Parrot 1.9.0 "Blue-fronted Amazon" released! | parrot.org | Roadmap: icanhaz.com/parrotroadmap | Latest modified TT's: icanhaz.com/parrotbugs
Set by moderator on 16 December 2009.
NotFound Maybe a little magic with :flat will help, but I don't know how to use that in nqp or rakudo. 00:00
ash_ jnthn was suggesting that we override the method with a new one after compilation is finished
what is :flat ? 00:01
NotFound A parameter modifier.
To get lists of arguments as an array.
ash_ well, no its the result thats the problem 00:02
NotFound You can return the lists of roles flattened and let the caller packi it. 00:03
ash_ at compile time the result is a ResizablePMCArray, but at runtime it would be nice/better if it was a perl6 Array (or List)
but List and Array aren't defined at compile time
NotFound ash_: you need someone more used to rakudo than me 00:13
ash_ NotFound: alright, just checking if there was not already something that did almost this 00:14
00:19 lucian joined, Whiteknight joined
dalek nxed: r305 | julian.notfound++ | trunk/ (2 files):
parse modifiers with arguments in stage 1
 00:24
00:45 plobsing joined 00:52 abqar joined
Coke (not defined at compile time) you just need to do the mapping /early/ 00:53
before any other compilation happens.
old partcl did this; not sure if partcl-nqp does.
ash_ Coke: is that possible though? since List depends on the method to define itself, can you do that? 01:06
i guess i should explain the dependency, currently RoleHOW#methods returns a ResizablePMCArray, at runtime it should return an Array which is a child of List, which uses the role Positional, so in order to define List you need RoleHOW#method to be functional 01:10
Coke: can i try calling Array and if it fails catch the exception and make a ResizeablePMCArray instead? 01:14
dalek rtcl-nqp: 8ecd953 | coke++ | (2 files):
implement [lrange] - all tests pass.
 01:17
01:26 bacek joined 01:28 lucian joined
Coke ash_: your question makes me slightly dizzy. 01:35
I'm not sure; maybe you could do the mapping when list is defined (but before all its methods are added) 01:36
dalek rtcl-nqp: 5248255 | coke++ | (3 files):
implement [lreverse] - all tests pass
 01:39
01:46 nopaste joined
dalek rrot: r43177 | plobsing++ | branches/pmc_freeze_cleanup/src/pmc (5 files):
use VISIT_PMC_ATTR in key, scheduler, sub, and task PMCs
 01:47
01:50 kid51 joined 01:52 eternaleye joined 02:02 eternaleye joined
ash_ Coke: I ended up doing a try catch around it, i try to make an array, if that fails i make a ResizablePMCArray instead 02:10
i only catch 1 type of error to be sure i don't accidentally handle something else
02:15 lucian joined
dalek rrot: r43178 | plobsing++ | branches/pmc_freeze_cleanup (2 files):
remove non-functional general looking code
 02:19
02:30 JimmyZ joined 02:32 brrant joined 02:44 JimmyZ_ joined 02:57 plobsing_ joined 03:06 bacek joined 04:15 cognominal joined 04:27 patspam joined 04:29 TonyC joined
JimmyZ doesn't know how to change PMC_get_sub macro to use GET_ATTR syntax. 04:40
04:41 nopaste joined 05:35 TonyC joined 05:36 nopaste joined 05:41 TonyC joined 05:42 davidfetter joined 05:46 nopaste joined 06:07 bacek joined 06:13 TonyC joined 06:14 nopaste joined
JimmyZ oh hai, anyone active here? 06:26
06:31 abqar joined 07:44 bacek joined 07:58 iblechbot joined 08:20 cognominal joined 08:41 fperrad joined, bacek joined
fperrad bacek after last merge, parrot-nqp is broken 08:51
09:08 preflex joined 09:54 fperrad joined 10:04 bacek joined 10:34 gaz joined 10:43 clinton joined 11:29 lucian joined 11:36 lucian joined 12:11 bluescreen joined 12:13 bacek joined
mikehh nothing exciting happening 12:21
12:21 nopaste joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#31173), fulltest) at r43178 - Ubuntu 9.10 amd64 (g++ with --optimize) 12:36
Coke fperrad: yah, I pinged the list on that. 12:46
12:48 JimmyZ joined
Coke fperrad: I converted it into a ticket. 12:55
tt #1385
dalek TT #1385 created by coke++: context_unify3_simple breaks parrot-nqp 13:00
13:08 whiteknight joined
Coke msg pmichaud: so, we have a bunch of stuff in place for partcl-nqp now; I knocked out a bunch of the simpler commands this week; I'd like to focus on getting t/cmd_catch.t working 100%, as that includes the test for not compiling/executing the entire block at once. Any help you can provide on that front welcome. =-) 13:08
purl Message for pmichaud stored.
whiteknight good morning #parrot
JimmyZ doesn't know how to change PMC_get_sub macro to use GET_ATTR syntax. 13:17
whiteknight: good morning
whiteknight hello JimmyZ
JimmyZ saw there was a secret project on parrot. where is it now? 13:19
whiteknight JimmyZ: it fizzled out. I ran out of time. 13:20
JimmyZ whiteknight: are you free now? 13:21
whiteknight no.
JimmyZ oh, It's morning now. 13:22
whiteknight I have a few minutes now, maybe. What do you need?
JimmyZ hmm, I don't know how to change PMC_get_sub macro to use GET_ATTR syntax. 13:23
whiteknight I don't know what PMC_get_sub macro does. I'll have to look
dalek rrot: r43179 | whiteknight++ | trunk/docs/project/release_manager_guide.pod:
[RELEASE] Gerd++ has requested the 2.3 release.
 13:25
13:25 payload joined
Coke whiteknight: that's a starred release. 13:25
whiteknight JimmyZ: You can't replace PMC_get_sub with GET_ATTR. They do different things 13:26
JimmyZ whiteknight: I will check it again. 13:27
whiteknight: thanks.
13:29 plobsing joined 13:36 jan joined
JimmyZ whiteknight: PMC_get_sub is a bit like PMC_SUB macro; 13:43
whiteknight right
JimmyZ that also uses to get the attr of sub.pmc. 13:47
13:48 iblechbot joined
Coke msg pmichaud argh. my recent changes to use methods on TclString/TclList are now failing elsewhere because the HLL mapping isn't in effect everywhere. 13:49
purl Message for pmichaud stored.
Coke msg pmichaud (e.g. on t/cmd_catch.t - join is getting a String, not a TclString as one of the args.)
purl Message for pmichaud stored.
whiteknight matrixy doesn't use HLL mapping, and I'm thankful for that 13:59
13:59 patspam joined
whiteknight it's a system that never struck me as being too solid 13:59
it's sort of a chicken-and-egg problem. You can't HLL map without having a class to map to, but you don't have those classes till you create them at runtime 14:02
so constants at compile time aren't mapped 14:03
Coke and what are you using in place of it? 14:05
whiteknight I'm using Parrot's builtin PMC types directly
Coke (note that this sytem works fine in partcl)
whiteknight Matrixy isn't OO, so I don't need to worry about having methods
Coke partcl isn't OO either.
whiteknight oh.
moritz what about hll interop? 14:06
whiteknight moritz: what do you mean?
Coke now /there's/ something that's doomed to failure. =-)
moritz whiteknight: I thought HLL maps were a way to exchange Integers between two HLLs, and they come up magically as the "right" type in both HLLs 14:07
Coke moritz: no.
moritz: hll mapping is a way for a hll to map its types to core parrot types.
so that if you have something in core that generates an RPA, it can generate a List instead.
whiteknight moritz: Matrixy's integers are Integer PMCs. Other HLLs can translate those if necessary
Coke moritz: someone might extend that to map between mappings, but that's not what it was designed for.
whiteknight: do you have any container types? 14:08
whiteknight Coke: yes, but they aren't used in place of RPA
Coke np. how do you stringify them?
whiteknight Matrixy has some matrixy types. They stringify themselves
Coke ok. 14:09
whiteknight I don't stringify RPA's either. They're used internally only
I don't expose too many object types to the user
Coke ah, i suspect it's because I moved the mapping code in NQP from PIR, and now it's getting run too late. easy enough to fix. (though having BEGIN {} blocks would help.) 14:17
14:19 mikehh joined
JimmyZ good evening 14:52
purl Ah, evening. The tumultuous mind tarries and contemplates, reveling in the silence afforded by the diurnal proletariat. Good evening, indeed.
14:55 jsut joined
NotFound Winxed use a simpler approach: it doesn't have native types, uses parrot ones. 14:59
15:07 Psyche^ joined 15:08 bubaflub joined 15:09 PacoLinux joined 15:11 mikehh joined 15:16 payload joined
Coke I can't use parrot types for stringification; I want to make sure I support $S0 = $P0; so I have to use non-internal types when speaking to the user; I'd like to not have to manually translate between core types and partcl types all the time, so I find it easiest to do the mapping. 15:19
(stringification being one of 2 or 3 examples where the core types need vtable overrides to work with partcl) 15:20
whiteknight I'm certainly not arguing against the use of maps. I'm just acknowledging that they are not robust
Coke again, they work fine in partcl. =-)
(it's just that partcl-nqp adds another layer of complexity in there, with the use of nqp-rx.) 15:21
whiteknight What I would really like is a system where class definitions can be froze at compile time
and that would allow a more robust system of using maps at compile time
Coke You can do this now by using the :immediate scope. 15:22
define and map everything in the :immediate scope, and it is available for the rest of the compilation. 15:23
(my problem was that I translated this to an INIT {} block in NQP, which translates to :init) 15:24
so I need to undo that and I should be fine.
whiteknight that's not really the same thing as a frozen constant, and :immediate has it's own problems
Coke I'm just trying to address the vague "robust" concern. 15:26
whiteknight in any case, I don't use them so I can only go off the information I hear. You say it works well enough, that's good enough for me 15:28
Coke needs to get partcl-nqp working on his wintop so he can hack on it over xmas break. 15:29
Anyone here using git fer windows? 15:30
whiteknight me
Coke how's the usability?
whiteknight good, if you don't do anything fancy
Coke you using the tortoise version, or the command line version?
whiteknight command line. I wasn't able to find a good version of tortoisegit 15:31
Coke you using github?
whiteknight or maybe I heard bad reviews
yes
Coke any problems setting up a key? (did you putty gen for that?) 15:32
*use
dalek TT #475 closed by doughera++: Could not build parrot-1.0.0 on ppc64
whiteknight key generation was actually pretty easy, it shipped with the ssh-keygen utility 15:33
I had problems because my wonderful system administrator helpful set my "home" folder to be somewhere off in the network share drive 15:34
and git couldn't find it
but if you don't do anything "helpful" like that, you should be fine
Coke can't do a clone from a work machine; will try at home with a more open fw. 15:35
whiteknight ok
15:37 payload joined
dalek rrot: r43180 | coke++ | trunk/DEPRECATED.pod:
This ticket was moved to Trac. No longer need pointer to old rt queue from here.
 15:53
15:56 theory joined
Coke what's the difference between config/gen and config/auto? 16:03
(and init, for that matter. =-) 16:06
plobsing AFAICT, auto detects things (and should be overridable by --ask), gen acts on that information to generate files 16:08
Coke ok. I have something that needs to be specified on the config command line, but have a sane default. I am thinking that 'init' make sense here. (there is no detection for this step.) 16:09
PerlJam Coke: I was just looking at partcl's glob grammar. Does tcl globbing really support alternation via | ? 16:15
plobsing If it set up the config process, I'd still put it under auto 16:16
s/if it set up/if it *didn't* set up/
16:20 mikehh joined
dalek nxed: r306 | julian.notfound++ | trunk/winxedst1.winxed:
implement dotted new in stage 1
 16:21
Coke PerlJam: nope. 16:23
probably a hold over from the copy from ARE.
I just made sure that t/tcl_glob.t passed =-) 16:24
can you fix it?
(where do you see the alternation? ah, nibbler?) 16:25
16:25 particle joined
dalek nxed: r307 | julian.notfound++ | trunk/ (3 files):
use Getopt.Obj() instead of 'Getopt::Obj', parrot TT #1372
 16:30
16:32 cognominal joined
Coke wonders if there is a way for a Configure.pl step to say "look, this is serious, Configure - give up." 16:32
particle i think there's a way for a step to die fatally if it fails 16:33
Coke ok. it's not "die", nor is it listed in docs/configuration.pod - I don't mind, per se, but this seems like something that should be available. 16:34
plobsing exit(1)?? 16:35
particle the term i remember is 'fatal'. i could be wrong, but try searching for that
under config/ and lib/ probably 16:36
16:36 lucian joined
particle i'm in the process of rebooting, so it'll take me a while to help 16:36
Coke exit_fatal() ?
particle sounds good
<purl>speakers so loud...</purl> 16:37
Coke hurm.
nope, that's in C. 16:38
exit++ 16:39
Coke wonders why Configure.pl is checking for tests for steps when run in plain vanilla mode. 16:40
16:46 Zak joined
Coke cotto-- 16:53
(missing build dep) 16:58
dalek rrot: r43181 | coke++ | trunk/config/gen/makefiles/root.in:
Fix file dep broken in r42521 for src/io/win32.c
rrot: r43182 | coke++ | trunk/config/gen/makefiles/root.in:
Add another missing build dep.
17:06 davidfetter joined
cotto_work good morning 17:32
dukeleto good moroning 17:35
plobsing is there a good way to tell GC about a Buffer* I have (and would like to keep)? 17:45
tewk_ plobsing, make sure it stays on the stack. 17:47
cotto_work plobsing, failing that you can also add it to the root set directly, though that's not a good general approach.
plobsing I'm pretty sure its on the stack, but it keeps getting gc'd out from under me. 17:48
tewk_ Its not on the stack then.
plobsing, code reference? is this the visit_info in pmc_freeze? 17:49
plobsing tewk_: it's the Buffer * in visit_info in pmc_freeze
the visit_info struct is being stack-allocated AFAICT, so the buffer should be on the stack 17:50
exact code location: branches/pmc_freeze_cleanup/src/pmc_freeze.c line 1161 17:51
tewk_ looking 17:52
dalek rrot: r43183 | coke++ | trunk/tools/dev/checkdepend.pl:
Check files in $(IO_DIR); fix sense of comparison for diagnostic output.
 18:03
bubaflub last #ps i was told someone was going to flip my commit bit, but i tried committing last night and got rejected. who should i talk to? 18:27
cotto_work Anyone listed as a metacommitter in RESPONSIBLE_PARTIES should be able to do it. 18:30
chromatic and Coke are probably your best bet.
dukeleto Coke: FIRE ON DECK 18:31
bubaflub: we will get that sorted out 18:35
bubaflub dukeleto: thanks.
dukeleto bubaflub: who was the 'someone' that told you your bit was getting flipped? 18:36
bubaflub: did they say they got your CLA?
bubaflub dukeleto: yep. last ps. let me check the logs.
18:37 cotto joined
cotto_work wb me 18:37
bubaflub dukeleto: irclog.perlgeek.de/parrotsketch/200...#i_1840624 18:38
dukeleto: unless that's chromatic's royal we, i'm not sure
Coke bubaflub: what's your trac id? 18:44
bubaflub Coke: bubaflub 18:45
purl it has been said that bubaflub is mailto:bobjkuo@gmail.com
bubaflub purl: bubaflub is Bob Kuo, bobjkuo@gmail.com
purl ...but bubaflub is mailto:bobjkuo@gmail.com...
bubaflub fine. have it your way.
cotto_work no, bubaflub is Bob Kuo, bobjkuo@gmail.com
purl okay, cotto_work.
bubaflub haha. 18:46
cotto_work If you don't mind waiting, we can give you a commit bit for Christmas. ;) 18:47
bubaflub cotto_work: now i feel bad cause i didn't get you anything...
bubaflub scrambles for cheap Christmas present...
Coke who is bubaflub's mentor? 18:50
cotto_work istr that dukeleto volunteered 18:51
Coke (because tag, you're it.)
cotto_work saw that coming.
Coke bubaflub: you now have a commit bit. It is typical for you to add your name to CREDITS mentioning any work you've done so far that was applied on your behalf. 18:55
do not break the build.
bubaflub Coke: roger wilco. i'll try committing an updated CREDITS.
Coke if you're bored, you could make tools/dev/checkdepend.pl pass again. =-)
cotto_work bubaflub, since I appear to be your mentor, let me know if you have any questions. 18:56
bubaflub cotto_work: cool. i have patches in the hopper that convert some perl tests to PIR.
cotto_work: i'm just going to apply the patches to trunk, run a make fulltest, and if that passes commit
cotto_work: does that sound sane? 18:57
cotto_work sure
I'll review your commits and mention anything that looks odd.
bubaflub cool.
cotto_work If I'm not around, email me or msg me via purl. I try to be responsive to both. 18:58
Coke particle: jimmyz's cla was stuck in the moderator queue. enjoy. 19:02
bubaflub: your commit message was also stuck in the moderator queue. released, whee.
Coke hurls www.davidbrunton.com/2009/12/merry-...html#links 19:03
cotto_work It's nice to see people getting real wor^H^H^Hgolf done using Rakudo. 19:07
dukeleto bubaflub: so you are all set now? 19:08
dalek rrot: r43184 | bubaflub++ | trunk/CREDITS:
add myself to CREDITS
bubaflub dukeleto: bingo.
dukeleto bubaflub: looks like it!
purl No it doesn't, shut your hole!
bubaflub Coke: tools/dev/checkdepend.pl doesn't actually check that the Makefile exists before trying to open it (throwing a wonderful error). should i add a one line check + die if it doesn't exist? 19:13
Coke sadly, it doesn't work for me in rakudo.
bubaflub: I suppose. 19:14
bubaflub okey dokey
Coke (it requires more than just the makefile, but that's reasonable to check)
19:17 bacek joined
Coke bubaflub: I notice I misspelled 'construct' a few times, too. =-) 19:19
particle msg JimmyZ received your completed cla, thanks! 19:24
purl Message for jimmyz stored.
dalek rrot: r43185 | bubaflub++ | trunk/tools/dev/checkdepend.pl:
make sure Makefile exists before trying to open it
Coke bubaflub: that die should probably end with a "\\n" to avoid the line number output. Might also be good to make it "open or die", and to mention you need the full build. 19:27
bubaflub++
bubaflub Coke: so combine the open and die lines. got it.
Coke yah. that will cover any other bizarre reasons they might not be able to read the file. 19:28
19:29 japhb joined
bubaflub Coke: this is strange... i've got the open ... or die ... but it's still failing on opening the file 19:36
Coke: what i mean is that it's still throwing the error rather than my die message
dukeleto bubaflub: nopaste? 19:37
purl nopaste is probably 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
bubaflub gist.github.com/261179 19:39
am i just being dull? 19:40
dull in the sense of slow.
no doubt i'm boring. 19:41
19:41 mikehh joined
dukeleto bubaflub: checking it out now 19:44
19:44 bacek_ joined
bacek Morning. Good morning 19:45
bubaflub mornin' bacek
dukeleto bacek: o hai
bubaflub: what does the full script look like? does it have autodie? 19:46
bacek aloha bubaflub and dukeleto
bubaflub dukeleto: negative on the autodie
dukeleto pours bacek an irish coffee
bacek dukeleto, sigh... I wish. But it's way too early... 19:47
bubaflub bacek: what's your timezone? IIRC, you're in England, right?
bacek bubaflub, nope. It's summer here :) 19:48
7am of good summer day.
clock?
purl bacek: LAX: Mon 11:48am PST / CHI: Mon 1:48pm CST / NYC: Mon 2:48pm EST / LON: Mon 7:48pm GMT / BER: Mon 8:48pm CET / IND: Tue 1:18am IST / TOK: Tue 4:48am JST / SYD: Tue 6:48am EST /
bubaflub bacek: heh. what a bad guess - wrong hemisphere 19:49
19:49 bubaflub left, bubaflub joined 19:51 mikehh joined
dalek TT #1385 closed by bacek++: context_unify3_simple breaks parrot-nqp 19:53
a: b0c5eed | fperrad++ | p (2 files):
move plumage in ports/plumage
 19:56
rrot: r43186 | bacek++ | trunk/ext/nqp-rx/src/stage0/Regex-s0.pir:
Update nqp-rx to use getattribute syntax in CallContext
 19:57
rrot: r43187 | fperrad++ | trunk/runtime/parrot/library/distutils.pir:
[distutils] when install, chmod only on cygwin or hpux platform
moritz bacek: isn't Regex-s0.pir a generated file? 19:58
bacek moritz, it is. pmichaud updated upstream already. I just patched ext/ version in-place 19:59
moritz bacek: ok
dalek l: 034a4ba | fperrad++ | p (2 files):
move plumage in ports/plumage
rkdown: d4e1c99 | fperrad++ | p (2 files):
move plumage in ports/plumage
 20:00
lscript: 336bdc8 | fperrad++ | p (2 files):
move plumage in ports/plumage
bacek msg whiteknight Haskel ftw! Check research.microsoft.com/en-us/um/peo.../index.htm With "VTABLE swap" we can solve "old-to-young" problem in generational gc easily 20:01
purl Message for whiteknight stored.
bubaflub Coke: my open or die is still throwing the wrong error. i'm heading home from $work right now and will hack on this tonight. feel free to msg or email me with other stuff you'd like to see in that script.
Coke you have a nopaste? 20:02
ok.
20:02 payload joined
Coke bacek: seems to work for me. Danke. 20:03
20:03 Zak joined
bacek Coke, nqp-rx? 20:03
Coke yes.
bacek Glad to hear it. 20:04
20:13 cognominal joined 20:17 TimToady joined
Coke Cheese it, it's the architect. 20:18
(no, the other one.)
TimToady: hio.
dalek nxed: r308 | julian.notfound++ | trunk/winxedst1.winxed:
implement constants and improve optimize step in stage 1
 20:20
20:27 xenoterracide joined
Coke msg bubaflub do you want a google wave invite? 21:01
purl Message for bubaflub stored.
Coke (or any other committers?)
NotFound Coke: I have one, but don't know anything interesting to do with it. 21:02
21:04 bacek joined 21:05 joeri joined
dalek nxed: r309 | julian.notfound++ | trunk/winxedst1.winxed:
more optimize propagation, some refactoring and several new predefs in stage 1
 21:08
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#31187), fulltest) at r43187 - Ubuntu 9.10 i386 (g++ with --optimize) 21:31
Coke ugh. the fact that pirc is designed to be compiled into its own binary is frustruting. 21:36
pirc?
purl pirc is kjs's or lexx/yacc based
Coke msg kjs So, I'm looking at adding Configure.pl --pirc=pirc; My plan is to kill 'pirc' and just make it the registered comiler for PIR/PASM. - does this make sense to you? 21:37
purl Message for kjs stored.
Coke msg kj (see kjs)
purl Message for kj stored.
japhb pmichaud, ping 21:46
Anyone know if pmichaud is around this week? 21:54
Also, will there be a #ps tomorrow? 21:55
cotto_work japhb, good question. I'll be around but I suspect it'll be sparse.
japhb cotto_work, OK, thanks 21:56
mikehh I should be around 22:00
23:33 plobsing joined
dalek nxed: r310 | julian.notfound++ | trunk/winxedst0.cpp:
fix a bug in annotations and add some more in stage 0
 23:40
nxed: r311 | julian.notfound++ | trunk/winxedst1.winxed:
small fixes in hash initializer and identifier undefined report in stage 1
 23:50
23:53 tetragon joined
japhb Is there a way to specify an inout param in NQP-rx op__args() syntax? 23:56
cotto_work Is it not working the way you'd expect it to work? 23:58
dalek TT #1386 created by kurahaupo++: Convert t/pmc/exporter.t to PIR

Powered by Cro and the Raku® Programming Language.

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

Thank you!