanism we've used is the :load flag 04:23
and at the end of a module, we create a :load (but not :init) sub that invokes the module's mainline for us when the module is loaded via load_bytecode
plobsing pmichaud: but the load flag doesn't work in other contexts. what if I'm an embedding application that wants to run a modules :main?
pmichaud I'm not saying that load_bytecode isn't misdesigned. 04:24
I've argued for years that load_bytecode should return a "handle" to whatever was loaded so that the caller can decide what do with it and how to invoke it.
I could never convince allison++ why it was needed.
plobsing pmichaud: PackFile PMCs provide those handles. we are getting there. 04:25
I may have just done things in the wrong order.
pmichaud anyway, I'm somewhat blocked on my tasks (benchmarking nqp and rakudo) until parrot head is working again for them. 04:26
and jnthn++ is likewise blocked on the things he was working on, iirc
plobsing now that you've pinpointed the problem, it shouldn't be long 04:27
pmichaud good, glad to help :)
afk for a while 04:28
plobsing the good news - I fixed NQP 04:41
dukeleto plobsing++
plobsing: bad news?
plobsing the bad news - I broke PCT in the process
I feel like I've knocked over a leaning tower of abstractions
04:42 bacek_ joined
dukeleto plobsing: good lord, chap. 04:46
plobsing: what broke in PCT? 04:47
plobsing not entirely sure on that one. t/compilers/pct/complete_workflow.t has major fail ATM though. 04:48
dalek rrot/opmap_aware_pmcs: 6e990fc | bacek++ | src/pmc/packfilebytecodesegment.pmc:
First baby step for generating ByteCode
 05:00
rrot/opmap_aware_pmcs: 21950b1 | bacek++ | src/pmc/packfile (2 files):
Add skeleton how to pack PackfileBytecode
bacek_ cotto, around? 05:04
05:08 bacek__ joined 05:10 bacek joined 05:13 bacek_ left
dukeleto bacek: how are you feeling? 05:13
05:14 Matt221 left
bacek dukeleto, pretty much ok. Can be discharged from hospital right now. Unfortunately I booked my flight for tomorrow. 05:14
So, I have to spend one more day here
05:15 bacek__ left
dalek rrot: 9dda543 | plobsing++ | src/packfile/api.c:
only run :load subs on load_bytecode
 05:18
rrot: 533bde3 | plobsing++ | runtime/parrot/library/PGE/Perl6Grammar.pir:
Revert "mark PGE main subs :main"

This reverts commit 2b0abde5604f07f58bae5049142f79e5d9afba16.
rrot: 7529cd3 | plobsing++ | compilers/tge/TGE/Compiler.pir:
Revert "tag *every* tge sub :main. emulates old broken pick-first-sub behaviour"

This reverts commit 0ee91bbca15d0d8af52053919547d6b726943d35.
cotto bacek, hi 05:20
bacek cotto, aloha
cotto I see you're hacking on the packfile pmcs
it gladdens my heart
bacek cotto, yes. I have few free hours before leaving hospital. 05:21
Looks about good size task to finish it.
cotto I'd forgotten how productive bacek-hours could be
bacek cotto, :)
cotto wb, magical coding robot
bacek there is only one small problem... No coffee 05:22
And I quit smoking.
cotto melange--
stupid 500s
05:22 particle left
bacek So, my productivity isn't so high. 05:22
05:23 fbrito left
dukeleto bacek: it is good to reset your coffee tolerance 05:24
cotto not pleasant, but good 05:25
bacek dukeleto, heh. About a week already. I almost ready to kill for a good cappuccino with a good cigarette. 05:26
dukeleto bacek: mmmmmmmmm
s/cigarette/cigarillo/
bacek: what state is cardinal in now? 05:27
bacek s/coffee/single malt/, s/cigarette/havana cigar/ :)
dukeleto, same as before. It have to updated to nqp-rx. Too beg task for me for couple of days.
dalek rrot: cbccec7 | plobsing++ | src/packfile/api.c:
[codingstd] fixups for recently added packfile_main() accessor function
bacek (And I don't like ruby :)
plobsing OK, I think the emergency is over now. sorry for the interruption. 05:28
dukeleto plobsing: what emergency ? Does stuff work now? 05:29
plobsing yes. all of it.
pct, nqp, rakudo
cotto verifies 05:30
plobsing I didn't have time to spectest rakudo, but it works where it was failing before.
cotto plobsing, thanks. I'm happy to verify that rakudo's spectest works. 05:31
05:35 bacek_ joined 05:40 rurban_ joined, bacek left 05:43 rurban left, rurban_ is now known as rurban 05:48 particle joined 05:54 bacek__ joined
cotto watches #perl6 with great amusement 05:57
05:59 bacek_ left 06:02 bacek_ joined
dalek rrot/opmap_aware_pmcs: 3ccc55f | bacek++ | src/pmc/oplib.pmc:
Add OpMap.get_pointer to be used in OpMapping.get_pointer
 06:04
rrot/opmap_aware_pmcs: 83957ef | bacek++ | src/pmc/packfileopmap.pmc:
Return mapping value, not mapped value for op.
rrot/opmap_aware_pmcs: ab807d0 | bacek++ | src/pmc/packfileopmap.pmc:
Initial shoot on PF_BC_OM creation
bacek_ cotto, ping
cotto bacek_, pong
bacek_ ~/src/parrot (opmap_aware_pmcs)$ ./parrot examples/pir/make_hello_pbc.pir
Hello World
Segmentation fault
kind of progress :)
cotto not really. I got a spectacular segfault too.
bacek_ At least it _runs_ already :) 06:05
cotto oh
that is progres
s
bacek++
bacek_ ~/src/parrot (opmap_aware_pmcs)$ ./parrot generated_hello.pbc
Hello World
Segmentation fault
this is more clear example :)
06:07 bacek__ left 06:12 dylan_ joined
dalek rrot/opmap_aware_pmcs: 8c7adc4 | bacek++ | t/pmc/packfileopmap.t:
Add more tests for OpMap
 06:22
rrot/opmap_aware_pmcs: abb9534 | bacek++ | examples/pir/make_hello_pbc.pir:
Fix index of FIA for get_returns
bacek_ cotto, done
~/src/parrot (opmap_aware_pmcs)$ ./parrot examples/pir/make_hello_pbc.pir 06:23
Hello World
~/src/parrot (opmap_aware_pmcs)$ ./parrot generated_hello.pbc
Hello World
cotto it's really nice to have you back in action
bacek_ :)
I don't like PackfileConstTable API. It's really weird. 06:24
cotto at least the tests survived
bacek_ because API is way too stupid...
cotto which part of the api? 06:26
06:36 bacek__ joined 06:42 bacek_ left 06:43 bacek_ joined
dalek rrot/opmap_aware_pmcs: 10cbb73 | bacek++ | src/pmc/packfileconstanttable.pmc:
Add PF_ConstTable.push_* functions for convinience.
 06:48
rrot/opmap_aware_pmcs: 10d5f6f | bacek++ | examples/pir/make_hello_pbc.pir:
Simplify make_hello_pbc
06:49 bacek__ left
dylan_ hi parrot people. i'm having a lot of trouble getting through the doco on the PAST. 06:55
is it using perl6?
bacek_ dylan_, no, PAST is part of PCT/nqp-rx 06:56
dylan_ are there better examples than the squaak
bacek_ dylan_, squaak is really good example.
dylan_ i got quit lost around ch. 3
bacek_ You probably talking about generating PAST from your Grammar, aren't you?
dylan_ yes 06:57
the whole things a bit mysterious.
bacek_ hmm
Let me check 06:58
dylan_ do you suggest reviewing PCT doco first?
bacek_ dylan_, no.
What is your questions about chapter3? 06:59
dylan_ trying to work out this idea od attributes, e.g. like :namespace
where do i learn the syntax/semantics for that? 07:00
bacek_ dylan_, erm. Where do you read squaak example? 07:01
dylan_ i'm here: en.wikibooks.org/wiki/Parrot_Virtua...irst_Steps
on the wikibook.
bacek_ let me check how up-to-date is it 07:02
dylan_ problem 3. see the syntax? its a mystery.
bacek_ dylan_, it's outdated by about 1.5 years. Things are changed :)
dylan_ oh bugger. 07:03
where to go?
bacek_ examples/languages/squaak/doc/tutorial_episode_3.pod 07:04
I'm trying to find HTML version of it somewhere
docs.parrot.org/parrot/2.11.0/html/...3.pod.html 07:05
here we go
And examples/languages/squaak contains whole Squaak implementation 07:07
Hope it will help :)
dylan_ ok. i was there. but my question was this line:
make PAST::Var.new( :name(~$/), :scope('package'), :node($/) )
bacek_ Ah. 07:08
dylan_ where do i lear about :name, :scope, etc?
bacek_ Ok. This is PIR
No, NQP
Not Quite Perl8
Perl6
sorry
:scope('package') is named argument to function 07:09
dylan_ ok, i'll read there now. thaks muchly.
bacek_ In Perl5 world it will be something like PAST::Var->new( name => ''.$/, scope => 'package', node => $/)
dylan_ hah. thought so. i'm way out of date. 07:10
bacek_ dylan_, good luck with squaak. About 3 years ago it was one the reasons for me to join parrot :) 07:11
dylan_ i'm actually wanting to use these tools to make a translator first (without making a full blown language). do you think the PCT can be used to do thar easily? 07:13
07:19 bacek_ left
dukeleto dylan_: what kind of translator ? 07:25
dylan_ a translator to convert a language i wrote called foo into fortran95. (don't laugh). my strategy is to use the actions to emitprints, instead of low level PIR or whetever you call it. do you think PCT can handle that? 07:29
just now, i have more than quarter million lines of code in this language foo. 07:30
cotto dylan_, interesting. Is this for work or fun? 07:38
dylan_ both. must update my crappy translator.
cotto how important is speed? 07:42
dylan_ not that important. more important is maintainability and possibility to convert to c++. 07:47
cotto ok. Translating 250,000 lines of code will definitely take a while to run. nqp-rx is a powerful tool but the code it generates isn't especially fast. 07:49
dylan_ ok. typically with my roll-my-own translator it takes 10 minutes with 4 fast processors. but in development cycle you don't usually translate all that, just the bits you change. 07:50
cotto pmichaud had a proof of concept lolcat translator, but I don't think it ever got posted 07:51
If you don't mind occasionally running into bumps, Parrot and nqp-rx might be an interesting way to approach your problem. It's still in a very active stage of development. 07:56
dylan_ yes, its very lovely, that's for sure, i'm going to persist several days yet. thanks guys. 07:57
cotto great. You're always welcome to ask questions if you run into something that surprises or confuses you.
dalek rdinal: c72a79c | (Gerd Pokorra)++ | src/parser/actions.pm:
apply patch from "bluescreen10"
 08:01
08:02 bacek joined
tadzik hello 08:06
08:07 chromatic left
cotto hi tadzik 08:07
also, good night
cotto sleeps
bacek cotto, wake up! 08:08
dalek rrot/opmap_aware_pmcs: fcedaff | bacek++ | / (2 files):
Fix OpMap.load_lib
rrot/opmap_aware_pmcs: 7654375 | bacek++ | t/pmc/packfileopmap.t:
Add (failing) test for mapping ops from dynext
rrot/opmap_aware_pmcs: c6c2678 | bacek++ | src/pmc/packfileopmap.pmc:
Fix mapping ops from dynext
bacek opmap_aware_pmcs is ready for merging :)
sorta
tadzik yay, cardinal works! 08:10
bluescreen++
cotto bacek, it worked 08:13
bacek msg plobsing why do you need bytecode->op_func_table and bytecode->op_info_table? Or they are filled during loading of PBC?
aloha OK. I'll deliver the message.
bacek cotto, of course! :)
cotto waking someone up over irc is impressive 08:14
I really need to sleep, but it's awesome that you got that working so quickly.
bacek it's just another way of communication.
I'm going to merge it back to master tomorrow morning
It's self-contained piece of work anyway 08:15
cotto ok. It can't mess anything up.
next stop, PIRATE
'night
bacek night 08:16
dalek rrot/opmap_aware_pmcs: 16c87fb | bacek++ | / (1958 files):
Merge branch 'master' into opmap_aware_pmcs

Conflicts:
  \tMANIFEST		 08:18
08:20 theory left, M_o_C joined 08:26 fperrad joined 08:29 bacek_ joined 08:34 bacek left
dalek rrot/opmap_aware_pmcs: fe5dbbe | bacek++ | src/pmc/packfileopmap.pmc:
Fix function names after merging with master
 08:40
rrot/opmap_aware_pmcs: 27d29b7 | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Fix compiler warning
rrot/opmap_aware_pmcs: c2db800 | bacek++ | t/pmc/class.t:
We have 2 more core PMCs. Adjust default value for new Classes.
rrot/opmap_aware_pmcs: 4f97261 | bacek++ | t/pmc/opcode.t:
Update test to modern OpCode.
dukeleto builds plumage 08:42
msg sorear i think plumage needs to be added to the file that tells dalek about repos 08:46
aloha OK. I'll deliver the message.
dukeleto Plumage works. I don't understand the /topic 08:49
dukeleto goes to bed
dalek nxed: r726 | NotFound++ | trunk/winxedst1.winxed:
optimize string constant indexed with int constant
 08:55
rrot/opmap_aware_pmcs: 5bc87fe | bacek++ | src/pmc/packfile (2 files):
Fix codetest - proper coda
rrot/opmap_aware_pmcs: 96d9d48 | bacek++ | src/pmc/ (2 files):
Fix codetest - space
rrot/opmap_aware_pmcs: 795f10e | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Fix codetest - POD syntax.
rrot/opmap_aware_pmcs: 263f60d | bacek++ | src/pmc/packfile (2 files):
Fix codetest - trailing space
rrot/opmap_aware_pmcs: 74b1055 | bacek++ | / (2 files):
Re-enable examples/make_hello_pbc
 08:58
08:59 cogno joined
bacek_ seen tcurtis 08:59
aloha tcurtis was last seen in #parrot 12 hours 40 mins ago joining the channel.
bacek_ kids... 09:00
msg tcurtis What about resurrecting tree-optimizations? I do need it :)
aloha OK. I'll deliver the message.
09:07 bacek_ left 09:19 bacek_ joined 09:28 cogno left, PacoLinux left
dalek rrot/opmap_aware_pmcs: 7f66d12 | bacek++ | MANIFEST:
Regenerate MANIFEST
 09:30
09:51 rfw left 09:57 bacek__ joined 10:02 bacek_ left 10:05 cogno joined
dalek r: 91c4298 | bacek++ | src/PIR/Grammar.pm:
Use :qq to handle double-quoted strings.
 10:12
r: 0312732 | bacek++ | README:
Merge branch 'master' of github.com:parrot/pir
r: d44f759 | bacek++ | / (2 files):
Move hacks into src/hacks.pir
r: e134c19 | bacek++ | / (3 files):
Fix creating of OpLib
rrot: c13f8ca | NotFound++ | t/pmc/bytebuffer.t:
test ByteBuffer get_chars method out of bounds condition
 10:15
moderator Parrot 2.11.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Keep up with GCI | Get Plumage working 10:18
dalek rrot: 59d037f | NotFound++ | runtime/parrot/library/Test/More.pir:
fix message and diag of Test;More throws_type
 10:20
r: 86d8bc3 | bacek++ | / (2 files):
Move generated files into gen directory
 10:23
10:27 cogno left 10:28 mtk left
tadzik moving a compiler from PCT to NQP-rx is something more than changing "is PCT::Grammar" to "is HLL::Grammar" and making that work? 10:31
moritz the "making that work" step might be non-neglible 10:32
tadzik I know that
dalek nxed: r727 | NotFound++ | trunk/t/postincdec.t:
improve post ++ and -- basic tests
 10:34
10:34 mtk joined 10:35 Themeruta joined, Themeruta is now known as NotFound_gf
NotFound_gf hi 10:35
tadzik hi :) 10:36
for the good start I moved LOLCODE to nqp-rx :) 10:47
moritz does it work?
tadzik seems so
although when running setup.pir I don't see it compililng the grammar
and when I removed gen_grammar.pir it doesn't rebuild it and the thing doesn't work at all :) 10:51
moritz huh 10:52
tadzik looks like the old setup.pir doesn't handle the situation much
moritz but once it's built, the test suite passes?
tadzik it's not built anymore :) But when it did, it fails the same tests as before 10:53
wowowow. No, it doesn't work at all :) 10:54
Parse errors: Unknown TAP token: "Cannot find TOP regex in Grammar"
now Unknown TAP token: "Parent isn't a Class.". That seems familiar cardinal fails identically 10:55
must be some deprecation in the new object model or something
shame that lolcode has no classes except the grammar. 10:56
moritz I remember getting that error too when I tried the port 10:57
tadzik int eresting; I'll try making a new lolcode with mk_language_shell and moving the files, so I'll have a shiny new setup.pir working with nqprx 10:58
11:00 cogno joined 11:22 aantn left 11:25 cogno left 11:30 M_o_C left
dalek rrot/opmap_aware_pmcs: 81e718d | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Add PFBC.get_integer_keyed_int for introspection purposes
 11:32
r: 8e53a54 | bacek++ | src/POST/Compiler.pm:
Remove fixups. They are gone.
 11:34
r: ea3c470 | bacek++ | src/hacks.pir:
Remove FixupTable and change RawSegment to BytecodeSegment
r: 314dec5 | bacek++ | src/POST/Compiler.pm:
Start refactoring PBC emitting.

Rename op-part generating methods to to_op from to_pbc.
r: 76e2fb8 | bacek++ | src/POST/Compiler.pm:
Fix generating Op in PCC call. NQP is really NQ
11:39 redicaps joined 11:44 aantn joined
tadzik Unable to obtain PAST from Regex;Match -- does this sound familiar to you? 11:52
Coke ... whoa, bacek is in hospital? 11:58
bacek__ Coke, sorta 12:05
12:05 bacek__ is now known as bacek
bacek tadzik, forgot to add proper method into Actions 12:06
tadzik bacek: any chance to investigate which one? 12:12
bacek tadzik, try --target=past
You should see something,
Or may be not.
tadzik no, gives the same errors 12:13
Rakudo's <?DEBUG(1)> would be useful
12:17 bacek_ joined, darbelo joined
bacek_ tadzik, <?DEBUG> than 12:19
in nqp
tadzik oh 12:20
Coke msg dukeleto plumage's sanity.t fails nearly every test for me. opened a ticket. 12:21
aloha OK. I'll deliver the message.
Coke aloha, ping?
aloha Coke: No clue. Sorry.
Coke slowbot.
anyone know how to do "prove -v" with setup.pir ? 12:22
12:22 bacek left
tadzik there's enormormous amount of calls to ws and ww 12:24
dalek rrot/opmap_aware_pmcs: 55002ed | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Add main_sub to BytecodeSegment
 12:32
12:39 aantn left
dalek r: 74a163e | bacek++ | src/POST/Compiler.pm:
Use Bytecode instead of Raw segment.
 12:40
r: 3286f15 | bacek++ | src/POST/Compiler.pm:
Small fixes for handling to_pbc(POST::Op)
r: 298813f | bacek++ | src/hacks.pir:
Fix PackfileConstantTable.push hack
r: e1e5a28 | bacek++ | src/POST/Compiler.pm:
Add stub to_op(Label).
r: a3a157c | bacek++ | src/POST/Compiler.pm:
Bring FIA back for default returns
r: 3a28ad4 | bacek++ | src/POST/Compiler.pm:
Fix generating pbc for POST::Op
r: b028854 | bacek++ | src/POST/Compiler.pm:
Select main sub in same way as imcc
12:41 contingencyplan left 12:42 contingencyplan joined
NotFound_gf Coke: looks like a shebang problem: the shell can't locate parrot-nqp just by its name. 12:42
arnsholt tadzik: IIRC the unable to obtain PAST message means that the ast property of the Match object contains something other than a PAST node
In my code it usually means that I've brainoed and made an action method return the AST object instead of make()ing it
12:46 aantn joined
tadzik yeah, that's fixed now. Btw, plenty of compilers tend to have their own literal methods, written in pir. What's a good way to include those in the grammar? 12:46
I tried wrapping them in Q:PIR and adding to grammar.pm, got some strange errors, but is there a better way? 12:47
12:48 bluescreen joined 12:49 cogno joined
bacek_ tadzik, why do you need PIR in Grammar??? 12:49
arnsholt Not sure what you mean by "literal method"
tadzik bacek_: some special quote tokens as I see, cardinal has them too 12:50
bacek_ tadzik, put them into separate pir file. Same way as NQP did. 12:51
bacek_ sleep
it's almost tomorrow again
tadzik but how would setup.pir know about them?
moritz it doesn't need to
dalek r: 3347e7d | bacek++ | src/POST/Compiler.pm:
Partially migrate sub invocations to new Bytecode
bacek_ msg cotto PIRATE passing 03-pbc/01-basic test now.
aloha OK. I'll deliver the message.
moritz they are just .include'ded in Compiler.pir or so
tadzik I see 12:52
12:53 bacek_ left 12:58 aantn left, aantn joined 12:59 lucian joined
darbelo dukeleto: ping 13:03
13:04 cogno left 13:05 cogno joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#2044) fulltest) at 2_11_0-689-g59d037f - Ubuntu 10.10 i386 (g++-4.5) 13:10
pmichaud 12:47 <tadzik> I tried wrapping them in Q:PIR and adding to grammar.pm, got some strange errors, but is there a better way? 13:14
one can always put the PIR directly into (grammar) methods
method xyz() { Q:PIR { ... } }
or even
method xyz() { pir::op(...); pir::op(...); } 13:15
tadzik pmichaud: the first example was my first trial. The .param and .return will not work anymore, no? 13:16
moritz I'm taking a very wild guess here, but I think that's not the actual problem
13:16 cogno left
pmichaud tadzik: you can put params there also 13:16
moritz but rather that the interaction of hand written parsing methods with regexes/cursors has changed
pmichaud method xyz($x) { ... }
moritz: yes, you may be correct :-) 13:17
I'm only going on the question I have before me :)
dalek rrot: 4ff4ae3 | NotFound++ | runtime/parrot/library/LWP/Protocol.pir:
fix LWP;Protocol request without content
pmichaud is there a github repo I can look at? 13:20
13:20 cogno joined
tadzik strange how now yarn_literal.pir is not included anywhere, and the compiler does not complain about using it in a grammar 13:20
pmichaud afk, taking kids to school -- bbiaw 13:21
NotFound_gf tadzik: using a file in a grammar?
tadzik NotFound_gf: well, using the yarn_literal token 13:22
NotFound_gf tadzik: A sub?
tadzik NotFound_gf: github.com/tadzik/lolcode-nqprx/bl...ar.pm#L140 13:24
NotFound_gf tadzik: if that rule is assumed to be a pir .sub, pir doesn't care if doesn't see it, it will be looked up at runtime. 13:26
13:28 lucian left
tadzik I see 13:30
Coke NotFound_gf: any test that expects a bare shebang like that to work is borked. 13:33
13:34 cogno left
pmichaud why is yarn_literal being written in PIR, ooc? 13:37
13:39 cogno joined
tadzik no idea 13:39
13:42 rurban_ joined
pmichaud I mean, I understand why it was done that way before nqp-rx existed, but HLL::Grammar provides some builtins for parsing quoted strings 13:44
13:45 rurban left, rurban_ is now known as rurban 13:46 PacoLinux joined 13:47 fbrito joined, PacoLinux left 13:48 PacoLinux joined 13:49 whiteknight joined
Coke when in doubt, bitrot. 13:50
13:50 PacoLinux left
whiteknight good morning, #parrot 13:50
13:50 PacoLinux joined 13:51 PacoLinux left
atrodo Sometimes, I enjoy pre-bitrotting my code to save me the hassle in the future 13:51
13:51 PacoLinux joined 13:52 PacoLinux left, PacoLinux joined
dalek nxed: r728 | NotFound++ | trunk/examples/ajax.winxed:
delete no longer needed chunked decoding from example ajax
 13:54
tadzik does the '{*} #= declare' syntax still work in nqp-rx? 13:55
13:57 aantn left
pmichaud I know it's no longer acceptable Perl 6, so it's being phased out 14:00
I think nqp-rx still supports it, though.
yes, it does support it still 14:01
14:25 cogno left 14:38 redicaps left 14:40 aantn joined 15:03 Psyche^ joined 15:04 cogno joined, cogno left, cogno joined 15:06 Patterner left, Psyche^ is now known as Patterner 15:08 bluescreen left
dalek rrot: 9d13ffa | Whiteknight++ | / (2 files):
export the function for separating out the path extension
 15:12
15:23 bluescreen joined 15:27 dylan_ left, plobsing left, gbacon joined
TimToady note also that we've since stolen the {*} notation for use as the "you are here" in protos 15:31
whiteknight what is "you are here"? 15:32
moritz it's basically a term for "start your real work at this point" 15:34
TimToady where the multis are notionally called from the proto, in the new proto semantics
moritz feels that TimToady did a much better job describing {*} 15:35
TimToady so a proto can specify wrapping behavior around its multi dispatch
moritz but then he's the one who invented it in the first place :-)
TimToady {*} basically stands for the multi dispatcher itself
15:36 cogno left
TimToady so proto foo () {*} can generally be optimized away to the current multi semantics 15:36
but since protos are lexically scoped, this gives us a way to name multi dispatchers
so when you say &foo, it's really the name of the dispatcher, as cloned into the current lexical scope 15:37
hence represents all the currently visible candidates
15:38 lucian joined
TimToady rakudo has not yet switched to that viewpoint, but jnthn's 6model will use the new scheme 15:38
to get a clean semantics we're banking on the idea that sub invocation will become much cheaper, and aggressive inlining is usually possible 15:39
15:42 aantn left
whiteknight okay. Thanks 15:43
dalek rrot: ec9e3a0 | Whiteknight++ | / (2 files):
update OS PMC to do PATHEXT matching on Win32 to determine executability. This completely ignores newer Win32 permissions models, but is a good start and gets this test passing. We can cover the more advanced cases later, if necessary
 15:47
15:49 theory joined
dalek TT #731 closed by whiteknight++: can't tell if an OS file is executable from parrot 15:54
TT #731: trac.parrot.org/parrot/ticket/731
16:01 gbacon left
dukeleto ~~ 16:09
16:11 Kristaba joined
dalek rrot: c834f3a | mikehh++ | src/dynpmc/os.pmc:
add func docs, remove trailing whitespace
 16:16
16:20 hercynium joined
dukeleto fperrad++ # beating me to fixing the plumage test 16:24
16:26 gbacon joined
dalek rrot: 362335a | Whiteknight++ | / (2 files):
remove unused multis from integer.pmc, which fixes one TODO'd test
 16:29
rrot: 851d079 | Whiteknight++ | src/dynpmc/os.pmc:
Merge branch 'master' of github.com:parrot/parrot
dukeleto whiteknight: have any ideas for more GCI tasks? I am kind of burned out. 16:31
tadzik dukeleto: I'm sharpening my pge-to-nqprx skills porting lolcode
moritz fears porting languages to nqp-rx is too hard for students 16:32
dukeleto darbelo: pong
moritz: i would really like to provide some utility that helps convert PGE langs to nqp-rx 16:33
tadzik: can you please document what you have to do? Those instructions are very valuable for others.
tadzik: it would make a *great* blog post
darbelo dukeleto: I just noticed the gitorious plumage repo is falling out of date, do we have any plans for it? 16:34
dukeleto darbelo: it is no longer canonical
darbelo: it is just an out-of-date mirror now
whiteknight plumage is on parrot/plumage in github
dalek rrot: 2d13d16 | Coke++ | runtime/parrot/bin/prove.pir:
this script is no longer called tapir.
darbelo I know, but stale mirrors are a pain, I'd rather kill it or make it an updated mirror, 16:35
dukeleto darbelo: it is slightly confusing though.
darbelo: i agree
whiteknight we don't have the authority to kill it
japhb owns it
we can only work with what we have
dukeleto darbelo: japhb started plumage on gitorious, so it is his pony
darbelo: but if you contact him and tell him what is up, i am sure he will be ammenable to fixing the situation
darbelo: i think he has just been really busy
16:37 cogno joined 16:41 chromatic joined
tadzik dukeleto: I must first figure out what decisions were right :) 16:44
dukeleto: has there been any successful attempts so far?
dalek TT #1887 closed by whiteknight++: Integer PMC i_multiply bug
TT #1887: trac.parrot.org/parrot/ticket/1887
TT #710 closed by whiteknight++: IO - On Win32, huge strings printed by Parrot do not appear in the console ...
TT #710: trac.parrot.org/parrot/ticket/710
dukeleto tadzik: not sure, good question. 16:45
16:46 lucian left
Coke how does one re-open a ticket on github? 16:49
ah, found it.
dukeleto Coke: what did you break now? 16:53
16:54 cogno left
whiteknight msg Andy can you please take a look at TT #152? In most cases I think we can remove PARROT_CATCH_NULL with no issues since it's been the default for so long. However, in include/parrot/interpreter.h I see something about SPLINT, and I figure I should ask you about it first. Thanks 16:58
aloha OK. I'll deliver the message.
Coke dukeleto: fperrad's fix doesn't make those tests pass. 16:59
(on plumage)
dukeleto Coke: does "make test" pass for you? 17:00
Coke: parrot setup.pir test 17:01
Coke: is what I meant
/home/leto/git/plumage/plumage.c:4226: undefined reference to `Parrot_api_make_interpreter' 17:02
that wasn't happening last night
whiteknight anybody here with a working rakudo and a few minutes to investigate a bug?
Coke dukeleto: ... no. hence the ticket.
dukeleto Coke: it was working for me last night 17:03
Coke yes, and it wasn't working for me.
whiteknight dukeleto: try a realclean.
Coke which is why i was trying to generate verbose output about what was expected and what was got (which doesn't seem to be generated).
whiteknight Parrot does export Parrot_api_make_interpreter
17:04 M_o_C joined
dukeleto whiteknight: realclean in plumage? 17:04
whiteknight: i assume you mean realclean in my parrot
whiteknight dukeleto: either
dukeleto whiteknight: which i will try
whiteknight I mean, after the embed_api merge, libparrot does export that symbol 17:05
Coke why would realclean in the build dir of parrot help plumage running against an installed parrot?
dukeleto whiteknight: setup.pir doesn't have realclean, but "parrot setup.pir clean" does not work either
Coke oh, that's something else.
nevermind.
fbrito we have some impossible GCI tasks: www.google-melange.com/gci/task/sho...9329282870 and www.google-melange.com/gci/task/sho...9360922947
Coke dukeleto: "parrot setup.pir clean" seems to remove files, yes.
whiteknight if your installed parrot is older than the embed_api merge, you need to install a more recent version
dukeleto fbrito: looking 17:06
whiteknight there's no real way to test the readline_interactive method on FileHandle
dukeleto fbrito: if you make it to 100%, than that counts
whiteknight I don't know what the problem is with OrderedHash
fbrito OrderedHash PMC test coverage is 92.7%. 17:07
dukeleto fbrito: but if there are static functions in OrderedHash, then those will never show up as covered
whiteknight fbrito: make it 100%
dukeleto fbrito: is there anything in orderedhash that looks uncoverable ? such as dead code or static functions ?
fbrito: i can change the task to say "increase 5%" if that makes you feel better :) 17:08
fbrito hehehe, it does. sorry, but I am very literal :P
Coke dukeleto: why would static functions not show coverage if they are used by a non-static func?
dukeleto Coke: static functions are inlined at compile time and don't show up in code coverage 17:13
fbrito: which task do you want me to change the title of? orderedhash ? 17:14
17:14 gbacon left
Coke dukeleto: huh. I'd expect gcov &/or our compiler options to be smart enough to handle that. 17:14
fbrito dukeleto: yes. and maybe update the test coverage % on the description
dukeleto Coke: this is very basic. static functions are never "called" because the C preprocessor/compiler (not sure which) inlines them 17:17
Coke: it happens before anything else can see what is going on
PerlPilot dukeleto: I think Coke wants an option to turn off that inlining. 17:18
Kristaba Can someone confirm me if the tailcall passing arguments system is supposed to be the same than invokecc? 17:19
17:19 PerlPilot is now known as PerlJam
Coke dukeleto: are you sure that's done when the function is just "static" and not "static inlnie" ? 17:20
*inline
(looks like g++ would do it regardless).
dukeleto Coke: i am actually not sure. I just pretend to know stuff 17:22
Coke looks like it's enabled with -OX, but could be overridden with something like -fno-inline 17:23
(just looking at SELECT cl.val
FROM t_ecm_filing f, t_ecm_obligation o, t_ecm_countrylogic cl
WHERE f.obligation_id = o.obligation_id
AND cl.country_code = '??'
AND f.filing_id = <cfqueryparam value="#Arguments.filing_id#" cfsqltype="cf_sql_numeric">
AND cl.logic_id =
GAH
gcc.gnu.org/onlinedocs/gcc/Optimize...ze-Options is the url i was looking at.
dukeleto Coke-- # pasting cold fusion into #parrot ;) 17:30
darbelo starts looking for exploits... 17:31
17:32 cogno joined, plobsing joined
dukeleto Coke: so do you think we should compile with -fno-inline for repos that do coverage tests? 17:33
Coke: perhaps we need a Configure.pl argument?
Coke --cflags="-fno-inline" 17:35
we're already doing a reconfigure, so adding another parameter should be easy. 17:36
dalek rrot: b9e9c3c | fperrad++ | compilers/pct/src/PCT/Grammar.pir:
[PCT] fix ww when called by generated code

sample from language GIL:
   $P0 = find_name 'ww'
   ...
   captob = $P0(captob, 'dba'=>"")		 17:38
17:47 plobsing left 17:52 fbrito left
fperrad plumage is working again 17:55
17:55 cogno left 18:02 plobsing joined 18:03 fbrito joined
dalek umage: 1af2147 | fperrad++ | t/ (4 files):
fix shebang of t/*.t
 18:04
umage: 5f8e803 | fperrad++ | t/sanity.t:
tests use the fakexecutable installable_plumage
umage: 80fc017 | fperrad++ | src/plumage.nqp:
fix the following error:

current instr.: 'command_info' pc 3997 (src/plumage.pir:1476)
parrot seems more strict now
whiteknight ...and parrot/plumage is now feeding into dalek
dalek r: 3a28ad4 | bacek++ | src/POST/Compiler.pm:
Fix generating pbc for POST::Op
 18:05
r: b028854 | bacek++ | src/POST/Compiler.pm:
Select main sub in same way as imcc
r: 3347e7d | bacek++ | src/POST/Compiler.pm:
Partially migrate sub invocations to new Bytecode
whiteknight ...and so is parrot/pir
cotto nom 18:08
18:08 fbrito left
fperrad whiteknight, have you seen smolder.parrot.org/app/projects/smoke_reports/3 ? 18:10
18:14 gbacon joined
dukeleto yay for passing test suites! 18:14
18:15 mtk left
dukeleto whiteknight: i still get /home/leto/git/plumage/plumage.c:4277: undefined reference to `Parrot_api_string_export_ascii' 18:16
whiteknight: i just did a realclean and installed a fresh parrot from the master branch
whiteknight: am i doing something wrong?
whiteknight: this is in plumage
18:16 cogno joined 18:22 JimmyZ joined
dukeleto whiteknight: i am rm -rf'ing and starting totally fresh, trying to reproduce 18:23
Could not upload report to Smolder at smolder.parrot.org
HTTP CODE: 500 (Internal Server Error)
smolder--
whiteknight: false alarm. Evidently 'realclean' wasn't really cleaning. Sometimes "git clean -fdx" is the only solution 18:24
whiteknight dukeleto: weird. File a bug report 18:26
dukeleto whiteknight: if i hit it again, i will 18:27
18:28 particle1 joined
cotto_work ~~ 18:29
18:31 particle left
dukeleto cotto_work: buenas dias 18:33
moderator Parrot 2.11.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Keep up with GCI 18:34
dukeleto what does "keep up with gci" mean? Does it mean "try not to drown in GCI student commits?" 18:34
we are pretty short on tasks right now.
we also have many branches on github from GCI. We should short those out sooner rather than later, or the bitrot elves will eat them. 18:35
s/short/sort/
is GCI our only goal for this week? 18:36
18:36 plobsing left 18:37 plobsing joined
cotto_work dukeleto: Rakudo was too, but plobsing++ took care of it last night. 18:38
18:38 mtk joined
cotto_work dukeleto: is plumage usable now to run tests on HLLs? 18:38
darbelo It should be. 18:39
If it isn't I might be able to spare a few tuits.
dukeleto cotto_work: plumage works
cotto_work Shiny. Let's start using it then. 18:42
cotto_work wonders if we can build with tcc 18:43
whiteknight we want plumage able to submit smoke reports
cotto_work "no" appears to be the answer 18:44
dukeleto whiteknight: what do you mean? 18:45
whiteknight nevermind, I take that back. Plumage is currently submitting smoke reports
fperrad++
dukeleto whiteknight: yep :)
whiteknight smolder.parrot.org/app/projects/smoke_reports/3
cotto_work v nice
seen kid51 18:46
aloha kid51 was last seen in #parrot 16 hours 26 mins ago saying "But, Yuki`N, you still haven't told us *where* you want to use it.".
18:47 bluescreen left
cotto_work seen matt 18:50
aloha Sorry, I haven't seen matt.
cotto_work seen matt_
aloha matt_ was last seen in #parrot 3 days 13 hours ago saying "Im gonna go now. Good night!".
cotto_work seen matt__
aloha Sorry, I haven't seen matt__.
dukeleto cotto_work: we need more goals, or the GCI goal needs to be more specific 18:52
Coke plumage does NOT work. 18:55
cotto_work dukeleto: is there a way to delete branches through github's web interface?
Coke (if by work you mean "pass all tests")
dukeleto Coke: show us the errors. It works for whiteknight and me 18:56
darbelo Coke: It does. Update.
dukeleto cotto_work: probably, why do you ask?
18:57 bluescreen joined
dukeleto cotto_work: git push origin :branch_to_delete 18:57
cotto_work: is how to delete from the command line, which is described in git_workflow.pod :)
darbelo Coke: see 80fc017 by fperrad++
cotto_work dukeleto: yup. I was just looking for a lazier way. 18:58
Coke dukeleto: Failed tests: 3-18 19:00
dukeleto Coke: what do you want to delete?
Coke: a smoke report would be useful
Coke (t/sanity.t) - that's after doing a clean/build/test at 80fc017
cotto_work Coke: same thing here 19:01
reported to smolder
Coke dukeleto: This step needs a parrot built with zlib
dukeleto also, as a data point, I had to rm -rf the directory that I install parrot into (in my source tree) to get Plumage to compile 19:02
Coke but you can use cotto's. There's nothing to see in the error output, the tests fail, done.
plumage compiles just fine for me.
dukeleto some bug with "make realclean", probably
those tests could have more useful failure output 19:03
Coke cotto_work: 14:00 <@dukeleto> Coke: what do you want to delete? 19:05
19:07 JimmyZ left
cotto_work a couple of merged gci branches 19:07
19:08 cogno left
dalek TT #1764 closed by whiteknight++: Infinite loop in exit handler 19:15
TT #1764: trac.parrot.org/parrot/ticket/1764
19:17 bluescreen left
Kristaba Does anyone here know pretty good the tailcall operation? 19:25
whiteknight Kristaba: sure, what's up?
Kristaba whiteknight: Oh, I have a complex problem with this 19:26
whiteknight: When a sub is called, and when this sub do a tailcall, the tailcalled sub seems to be called 2 times (one time at the tailcall instruction, and one time if the main sub do "get_results") :/ 19:27
Coke Kristaba: do you have a small PIR snippet to demonstrate the issue?
19:27 bluescreen joined
Kristaba pastebin.com/sU0U1J24 19:27
whiteknight Kristaba: nopaste your code?
Kristaba The output is "main\\nfoo\\n [the too few argument error message]", the current instruction is the line 16, and the called instruction is the line 5 19:28
whiteknight Kristaba: Okay. Don't use invokecc directly. Call f like "f()"
Kristaba (Using the zero-arity branch)
whiteknight same with tailcall op. use .tailcall b("0", "from_foo") 19:29
Kristaba whiteknight: Isn't my code, it's a test code at t/op/calling.t
whiteknight and use .return() instead of returncc
Kristaba: oh. that test is wrong then
Coke whiteknight: assuming the purpose is to text get_results & set_args & friends, that doesn't help.
whiteknight: not necessarily.
Kristaba There are many test that use this type of old code :/ 19:30
whiteknight Coke: set_args and get_results and friends are included in the PIR call sugar
their use is incorrect in PIR
19:30 rfw joined
Coke whiteknight: no, it isn't. it might be uncommon and not recommended, but "incorrect" is a stretch. 19:30
19:30 plobsing left
Kristaba The line 15 is a line I added to fix the problem that append in this branch when get_parms is used instead of .param 19:31
Coke if it really is meant to be incorrect to use them, then that code shouldn't even compile.
19:31 dmalcolm joined
dalek TT #833 closed by whiteknight++: Exception from :init sub in PIR compreg causes problems 19:32
TT #833: trac.parrot.org/parrot/ticket/833
TT #1764 reopened by coke++: Infinite loop in exit handler
TT #1764: trac.parrot.org/parrot/ticket/1764
19:32 plobsing joined
Kristaba So, anyone has an idea of what I would have to do? 19:33
Because many test I have to fix have this kind of problem :(
19:34 Yuki`N joined
Yuki`N dukeleto, how do you feel about Configure.pl forking to handle "tee" emulation? 19:34
dukeleto Yuki`N: sounds fine to me 19:35
Kristaba whiteknight: See the description of my task : www.google-melange.com/gci/task/sho...9413283999 19:36
whiteknight: I think the fix I have to do are really dirty, and all of them affect tests which use invokecc, set_args, get_params, etc... 19:38
So I don't know what to do, I think the major part of these tests should be rewriten instead to fix them that way... 19:40
19:47 aantn joined 19:48 cogno joined
dalek TT #1764 closed by coke++: Infinite loop in exit handler 19:49
TT #1764: trac.parrot.org/parrot/ticket/1764
Yuki`N Hmmm, I'm having trouble forking twice to handle stderr and stdout. :| 19:59
cotto_work dukeleto: is there a compelling reason to use Parrot_hsh_* instead of Parrot_hash_* for hash function names? 20:02
dalek rrot: 6231646 | fperrad++ | runtime/parrot/library/distutils.pir:
[disutils] add get_submitter()
 20:03
TT #833 reopened by whiteknight++: Exception from :init sub in PIR compreg causes problems 20:05
TT #833: trac.parrot.org/parrot/ticket/833
20:06 bluescreen left 20:10 Yuki`N left 20:11 Yuki`N joined 20:12 Nol888 joined 20:14 Yuki_N joined 20:15 cogno left 20:17 Yuki`N left, cogno joined 20:19 Nol888 left 20:21 Yuki_N is now known as Yuki`N
dalek rrot: 8b188db | Whiteknight++ | / (3 files):
Fix for TT #1029. The return continuation was being mangled in the PIR compreg function, then we weren't able to account for it in the NCI.invoke VTABLE. cache the old value before we call the function, then sanity check the new return continuation value to see if we should update the cache. All tests pass.
 20:31
rrot: 07c6d61 | Whiteknight++ | / (3 files):
Merge branch 'master' of github.com:parrot/parrot
Yuki`N whiteknight, I got the tee-style Configure.pl working but as a side-effect I had to redirect stderr to stdout. 20:36
20:36 aantn left
dalek TT #1435 closed by gerd++: Tru64 report 20:39
TT #1435: trac.parrot.org/parrot/ticket/1435
TT #1029 closed by whiteknight++: taillcall into PIR compiler broken
TT #1029: trac.parrot.org/parrot/ticket/1029
20:39 aantn joined
aantn rfw: why are you missing from socghop.appspot.com/gci/program/sho...le/gci2010 ? 20:43
rfw aantn: it takes a while to load
mainly because it loads in chunks of 150 records
aantn rfw: okay 20:44
Coke whiteknight: oooh, #1029. Danke.
rfw aantn: you scared me for a second
20:48 cogno left
Yuki`N [configure_mod 8889182] 20:48
All numbers!
8889182311ffabfeaec80a80a316600770eaee3e
dukeleto cotto_work: not really. I think 3 or 4 letter short names should be fine 20:49
Yuki`N That is the lowest entropy SHA1 I have ever seen.
aantn cotto_work: I'm having some trouble with socghop.appspot.com/gci/task/show/g...9360578422
cotto_work: the first two untested lines don't seem to be testable 20:50
dukeleto aantn: what's up?
aantn dukeleto: ^
whiteknight aloha: coverage? 20:51
aloha whiteknight: coverage is cv.perl6.cz or tapir2.ro.vutbr.cz/cover/cover-results/
dalek rrot: 1e56d3f | fperrad++ | src/dynpmc/os.pmc:
[dynpmc] fix build on MinGW32
20:52 bacek joined
whiteknight aantn: those two lines look like dead code. 20:52
aantn: rip those lines out and replace them with PARROT_ASSERT()
aantn whiteknight: what does PARROT_ASSERT() do?
whiteknight aantn: very similar to assert from the standard library. 20:53
aantn: do PARROT_ASSERT(!PMC_IS_NULL(namespace)) for instance
or PARROT_ASSERT(!PMC_IS_NULL(info))
20:53 plobsing left
Yuki`N Can someone review github.com/nol888/parrot/commit/98...a5bfd92da? 20:54
aantn whiteknight: when is a class's nameset set? before of after init_pmc? 20:57
whiteknight during init_pmc 20:58
aantn whiteknight: by init_pmc? 20:59
whiteknight line 355
init_pmc calls init_class_from_hash
init_class_from_hash calls the method on NameSpace PMC to set the class
Yuki`N dukeleto, can you review github.com/nol888/parrot/commit/98...2a5bfd92da as a possible solution? 21:00
whiteknight aantn: actually, like 360
sorry
aantn whiteknight: doesn't line 355 unset the namespace if it was already set?
whiteknight aantn: yes, I was mistaken. 355 unsets the class in the old namespace. line 360 sets it in the new namespace 21:01
aantn whiteknight: but if the namespace can't be set before then (before init_pmc calls init_from_hash) then the code to unset the existing namespace can also be removed
whiteknight aantn: no. You can see from the coverage report that the code IS being called. Which means there is an old_ns 21:02
oh wait, wrong again
I keep looking at the wrong lines
aantn: init_class_from_hash is called from other places 21:03
clone_pmc, METHOD name
the one in clone_pmc cannot have an old_ns either
but the one in METHOD name might 21:04
because that method renames the class, which associates it with a new namespace
aantn whiteknight: okay, i see
whiteknight: i'm writing a test for thatn now 21:05
whiteknight okay
21:05 fperrad left
whiteknight aantn++ 21:07
21:13 bacek_ joined
dukeleto Yuki`N: why did you modify lib/Parrot/Configure/Utils.pm ? 21:15
Yuki`N: why must the parent redirect STDERR to STDOUT ? 21:16
Yuki`N dukeleto, STDERR doesn't get caught by the fork, so it doesn't get logged otherwise.
Utils.pm had a prompt() sub, and I modified it to print to stdout the response to prompts so that they're logged as well. 21:17
21:17 cogno joined
aantn whiteknight: look at the section which starts with: if (VTABLE_exists_keyed_str(interp, info, resolve_method_str)) { 21:17
whiteknight: I can't give you a line number, because I've modified the file 21:18
bacek_ aloha, humans
21:19 bacek left
aantn whiteknight: I can test that superficially by passing a resolve_methods list to init 21:19
(via the init hash)
but is there a way to really test if that line is doing what it should be?
i.e. is there any situation where a non-empty list of resolve_methods would be passed to init_from_hash 21:20
dukeleto Yuki`N: ok, sounds reasonable, i guess. did you send a pull request ? 21:22
aantn I can't find any good documentation on resolve_methods, but as far as I can understand, it's used to check whether a class implements a role or not 21:23
Yuki`N dukeleto, no, not yet, I wanted to get feedback and whatnot.
whiteknight aantn: I don't know. I have to look 21:26
aantn whiteknight: okay, thanks 21:27
whiteknight aantn: Yeah, it looks like something tricky that has to do with roles 21:28
I would do the superficial test for now. Add a TODO note-comment to your test to mention the shortcomings
aantn whiteknight: okay, thanks
whiteknight I'm heading home now. Later 21:29
sorear ping dukeleto
aantn whiteknight: k, thanks
21:30 cogno left 21:31 whiteknight left, plobsing joined 21:32 darbelo left
dukeleto sorear: pong 21:34
sorear where is plumage? 21:40
Yuki`N dukeleto, pull requete'd 21:41
*request'd
21:42 M_o_C left
tadzik sorear: github/parrot/plumage iirc 21:42
21:42 rurban_ joined
tadzik plumage? 21:42
aloha: plumage?
aloha tadzik: plumage is a package manager for Parrot or github.com/parrot/plumage
dukeleto sorear: github.com/parrot/plumage 21:43
Coke plumage is now working for me after a reinstall of parrot. 21:44
if plumage requires a specific version of parrot, it should probably die during the build rather than fail at runtime. 21:45
21:45 rurban left, rurban_ is now known as rurban 21:48 cogno joined 21:50 bacek__ joined
sorear tadzik: is dalek working for you now? 21:53
tadzik sorear: no idea, haven't commited anything in a while
hmm, is parrot/bf tracked? 21:54
sorear dukeleto: github.com/perl6/mu/blob/master/mi...k-push.txt # Best way to track a github repository
sadly, I can't do it - it requires admin access to the repo 21:55
21:55 bacek joined 21:56 bacek_ left 21:57 aantn left 22:00 bacek__ left
dukeleto sorear: i can do it 22:03
sorear: i thought there was a different way, some metadata file that you maintained
22:04 cogno left
dukeleto sorear: dalek-conf.json it seems to be called 22:04
22:06 bacek_ joined
dukeleto sorear: now i understand 22:07
Yuki`N: submit your task? 22:09
22:11 Yuki`N left 22:12 bacek left 22:15 user1 joined 22:24 user1 is now known as la_step, la_step left
cotto_work ~~ 22:26
22:26 bacek__ joined, fbrito joined
dalek rrot/gci_hash_func_rename: fbce216 | cotto++ | / (35 files):
have we learned nothing from creat?
 22:29
22:32 bacek_ left, bacek_ joined 22:37 bacek__ left
mikehh t/pmc/integer.t - TODO passed: 127 - ok 127 # TODO unresolved bug, see TT #1887 - TT #1887 was closed a few hours ago, nobody unTODOed the test (not sure how to) 22:41
cotto_work mikehh: thanks for noticing! 22:42
mikehh line would line 627 - todo($I0, 'i_multiply Integer PMC by BigInt PMC', 'unresolved bug, see TT #1887') -> ok($I0, 'i_multiply Integer PMC by BigInt PMC') work? 22:43
not sure if that get's the intent of the test 22:44
forget the initial line 22:45
seems to work, should I commit, or is someone else working on it? 22:50
dalek rrot: 826469e | mikehh++ | t/pmc/integer.t:
remove resolved TODO from test
 22:56
rrot: 58adb70 | mikehh++ | t/pmc/integer.t:
forgot to update copyright year again
 23:04
23:20 bluescreen joined 23:21 bluescreen left, bluescreen joined 23:28 bacek__ joined
cotto_work hi bacek_ 23:33
23:34 bacek_ left, bacek__ is now known as bacek
bacek aloha, cotto_work 23:34
dalek rrot/opmap_aware_pmcs: d66af7f | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Add PFBCS's direct manipulation functions.
rrot/opmap_aware_pmcs: 348d61c | bacek++ | src/pmc/packfilebytecodesegment.pmc:
More PFBCS accessors
rrot/opmap_aware_pmcs: 73c23da | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Add .opmap for fetching OpMap.
rrot/opmap_aware_pmcs: 1dae6e2 | bacek++ | src/pmc/packfilebytecodesegment.pmc:
Reorder vtables.
rrot/opmap_aware_pmcs: 1b566eb | bacek++ | src/pmc/packfileopmap.pmc:
More accessors to OpMap to fetch Op PMC by name.
r: 57d1f3d | bacek++ | src/ (2 files):
First step of fixing fixup_labels. Calculate offset of label within op
 23:37
Kristaba cotto: I just finish my task : www.google-melange.com/gci/task/sho...9413283999 23:40
cotto: Tell me if you think that something needs to be redone 23:41
dukeleto cotto_work: that task is all you. The commit looks good, feel free to complete the task if the tests pass :) 23:42
dalek r: 57d1f3d | bacek++ | src/ (2 files):
First step of fixing fixup_labels. Calculate offset of label within op
cotto_work Kristaba: great. I'll look at it later this evening when I can throw some dedicated time at it. 23:43
dukeleto: you mean Kristaba's?
Kristaba cotto: Ok, don't worry, in France it's about 1 am, so I'm not really hurried ;) 23:45
dukeleto cotto_work: yep
23:49 whiteknight joined 23:51 fbrito left
dalek m-eta-wink-kzd: 943b658 | plobsing++ | src/Ωη.coda.winxed:
distinguish between init_pmc vtable overrides for NullOptimization
 23:53
m-eta-wink-kzd: d837de9 | plobsing++ | src/Ωη.coda.winxed:
workaround set_integer_keyed in NullOptimization
m-eta-wink-kzd: 5dbd234 | plobsing++ | src/ (2 files):
liberal sprinkling of string() coercions to get valid winxed output
r: b82088d | bacek++ | src/POST/Compiler.pm:
Fix stupid typo in fixup_labels.
 23:54
mikehh whiteknight: ping
whiteknight pong 23:55
mikehh whiteknight: having a problem with t/pmc/integer.t, I sorta fixed a problem, then looked at your commit from a few hours ago 23:57
whiteknight mikehh: what problem? What platform? 23:58
mikehh you sorta fixed the wrong todo, but what you did works withh g__, but fails with g__ --optimize (complex multiply)
dalek r: b82088d | bacek++ | src/POST/Compiler.pm:
Fix stupid typo in fixup_labels.
mikehh damn, g++ 23:59
whiteknight mikehh: weird. Let me fire up the VM and give that a spin

Powered by Cro and the Raku® Programming Language.

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

Thank you!