Parrot 1.8.0 Zygodactyly released | Priority: define 2010 Roadmap | Priority: trac.parrot.org/parrot/wiki/ClassV...eOverrides | Latest modified TT's: icanhaz.com/parrotbugs | Parrot Languages: icanhaz.com/parrotlang
Set by moderator on 8 December 2009.
00:25 tetragon joined
dalek rrot-plumage: 40c840b | japhb++ | src/ (3 files):
[LIB,plumage] Improve metadata validation, resilience, and error reporting
 00:29
00:38 colomon___ joined 00:51 kthakore joined
kthakore hiya 00:51
00:51 kid51 joined
kthakore is there a page for rakudo that is update regularly that says what is new ... maybe even give snippets? 00:51
kid51 How about: www.rakudo.org/ 00:53
kthakore kid51: um ... 00:55
kid51: first thing I clicked on
rakudo.org/node/59
then I see Rakudo now supports unpacking of arrays, hashes and objects in signatures
that is great ... but wth does that mean for a programmer
kid51 www.rakudo.org/status
kthakore CODE!!!
kid51: .... I think you missed the point 00:56
kid51: I am saying spec tests don't mean anything to me as I don't even know what I can do with it
kid51 kthakore: Perhaps you want #perl6 on freenode. That's more specific to Rakudo
kthakore kid51: ok
kid51: I am just saying we should have a page like that for people looking at rakudo 00:57
like here go try this
use perl v6; say "This is teh awesome code"
then I can go and see that it is kewl
kid51: I only bring this up because I was asked by a friend what does the esc tests mean.... in CODE 00:58
japhb kthakore, the problem is that the available stuff is so massive that we are literally writing books about it. There's no decent way to summarize.
00:59 abqar joined
kthakore japhb: ah ok 00:59
japhb: where are these books?
japhb perl6book?
ah, not in the bot yet
kthakore oh that one ...
ok
when you guys write the tests ... do you not write tests in perl6? 01:00
japhb ah, you know it, good
yes
There's a massive archive of them
s/archive/repository/
perl6.org/specification/
kthakore japhb: so can we not just have few of those as good examples and have like a rss feed of tem
japhb See "official test suite" 01:01
kthakore so new people can see oooh I can do this now
japhb That would be a pretty high-traffic RSS feed ...
kthakore japhb: yup !! and that is good!
japhb For a while kyleha++ was posting every time he wrote tests for a new ticket
kthakore, well, suggest it on #perl6, you may get a taker. :-) 01:02
kthakore ok ... new project for me I guess
japhb: ok
I was just flabergasted when I was asked out of the blue
(even though I have not a single inkling of p6 beyond say)
japhb nodnod 01:03
kthakore logging and posting this somewhere
particle kthakore: you want a feed of threads like groups.google.com/group/perl.perl6....b80ab11b99
or you could subscribe to p6c and filter messages 01:04
kthakore We're sorry, but we were unable to find the topic you were looking for. Perhaps the URL you clicked on is out of date or broken 01:05
i gtg
dalek TT #1349 closed by jkeenan++: [patch]changed arrayiterator.pmc to use GET_ATTR syntax 01:32
TT #1280 closed by jkeenan++: [patch]removed unused codes in oo.c
TT #1216 closed by jkeenan++: [PATCH] Fix dependency in docs/Makefile
TT #1357 closed by jkeenan++: [patch]changed capture.pmc to use GET_ATTR syntax
kid51 Does anyone know the final plans for the online Roadmap meeting this weekend? 01:34
If so, we should post on list and change the channel topic.
dalek rrot-plumage: d3e30c7 | japhb++ | src/lib/Plumage/Project.nqp:
[lib] Plumage::Project: fix stupid bug in fetch_git()
 01:36
rrot-plumage: 652b7d0 | japhb++ | src/ (3 files):
[LIB,plumage] Save away a copy of the project metadata at install, remove it at uninstall
01:50 JimmyZ joined 01:51 JimmyZ_ joined, JimmyZ_ left 01:59 JimmyZ_ joined, JimmyZ_ left, JimmyZ left 02:00 JimmyZ_ joined 02:01 JimmyZ joined
JimmyZ cla? 02:19
purl hmmm... cla is Contributor License Agreement or www.perlfoundation.org/contributor_..._agreement or www.parrot.org/foundation/legal or www.parrot.org/files/parrot_cla.pdf or www.lowcarbfriends.com/bbs/main-low...-acne.html
kid51_at_dinner Hmm, that response needs some trimming. 02:31
kid51 trimmed of spam 02:34
JimmyZ good evining, kid51 ;) 02:37
02:46 szabgab joined
kid51 JimmyZ, Belated hello to you 03:11
JimmyZ: I think we've applied all your most recent patches. Of the recent TT, only 1367 is still open, and I'll close that if there are no problems in 3 days. 03:13
JimmyZ kid51: #886 #896 03:14
kid51: #896 was not a good patch, since I'm killing PARROT_SOCKET(); 03:16
kid51 Then you should post something about that in the TT.
#896 is Win32-specific. Since I don't have access to Win32, I'm not in a position to do anything about that. 03:17
Suggestion: Name your patches something more self-documenting than 'cage.patch'; 'cage.1.patch'; etc. 03:18
Otherwise, when I download them for testing, I have to rename them so that they don't wipe out your *other* patches ;-)
JimmyZ kid51: yes, but before I had killed it. #896 is still worthy to commit. 03:19
kid51: that's old patch, so they were using old patch name. 03:20
kid51 What I am going to do is copy and edit the patch to eliminate the two Win32-specific files. I will then run the remaining patch through make test on Linux. 03:25
nopaste "kid51" at 70.85.31.226 pasted "tt 886: patch application problems" (22 lines) at nopaste.snit.ch/19033 03:27
kid51 Had problem with two files.
config/gen/platform/generic/exec.c
src/pmc/codestring.pmc
It's quite possible that those files have changed enough since you created the patches that the 'patch' program can't figure out what to do. 03:28
and, given that we just applied your other patch on codestring.pmc, it's more than just possible.
So, can you look at those two files?
nopaste "kid51" at 70.85.31.226 pasted "JimmyZ: tt886: give 1 patch with these files and another with 3 win32 files" (7 lines) at nopaste.snit.ch/19034 03:33
kid51 I will try to get back to this tomorrow.
JimmyZ kid51: will try :) 03:34
kid51 Thanks.
japhb Tene, ping 03:39
dalek rrot-plumage: c7f94d2 | japhb++ | docs/hacking/add-action.pod:
[DOCS] Update add-action hacking doc
 03:48
rrot-plumage: fa45edd | japhb++ | TODO:
[META] Finally, able to remove a few TODO items
04:06 pjcj joined 04:14 brooksbp joined
Coke tene: any luck with break? 04:28
vi 04:34
ls
... grrr.
04:59 chromatic joined
Coke pmichaud: you about? 05:09
anyone able to help out with an AST question? 05:31
05:41 szabgab joined
PerlJam Coke: I'm awake; I can try to help; no promises :) 05:52
Coke bah. i nearly have it. I'm not asking for help now.
;)
05:54 plobsing joined
Coke Q'pla! 06:02
06:04 theory joined 06:23 DrForr_ joined 06:33 cotto joined 06:34 colomon joined
Coke incoming. 06:53
purl duck!
Coke PerlJam: I was trying to figure out how to get tclstring's "get_integer" to work with nqp-rx's rules. Wasn't too bad once I stared at the auto-generated code from [expr]
so now [time {expr 2+2} 3.2] will barf with {expected integer but got "3.2"} 06:54
dalek rtcl-nqp: c5072a4 | coke++ | t/cmd_ (3 files):
Remove references to obsolete RTs.
 06:56
rtcl-nqp: 5fd27b2 | coke++ | src/class/tclstring.pir:
fix boolean check
rtcl-nqp: 480b2d6 | coke++ | lib/test_more.tcl:
Avoid requiring parens in expressions in [like]
rtcl-nqp: f1e04a0 | coke++ | src/Partcl/commands/main.pm:
Numeric context on time count is insufficient, must be integer.
rtcl-nqp: 7d0856d | coke++ | src/class/tclstring.pir:
Add a 'get_integer' override that uses the Grammar's <integer>
rtcl-nqp: 68c6e9c | coke++ | build/Makefile.in:
This test now passes.
rtcl-nqp: 041316c | coke++ | TODO:
Swap a new one in at the top of the pile
rrot: r42980 | cotto++ | trunk/include/parrot/runcore_profiling.h:
[profiling] make the profiling runcore struct a little more efficient
 07:20
07:21 spinclad joined 07:45 mikehh joined 08:06 fperrad joined
JimmyZ cotto: I submit a cla. 08:12
08:14 colomon__ joined
cotto JimmyZ, great! Where are you? 08:14
JimmyZ cotto: China
cotto Ok. If you've got patches, feel free to keep submitting them. 08:15
anyone: what'd be the best place to put documentation on the profiling runcore?
JimmyZ cotto: I will. 08:16
cotto: I think parrot/docs is the best place? 08:17
cotto There doesn't seem to be too much of a standard. I blame the read-only pod2html infrastructure. 08:18
chromatic docs/dev
purl rumour has it docs/dev is the starting point for dev docs now
08:19 colomon joined
cotto sounds good 08:20
chromatic That's where I've always intended to document the runcore API. 08:22
cotto I figure that adding a third environment variable means I probably need to start documenting something. 08:23
(It'd be really nice if I could pass cli options to the runcore, btw.)
dalek rrot: r42981 | cotto++ | trunk/src/runcore/profiling.c:
[profiling] cleanup and commenting
 08:26
08:27 integral joined 08:29 iblechbot joined
fperrad seen japhb 08:30
purl japhb was last seen on #parrot 4 hours, 50 minutes and 29 seconds ago, saying: Tene, ping
chromatic CLI handling would be very nice. 08:31
cotto Do you know what a good way to handle it would be?
I assume it'd need to use char*s since Parrot's string initialization code won't have run by that point. 08:32
fperrad how to purge the queue of message from purl ? 08:33
cotto say messages erase
mikehh use messages erase in purl 08:34
fperrad thanks
chromatic We could initialize the string system first. There's no reason that shouldn't work.
cotto Sure. If we're never going to allow CLI args to effect the string system, that'll make life less painful. 08:36
chromatic If we had an argument to tell Parrot which encoding to use to process arguments, for example? 08:39
I shouldn't get into these philosophical bootstrapping questions while, in another window, I have a character in the middle of a philosophical psychological breakdown over the concepts of mortality and the notion of existence. 08:40
This is why Zelazny wrote so few computer programs. 08:41
cotto --encoding=not_this_one 08:42
I don't see the problem.
;)
dalek rrot: r42982 | cotto++ | trunk (4 files):
[profiling] add stub documentation and configuration support for annotations
 08:43
08:43 colomon joined
chromatic True, there's always LANG 08:43
cotto It's easy enough to detect that whenever is most convenient. 08:45
chromatic Lots of things break if you don't have LANG right. 08:46
I'm perfectly willing to point and laugh at bug reports that don't. 08:47
"I see that you're running Red Hat 7. What did you think would happen?" 08:48
09:05 payload joined
cotto chromatic, how long do you think it'd take to move string initialization? 09:07
09:08 colomon_ joined
JimmyZ TT #888 needs more love. Where is loooooove? 09:13
09:14 colomon__ joined
chromatic It shouldn't take too long. 09:17
Hm, looks like it already happens first. 09:18
src/main.c +59
cotto parseflags needs to get out of imcc 09:21
dalek tracwiki: v5 | cotto++ | CottoTasklist 09:26
tracwiki: trac.parrot.org/parrot/wiki/CottoTa...ction=diff
TT #1321 closed by fperrad++: Convert abc to nqp-rx 09:31
09:39 colomon__ joined 09:40 uniejo joined
dalek rrot: r42983 | fperrad++ | trunk (4 files):
[abc] fix test infrastructure
 09:48
tracwiki: v118 | fperrad++ | Languages 09:59
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
10:02 colomon joined
dalek rrot: r42984 | cotto++ | trunk/docs/dev/profiling.pod:
[docs] flesh out the profiling docs a bit
 10:05
10:11 payload joined 10:42 riffraff joined
dalek tracwiki: v119 | fperrad++ | Languages 10:48
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
cotto definitely time for sleep 10:49
night
dalek rrot: r42985 | cotto++ | trunk (2 files):
[profiling] Add initial support for (optionally) recording annotations in the pprof file

Annotations currently slow the runcore down by about 50%, but the implimentation is intentionally simplistic for now.		 10:54
rrot: r42986 | fperrad++ | trunk/tools/dev/fetch_languages.pl:
[languages] add close & winxed
11:01 cognominal joined 11:20 allison_ joined
allison_ where do I find the nqp-rx version of partcl? 11:26
Infinoid I believe that would be github.com/partcl/partcl-nqp 11:34
allison_ excellent! 11:35
purl plays air guitar
riffraff apparently, i can't build parrot lately, I keep getting a message: 'Parrot_Continuation_attributes’ has no member named ‘to_call_object’ 11:37
any idea on what is happening? tried realclean+configure+make to no avail
allison_ Infinoid: that's it, thanks
riffraff: this is on a checkout of svn trunk? 11:38
riffraff: and what platform/compiler, etc?
11:43 AndyA joined 11:45 colomon joined 11:55 preflex joined 11:59 colomon__ joined 12:04 colomon_ joined 12:18 colomon joined 12:25 payload joined
riffraff allison, sorry was trying to cook 12:37
yes latest revision
osx snow leopard
gcc 4.2.1
12:38 mikehh joined
dukeleto 'ello 12:44
12:46 whiteknight joined
whiteknight pmichaud: ping 13:08
nevermind, I'll send an email 13:13
dukeleto whiteknight: good moroning
whiteknight good morning dukeleto 13:14
dukeleto whiteknight: what hacketh you on this fine morning?
whiteknight nothing in particular, got some more Matrixy work to do, but running out of steam on it
dukeleto i want to make Tapir installable via Plumage, as well as able to be packaged as a single .pbc
whiteknight I don't really know anything about Plumage, but I would be happy to learn 13:15
dukeleto whiteknight: once i have Tapir useable by others, i will help switch various projects over, like matrixy
whiteknight: if i have a .pir file which .include's other .pir files, can i just generate a single .pbc from the original .pir? will the resulting pbc need those .included PIR files to be present, or does it "package" all the .includes into the PBC ? 13:16
i guess i can generate PBC's of the .included PIR files, then .include those. but I want a single, stand-alone file, at the end 13:17
whiteknight Matrixy and PLA are using the test facility in setup.pir 13:18
fperrad++ gave me setup.pir files the other day
dukeleto whiteknight: cool. what is your test harness? 13:19
whiteknight I have no idea 13:20
whatever setup.pir is using
I don't know a lot about distutils yes
yet*
dukeleto whiteknight: how does your test suite get invoked?
whiteknight "parrot setup.pir test" 13:21
dukeleto whiteknight: i just updated my matrixy repo and i don't see setup.pir in your project root 13:22
whiteknight: i see Configure.nqp
whiteknight oh, maybe I haven't updated matrixy yet 13:24
I have the file locally, I wanted to do more testing. Look at PLA instead 13:25
I would still like to use Tapir though, so I need to see if we can change that setting 13:29
dukeleto whiteknight: i think setup.pir can be made to invoke Tapir
Tapir may have just found a bug in Parrot. Time to put my yak-hole-spelunking gear on 13:30
whiteknight bug? awesome 13:31
dukeleto the entire Tapir test suite fails due to bad exit codes
it was passing a few days ago and i haven't touched it
whiteknight oh great 13:36
so is spawnw returning the wrong exit code?
dukeleto .local pmc exit_status $I0 = pipe.'exit_status'() exit_status = box $I0 13:37
that code is what seems to be failing for me now, returning an exit code of 1 when it shouldn't 13:38
whiteknight hmmm, so has anything changed in FileHandle.pmc, or anything in src/io/*?
dalek pir: 4dfb6d3 | dukeleto++ | TODO:
Update TODO, removing things that are done, adding new important stuff, and re-ordering based on need
dukeleto whiteknight: i am not sure, but i will inspectigate it 13:39
i wonder if parrot has tests for the exit_status method on pipes 13:40
13:45 JimmyZ joined 13:46 tetragon joined, patspam joined 13:49 JimmyZ_ joined
Coke wonders why allison was poking at partcl-nqp and hopes for a patch. =-) 13:53
13:54 payload joined
dalek rtcl-nqp: 870a66d | coke++ | (4 files):
add [flush] - now passing t/cmd_flush.t, even with extra tests [*]
 13:57
mikehh with both Ubuntu 9.04 amd64 and Ubuntu 9.10 amd64 (gcc/g++ 4.3.4 and 4.4.1 respectively) 14:00
t/op/exceptions.t - test31
- fails in testr only with g++ build, with or without --optimize
- fails in smoke and all cores EXCEPT testr with gcc build with --optimize
- passes all with gcc without --optimize
All tests PASS with all variants on Ubuntu 9.10 i386
dukeleto mikehh: wow. 14:01
mikehh All other tests PASS on Ubuntu 9.04 amd64 and Ubuntu 9.10 and64 (all 4 variants)
ha - I have tests in 12 variants in the last 5 hours or so 14:04
14:05 JimmyZ_ left
mikehh dukeleto: I also ran make corevm/make coretest a couple of times (not all variants) and it passes 14:07
14:08 JimmyZ joined
dukeleto mikehh: that is *highly* appreciated. mikehh++ 14:10
mikehh: you may also want to summarize your results, maybe weekly?, to parrot-dev. some people don't do irc
Coke I would rather focus all this testing into smolder. =-) 14:13
and then have folks rely on that.
dukeleto Coke: smolder is cool, but bugs me sometimes
Coke: i tried to run my own smolder instance, but it has horrible memory leaks. needs to be restarted daily. 14:14
Coke it beat the alternative.
dukeleto Coke: that is why it is so slow sometimes. hasn't been restarted lately
Coke dukeleto: I'm familiar with the architecture. 14:15
dukeleto Coke: the alternative of nothing, or the alternative of a better smolder? i think i would port smolder away from cgi::application
Coke dukeleto: I think i would find something we could use and use it rather than chasing more cycles away from core or languages. =-)
dukeleto Coke: yep 14:16
Coke: once i am done with Tapir, most peoples test suites in Parrot-land will be 10% of what they used to be, so we can spend a little time on a better continuous integration solution ;) 14:17
Coke excellent.
Coke wonders if everyone else aside from him prefers to have generated files during a build go into a "gen" directory. 14:18
Coke prefers the more .c -> .o style, where it's right there.
dukeleto Coke: hard to say. depends on the project/language/intent of the generated file :( 14:19
Coke NPQ -> PIR, e.g. 14:20
dukeleto Coke: same directory
Coke: but maybe if there are lots of files, you would want to have an nqp/ and a pir/ . but for less than a few dozen files, same directory 14:21
Coke dukeleto: in partcl-nqp, e.g., there's only a few .pms per dir. 14:22
dukeleto Coke: i would say same directory then. but it is mostly preference. for newcomers to the project, being in the same directory is a little more transparent
mikehh I do a smoke run as part of my tests, but I probably need to summarize my fulltest results 14:23
dukeleto the way some parrot files are generated are crazy and it is impossible to figure out what generated a file, even with ack
PerlJam Coke: "gen" nicely tells us what's generated and makes it easy to ignore generated files in an SCM
Coke PerlJam: the way the dirs are laid out, there is only one .pir file we'd want to not ignore that co-exists with .pm files. 14:24
dukeleto PerlJam: just as easy to ignore a glob in a directory in git, but I agree that it *logically* makes good sense to have generated files in their own home
Coke and your SCM tells you nicely what's generated. =-)
PerlJam: my problem with the gen/ dir is that it doesn't scale as you add files to the build. 14:25
PerlJam Coke: makes it easy to remove generated files for "make clean" and "make realclean" too :)
dukeleto Coke: gen/ should be hierarchical
Coke (you end up with a build rule for each file when a simple .pir.pm rule would do.)
PerlJam What dukeleto said
dukeleto Coke: gen/ should be a copy of the project root, with a hierarchy of generated files
14:25 PacoLinux joined
Coke dukeleto: ... that might work. 14:26
PerlJam now "gen" is starting to sound like the moral equivalent of "blib"
dukeleto then it is easy to figure out how generated files are related to their generators, and it is nicely organized
PerlJam: ruh roh 14:27
Coke if gen were meant for installing, sure.
dukeleto Coke: some generated files are installed
dalek rrot: r42987 | dukeleto++ | trunk/t/library/test_more.t:
[t] Tests for is() : comparing a Float PMC and a float with precision
 14:41
14:50 colomon__ joined
Coke dukeleto: yes. and those are more like .so or .pbc files, as opposed to the intermediate .o or .pir files. 14:57
so, my generic question was really just a mask for partcl-nqp. =-) 14:58
dukeleto someone should port pgegrep to nqp-rx 15:02
what should is(NaN,NaN,whatever) do ? 15:03
Coke fail
nan != nan
dukeleto Coke: i meant to say is(NaN,NaN,diag,precision) 15:04
Coke I would expect just a fail
dukeleto does nan != nan, when you are not talking exact arithmetic?
Coke: yes, i would too
Coke: but our code probably blows up
Coke that's fixable. write a failing test. =-)
15:05 bubaflub joined
dukeleto bubaflub: morgen! 15:05
Coke to check if is a nan, you want something like ok(isNan(nan))
bubaflub dukeleto: mornin'
Coke is pretty sure it ain't morgen in any german speaking locales. =-)
dukeleto bubaflub: you want to write some tests (and fix bugs if they are there) for is(NaN,foo,diag,precision) ? 15:06
Coke: only in my mind :)
Coke (not that isNan() exists.)
bubaflub yeah, in the t/op/inf_nan.t i think the checks are just for 'NaN'
dukeleto bubaflub: where foo can be any type of register, including a NaN
bubaflub i.e. casting to string 15:07
dukeleto bubaflub: also is(foo,NaN,diag,precision)
bubaflub: comparing two NaN's, with any precision, should always fail
bubaflub: because nan != nan, by definition 15:08
bubaflub: so we can always say that (x-y) > precision, regardless of what precision is
bubaflub dukeleto: ok. that makes sense. 15:09
dukeleto bubaflub: these tests should go in t/library/test_more.t
bubaflub: you will probably find some bugs in is() 15:10
bubaflub dukeleto: ok. i'll get a hackin'. should have a patch sometime today.
dukeleto bubaflub: i just added 4 tests. looks at the last 2 commits to trunk for examples
15:10 colomon___ joined
dalek rrot: r42988 | dukeleto++ | trunk/t/library/test_more.t:
[t] Tests for is() : comparing an integer and a Float PMC with precision
 15:14
15:29 JimmyZ joined 15:43 Psyche^ joined
japhb fperrad, ping 15:55
16:06 payload joined 16:23 jsut joined
Coke Hey, anyone want to write some shiny NQP? =-) 16:26
16:27 mikehh joined
fperrad pong japhb 16:29
pmichaud 16:26 <Coke> Hey, anyone want to write some shiny NQP? =-) 16:37
ME!!!!!
(alas, I have other plates to keep spinning at the moment) 16:40
allison Coke: (on Partcl) I was just looking for example code for Pynie 16:50
PerlJam allison: converting pynie to use nqp-rx? 16:51
allison PerlJam: yes 16:52
PerlJam cool
Tene PerlJam: converting pynie is coming up on my list RSN. I need to do squaak first, and then pynie is my next conversion. 16:53
PerlJam Coke: What shiny NQP do you need? (as I continue to procrastinate on important things)
Tene coke, japhb: family drama came up last night, so I didn't get anything done at all. 16:54
PerlJam Tene: Squaak would make a good test case for the NQP book IMHO
Tene I remember hearing about a hackathon this weekend? Who's organizing that? 16:55
purl: msg chromatic Any details on hackathon this weekend?
purl Message for chromatic stored.
allison Tene: there's a plan-a-thon this sunday 16:59
Tene I think it was about merging some of those PCC PMCs. 17:00
allison Tene: I may beat you to it on Pynie, although, I'm taking this as a bit of a chance for a rewrite
Tene allison: you read my abc migration commit, yes? 17:02
allison Tene: not yet, I'll look for it 17:04
17:06 iblechbot joined
Coke Tene: no worries. 17:06
purl no worries. is, like, my smoke harness code public
Coke PerlJam: item 1 in the todo list looks pretty doable if you grok regex and Actions. 17:08
(making optional signs part of a literal int)
japhb is bak 17:09
fperrad, still around?
17:09 Andy joined
Coke 7 is pretty much all regex, putting panics in the right spots. 17:09
japhb Tene: Will you be able to try again tonight possibly?
fperrad japhb, I'm here
Tene japhb: tonight looks much more reasonable. 17:10
Coke ponders issuing beverage coupons in #parrot as a kind of redeemable karma.
japhb Tene, cool bans
Coke redeemable for paypal or an actual drink at a yapc or something. =-)
Tene fperrad: You were trying to talk to me about XML grammar migrated to nqp-rx yesterday, but I never saw an actual question... did I miss something?
japhb Coke, no fair, you keep stealing all the good hackers. We need help too, you know! :-)
Tene Coke: There's a sandwich place near my office that delivers orders made online. 17:11
japhb fperrad, that last error you came across shouldn't have happened -- because I fixed the bug that caused it. Or so I thought. Did you roll back plumage at some point?
Coke Tene: exactly that sort of thing, yes. =-)
japhb: over the 8 years I've been working on that project, I manage to do this only infrequently.
I think pmichaud is my biggest score ever in terms of partcl contributions stealing tuits that would otherwise go elsewhere. 17:12
mdiep is probably �2.
er, #2.
(stupid keyboard keeps switching in german)
Tene So, who wants to buy me lunch in exchange for patches? :) 17:13
PerlJam Coke: so it's *your* fault the ng branch hasn't merged yet! :)
fperrad japhb, bugs occur with the latest Plumage
japhb Right ... and I just found a second code path that I hadn't fixed.
just a moment. 17:14
Coke in my defense, hopefully partcl-nqp is identifying pain points in nqp-rx =-)
japhb Coke, please add to plumage-requests (which would then become badly named, but oh well): wiki.github.com/perl6/nqp-rx/plumage-requests 17:15
It's a semi-official pain points list
fperrad Tene, no actual question 17:16
japhb fperrad, OK, pushed. Please try HEAD 17:17
Coke japhb: Coke Also Wants 17:20
japhb Coke, good idea!
dalek rrot-plumage: 03a9c49 | japhb++ | src/lib/Plumage/Project.nqp:
[LIB] Plumage::Project: Fixes (I hope) for three bugs reported by fperrad++
 17:21
rtcl-nqp: d41402c | coke++ | TODO:
minor text cleanup
 17:26
fperrad japhb, OK fixed
japhb fperrad, excellent, sorry about that
17:38 colomon___ joined
dalek rrot: r42989 | fperrad++ | trunk/runtime/parrot/library/distutils.pir:
[distutils] add another Plumage action : update
 17:42
17:42 lucian joined
Coke wonders if there is anything left of the old NQP bootstrap in nqp-rx or if it has been... compiled out. 17:42
nopaste "fperrad" at 93.10.251.115 pasted "[PATCH Plumage] update section instructions (for japhb)" (527 lines) at nopaste.snit.ch/19039 17:53
"fperrad" at 93.10.251.115 pasted "[Plumage parrot-linear-algebra] the new description that uses distutils (for japhb)" (64 lines) at nopaste.snit.ch/19040
"fperrad" at 93.10.251.115 pasted "[Plumage] dependencies nightmare : the clean of lua-batteries requires the installation of mt19937" (132 lines) at nopaste.snit.ch/19041
dukeleto 'ello 17:58
japhb fperrad, paste 19040 looks broken 18:00
Oh, you're saying that's the way it is ... ah, looking 18:01
fperrad, nope, I was right the first time, 19040 is broken, and the error does not appear in my copy. 18:02
18:03 davidfetter joined
japhb 19041 is ... wow, sucky 18:03
That will require more work (I knew this was going to come up, as the depsolver is not too bright, but I was hoping to put off that rework for a while, sigh) 18:04
18:05 payload joined
nopaste "fperrad" at 93.10.251.115 pasted "[Plumage parrot-linear-algebra] a better one (for japhb)" (64 lines) at nopaste.snit.ch/19044 18:05
japhb fperrad, OK, metadata updates pushed. 18:14
I've got some other stuff to attend to before I can deal with the depsolver issue. How late will you be online? 18:15
18:15 bubaflub left, bubaflub joined
dalek rrot-plumage: 6c368c0 | japhb++ | metadata/ (2 files):
[METADATA] Fix DOS line endings
 18:16
rrot-plumage: b8efc51 | japhb++ | metadata/ (17 files):
[METADATA] Update parrot_setup based projects to latest capabilities
rrot-plumage: 1643f9e | japhb++ | metadata/parrot-linear-algebra.json:
[METADATA] Convert parrot-linear-algebra to use parrot_setup
fperrad japhb, the depsolver issue is not critical 18:17
I prepare an email with new feature request
18:17 Zak joined
japhb fperrad, thank you 18:18
Coke pmichaud: I am seeing occasional failures in t/cmd_time.t: 18:30
# got : 'STRING: 14889 microseconds per iteration'
# expected : 'REGEXP: [0-9]+ microseconds per iteration'
PerlJam Coke: What does TODO #1 mean? 18:31
Coke I thought it was clear. is there a particular bit that is confusing?
PerlJam $ ./partcl -e 'set x 5; incr x -37; puts $x' 18:32
-32
purl -32
Coke ah. try [incr x boo]
that should die with "expected integer but got "boo"
(see [time {expr 2+2} 1] vs. [time {expr 2+2} boo+ 18:33
er, [time {expr 2+2} boo]
dalek rtcl-nqp: 9024067 | coke++ | src/ (14 files):
.pm indicates a p6 file. add a hint for vim
 18:34
rtcl-nqp: 22c0ed4 | coke++ | t/ (62 files):
Don't need perl to re-invoke these as tcl.
rtcl-nqp: e0d7c8d | coke++ | t/ (62 files):
all our t files are written in tcl; give vim a hint.
18:34 chromatic joined
Coke PerlJam: (does that help?) 18:36
PerlJam Coke: aye
Coke: I was focusing on the part that reads "but we need to modify the grammar ... to allow optional +/-" and I couldn't see what needed implementing 18:38
Coke right, we have to break it before we can fix it. =-) 18:39
PerlJam How is type information stored in partcl?
Coke ? 18:41
PerlJam I've got a thing called x, how do I know it's string or number or list or whatever? 18:42
(What are tcl's data types?)
Coke PerlJam: I don't think the language exposes that. 18:43
you have arrays, and "everything else". under the covers, everything else might have different representations, but you can always try to use them interchangably.
so, array == assoc array (like a hash, but only one level deep) 18:44
then there are strings, lists, dicts (multi-tiered hash), float, int, (bignums)..
but you can try to use a string as a list and it'll JFW. (assuming the string can be interpreted as a properly formed list.) 18:45
Tcl is much smarter about caching the various representations for speed than partcl - partcl just morphs to the new type.
dalek rtcl-nqp: 90f6d21 | coke++ | build/Makefile.in:
Add a fulltest target to just run everything in t/ [*]
 18:51
pmichaud 17:42 * Coke wonders if there is anything left of the old NQP bootstrap in nqp-rx or if it has been... compiled out.
It's gone. It only exists in the history.
japhb Modern VCS apparently weighs in on the philosophical question of whether anything is ever really gone, if someone still remembers it ... 18:52
pmichaud (nqp-rx requests) I'm happy if the page wiki.github.com/perl6/nqp-rx/plumage-requests gets renamed to "nqp-requests" or even just "requests".
Coke pmichaud: how goes!
japhb I'll try to rename the page 18:53
pmichaud goes well for me at the moment... I finally figured out something approximating how I want the language library interop to work
now just need to implement it and document it
(hopefully finishing that _today_) 18:54
it'll be marked as "experimental" through 2.0, but it gives us something to start working with
chromatic msg mikehh Can you bisect to figure out where TT #1368 started to occur? 18:55
purl Message for mikehh stored.
japhb OK, couldn't figure out how to rename it outright, so I copied Plumage Requests => Requests, moved the non-plumage stuff to the top (since it's way shorter), and replaced the content in Plumage Requests with a pointer to the new page. 18:58
Coke anyone interested in trying to debug issues with the p6 syntax highlighter?
ww.
pmichaud afk, errands 18:59
dalek rtcl-nqp: 65f8da0 | coke++ | src/Partcl/commands/string.pm:
make p6.vim happier
 19:14
19:22 bacek joined
cotto_work hi bacek 19:26
bacek morning cotto 19:30
cotto_work happy Saturday! 19:34
19:44 mikehh joined
Coke pmichaud: so if I get to the point where I have to write a lot of PIR in partcl-nqp, that's bad, right? 19:47
19:47 joeri joined
pmichaud depends on the PIR 19:48
but yes, nqp tends to want to make that not necessary
at the same time, it recognizes that sometimes that's not entirely possible
Coke basically, since tclstring is in PIR due to the vtables, I am forced (yes?) to add any methods in PIR also.
(or can I at least write the methods in NQP?) 19:49
pmichaud you could write the methods in NQP, and then simply have the vtables delegate to the methods
unless they're supposed to be anonymous -- that makes it harder, yes.
Coke they are supposed to be anonymous, yes, but that can come later. 19:50
pmichaud I might be convinced to bump the priority on a "is vtable('xyz')" option for subs in nqp
Coke (these methods or vtables are not exposed to the HLL=
but doing the dispatch trick will get me further faster. 19:51
pmichaud right
and it's easy to clean up later
(just adding the 'is vtable' flag to the subs when it's supported)
Coke can the PIR class & the PM class have the same name or should I juggle them? 19:53
pmichaud same is okay, just be careful about which one does the class creating 19:54
if you define the class in .pm, no need to redefine in .pir
if you define in .pir, then just use "module TclString {...}" in the .pm 19:55
(to avoid the class creation)
eventually that'll likely be "augment class TclString {...}" to match p6, but just haven't got there yet.
19:58 colomon___ joined
Coke will go with module since I don't know how to duplicate the HLL mapping atm. 19:59
pmichaud wfm
PerlJam trying to modify partcl has just made me silghtly depressed. 20:24
Tene howso? 20:26
PerlJam just that nothing is working quite like I think it should. Ergo, I'm not making good progress.
Coke anything I can help with? 20:32
20:32 colomon__ joined
PerlJam Coke: nah, I just have a headache and am feeling completely unproductive. Unless you can figure a way to kick my brain out of its slump ... :) 20:33
dalek p-rx: 8ab2413 | pmichaud++ | (3 files):
Start rewriting HLL::Compiler in NQP instead of PIR.
 20:35
rrot: r42990 | bacek++ | branches/context_unify3:
Third attempt to merge Context and CallSignature
 20:42
whiteknight bacek++ 20:43
21:26 slash joined
slash hi all 21:27
can someone help a total newbie? :D
PerlJam slash: what sort of help do you need? 21:28
bacek *incoming*
bacek hiding under desk
slash PerlJam: I'm trying to implement a smalltalk-like language with PCT
PerlJam: I want to create a root object class, "SObject", but for some reason I can't subclass Object 21:29
PerlJam: I get "Parent is not a Class" error
PerlJam can you paste the code somewhere? 21:30
dalek rrot: r42991 | bacek++ | branches/context_unify3 (2 files):
Rename CallSignature into CallContext
 21:32
Coke nopaste?
purl i heard nopaste was at nopaste.snit.ch/ (ask TonyC for new channels) or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/ or paste.scsys.co.uk (for #catalyst, #dbix-class, #moose and others) or gist.github.com/ or paste or gtfo or tools/dev/nopaste.pl or trac.parrot.org/parrot/browser/tru...nopaste.pl
rrot: r42992 | bacek++ | branches/context_unify3/config/gen/makefiles/root.in:
Update make dependencies.
rrot: r42993 | bacek++ | branches/context_unify3/t/pmc/callcontext.t:
Rename callsignature.t into callcontext.t
rrot: r42994 | bacek++ | branches/context_unify3/src/call/args.c:
Update src/call/args.c to use CallContext.
rrot: r42995 | bacek++ | branches/context_unify3/src/pmc/callcontext.pmc:
Update accessors in CallContext.
rrot: r42996 | bacek++ | branches/context_unify3/src/pmc/class.pmc:
Update to CallContext in Class PMC.
rrot: r42997 | bacek++ | branches/context_unify3/src/pmc/callcontext.pmc:
Copy ATTRibutes and methods from Context into CallContext.
rrot: r42998 | bacek++ | branches/context_unify3/src/pmc/context.pmc:
Remove Context PMC.
rrot: r42999 | bacek++ | branches/context_unify3/config/gen/makefiles/root.in:
Update makefile dependencies
chromatic That looks promising. 21:34
bacek chromatic, all scaffolding is done.
pmichaud hope that doesn't break rakudo or ng too badly.
bacek pmichaud, I'll update rakudo if it will be needed. 21:35
pmichaud I'm guessing "yes" -- there appear to be a lot of references to CallSignature in the rakudo source
21:35 slash joined
slash PerlJam: sorry, my connection went down 21:36
chromatic Scaffolding?
purl well, Scaffolding is entering CPAN as we speak or like php, you'll end up with a pile of shit or CatalystX::ListFramework::Builder
pmichaud I don't see quite as many references in the ng branch, though.
bacek chromatic, last bits - invoke_from_sigobject shouldn't allocate context. Sub.invoke should only allocate registers and push CallContext
nopaste "slash" at 94.163.178.50 pasted ".HLL 'smalltalk' .sub init_lan" (10 lines) at nopaste.snit.ch/19046 21:37
bacek chromatic, (scaffolding) PMCs are merged.
pmichaud slash: you probably want get_class or get_namespace instead of find_var
bacek: are you planning to merge to trunk before or after Tuesday's release? 21:38
Tene slash: if you're interested in smalltalk, you might want to look for my prototype implementation, ChitChat
bacek pmichaud, erm... Probably after. I have to go now. And I'm not sure that I'll be able to finish it today.
Tene svn.parrot.org/languages/chitchat/trunk/
way out of date, though
pmichaud bacek: after would probably be better.
bacek pmichaud, deal :) 21:39
pmichaud if it happens before release, I'm not sure we'd be able to bring rakudo up-to-date before its release on Thursday. 21:40
chromatic Does Rakudo have its own CallSig/Context handling? 21:41
pmichaud Yes.
iiuc, we had to do that to get the perl 6 mmd semantics needed, plus a bunch of other stuff.
slash Tene: I've used ChitChat as a "reference" for my (ugly) smalltalk implementation... but it was incomplete when I tried it, so I've decided to start writing my own implementation as an exercise to learn PCT 21:42
pmichaud I don't know that we do much with contexts, but I know we do a fair bit with CallSig
oh, rakudo master doesn't look so bad either. 21:43
it just looked bad at first because I acked the generated code
chromatic Just curious; I don't want to merge before the release either.
21:44 purl joined
cotto_work go bacek! 21:44
slash pmichaud: I get a "Null PMC access in get_string()" error if I use get_class 21:45
nopaste "slash" at 94.163.178.50 pasted ".HLL 'smalltalk' .sub init_lan" (9 lines) at nopaste.snit.ch/19047
pmichaud $P0 = get_root_namespace ['parrot';'Object'] 21:46
slash pmichaud: I think that the problem is related to my use of the HLL directive
pmichaud obj = subclass $P0, ['SObject']
slash pmichaud: I get the same error 21:47
pmichaud one problem you may run into is that the sub will get executed twice as you have it now 21:48
once at :init, and again as the first sub in the compilation unit
dalek rrot: r43000 | bacek++ | branches/context_unify3/lib/Parrot (3 files):
Include pmc_callcontext in ops generated files
 21:49
nopaste "pmichaud" at 66.25.4.52 pasted "this works for me (for slash)" (11 lines) at nopaste.snit.ch/19048
rrot: r43001 | bacek++ | branches/context_unify3/include/parrot/context.h:
Typedef CallContext attributes struct as Parrot_Context.
rrot: r43002 | bacek++ | branches/context_unify3/tools/build/nativecall.pl:
Update NCI builder to include pmc_callcontext.h
rrot: r43003 | bacek++ | branches/context_unify3/src (27 files):
Mass replace include pmc_context with pmc_callcontext
rrot: r43004 | bacek++ | branches/context_unify3/config/auto/pmc.pm:
Fix dependencies in config/auto/pmc.pm
rrot: r43005 | bacek++ | branches/context_unify3/src/call/context.c:
Update context.c to use CallContext instead ot Context.
rrot: r43006 | bacek++ | branches/context_unify3/src/pmc/multisub.pmc:
Include pmc_callcontext in MultiSub PMC
rrot: r43007 | bacek++ | branches/context_unify3/compilers/imcc (8 files):
Update IMCC to include pmc_callcontext.h
21:53 purl joined 21:54 allison joined 21:59 brrant joined, jan joined 22:00 purl joined
dalek p-rx: 8010f63 | pmichaud++ | src/HLL/Compiler.pm:
Add src/HLL/Compiler.pm, omitted from previous commit.
 22:08
22:37 purl joined 22:39 purl joined 22:48 Whiteknight joined
japhb Tene, is this the right time of day to remind you about => in NQP-rx? Or should I ping you later? 22:58
Tene japhb: right now I'm working on p6advent post. 22:59
so, good reminder, but it'll be a bit.
japhb Tene++ # Good on ya!
k
ping japhb 23:00
oooh, hachi++ 23:01
msg hachi thank you for the purl ping fix!
purl Message for hachi stored.
nopaste "slash" at 94.163.178.50 pasted "addparent nightmare - fails with "Object must be created by a class."" (14 lines) at nopaste.snit.ch/19050 23:03
slash I solved the problem with the namespace... and now I've got another problem with addparent :D 23:04
23:06 bubaflub joined
pmichaud try intc.'add_parent'(objc) 23:07
the addparent opcode may be a little obsolete or something.
slash mhhh still the same error 23:08
pmichaud oh, it doesn't want to add a namespace as a class 23:09
you probably need to do
$P2 = get_root_namespace ['parrot';'Integer']
$P2 = $P2.'get_class'()
to get the class object
er, actually, you want to do that for Object 23:10
I'm guessing add_parent doesn't automatically convert namespaces into class objects the way that the subclass opcode does
slash pmichaud: uhm, objc should already be a class object (created by subclass) 23:14
pmichaud oh, you're right.
slash pmichaud: subclass is the only opcode that gets a namespace
pmichaud isa and get_class also take namespaces
most class-related opcodes are supposed to be able to take a namespace as a proxy for a class object. apparently addparent doesn't. 23:15
regardless, that's not the issue here.... hrm.
oh, the addparent is working fine 23:18
japhb I'm researching via Google right now, but anyone know offhand of some good papers on package dependency solving?
pmichaud the problem is apparently the 'new' line
23:18 purl joined 23:19 purl joined
slash pmichaud: maybe I should subclass from SObject and then add Integer as parent... now I'm doing the vice-versa 23:19
pmichaud ohhhh!
you don't need to subclass Object at all
you should just do newclass
all classes are already subclasses of Object 23:20
I bet that's confusing Parrot a bit.
slash yeah, it works!
thank you pmichaud :) 23:21
nopaste "pmichaud" at 66.25.4.52 pasted "pir subclassing (for slash++)" (19 lines) at nopaste.snit.ch/19051 23:22
23:24 purl joined
slash pmichaud: it works, thanks :) 23:24
23:30 purl joined
Coke PerlJam: any luck? 23:38
msg pmichaud will restructuring where generated files go cause you to commit less? 23:52
purl Message for pmichaud stored.
cotto_work What were the plans for tomorrow's hackathon? istr merging CallSignature and Context and strstart elimination. 23:57
Coke wonders how we are going to eliminate strstart. 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!