Parrot 0.8.2 "Feliz Loro" Released www.parrot.org/news/2008/Parrot-0.8.2
Set by moderator on 23 December 2008.
00:08 AndyA joined 00:10 chromatic joined 00:11 Whiteknight joined
dalek r35462 | kjs++ | trunk/compilers/pirc/src (3 files): 00:14
: [pirc] PIRC didnt' properly unescape double-quoted strings. This is fixed now, but in a hacky and inefficient way. This must certainly be fixed at some point. Preferably there is a string_unescape_cstring() function returning a c-string, not a STRING. One solution would be to have a function string_unescape_cstring_to_cstring() that does the main work, and string_unescape_cstring calling that function and converting it into a STRING. Otherwise I'll probably
review: www.parrotvm.org/svn/parrot/revision?rev=35462
jonathan -> sleep 00:18
dalek r35463 | Whiteknight++ | branches: 00:35
: creating a branch to rip out STM
review: www.parrotvm.org/svn/parrot/revision?rev=35463
00:35 Limbic_Region joined 00:36 Whiteknight joined 00:54 Fayland joined
dalek r35464 | Whiteknight++ | branches/removing_stm/src (2 files): 01:00
: [remove_stm] remove code references to STM from core files
review: www.parrotvm.org/svn/parrot/revision?rev=35464
cotto Whiteknight, don't forget those silly stub STM PMC tests in t/pmc/stm*t. 01:06
Whiteknight Yeah, I'm pulling out the pmcs themselves right now. I'll get the tests after that 01:07
cotto They won't be missed.
Whiteknight++
Whiteknight I only do it because I'm a karma whore
:)
dalek r35465 | Whiteknight++ | branches/removing_stm (4 files): 01:09
: [remove_stm] removing the src/pmc/stm* PMCs
review: www.parrotvm.org/svn/parrot/revision?rev=35465
01:09 davidfetter joined
dalek r35466 | Whiteknight++ | branches/removing_stm (4 files): 01:11
: [remove_stm] remove the silly t/pmc/stm* tests for the PMCs I just yanked
review: www.parrotvm.org/svn/parrot/revision?rev=35466
01:13 tetragon joined
dalek r35467 | Whiteknight++ | branches/removing_stm (3 files): 01:16
: [remove_stm] remove include/parrot/stm/* and src/stm/* code files
review: www.parrotvm.org/svn/parrot/revision?rev=35467
01:17 AndyA joined
dalek r35468 | Whiteknight++ | branches/removing_stm/docs: 01:19
: [remove_stm] remove stm-related docs
review: www.parrotvm.org/svn/parrot/revision?rev=35468
r35469 | Whiteknight++ | branches/removing_stm:
: [remove_stm] remove stm-related docs from MANIFEST
review: www.parrotvm.org/svn/parrot/revision?rev=35469
r35470 | Whiteknight++ | branches/removing_stm (2 files): 01:21
: [remove_stm] remove src/ops/stm.ops
review: www.parrotvm.org/svn/parrot/revision?rev=35470
01:21 particle joined
GeJ Good morning everyone 01:22
Whiteknight good morning GeJ 01:23
cotto morning Gej
dalek r35471 | Whiteknight++ | branches/removing_stm/lib/Parrot/Pmc2c (2 files): 01:26
: [remove_stm] removing some references in the pmc2c perl code
review: www.parrotvm.org/svn/parrot/revision?rev=35471
r35472 | Whiteknight++ | branches/removing_stm/t: 01:27
: [remove_stm] removing some other stm-related tests
review: www.parrotvm.org/svn/parrot/revision?rev=35472
GeJ hello Whiteknight, cotto
dalek r35473 | Whiteknight++ | branches/removing_stm (2 files): 01:30
: [remove_stm] remove some more STM references from perl code
review: www.parrotvm.org/svn/parrot/revision?rev=35473
r35474 | Whiteknight++ | branches/removing_stm/runtime/parrot/library: 01:46
: [remove_stm] remove runtime/parrot/library/STM.pir
review: www.parrotvm.org/svn/parrot/revision?rev=35474
01:51 jimmy joined
jimmy test 01:51
cotto test? 01:52
purl testing 123 abc 4red 12blue 9green
jimmy I am using web irc at mibbit.com
cotto that's nice when someone's blocking the irc port 01:53
jimmy yes, the firewall blocked me. 01:54
cotto good old Universal Firewall Bypass Protocol, aka http 01:56
Whiteknight how do we renumber the ops? 01:57
cotto GeJ, have you mailed in a CLA?
Coke_away in a config'd parrot, "make opsrenumber"
Coke then re-config and rebuild.
also, update PBC_COMPAT
Whiteknight Coke++
that works fine
jimmy still don't know why there were duplicate lines in PBC_COMPAT. 01:59
Coke we can delete the duplicate. the whole file will probably be wiped as of 1.0
jimmy coke++ again. 02:00
TiMBuS i put my code on github and realized i dont have any copyright in there or anything. does parrot mandate a specific license or anything? 02:02
02:04 jimmy left
GeJ cotto: Oh, I thought that it worked th other way around : you propose my name during #ps, and if I get approved I send the CLA. 02:04
But I can send one tomorrow anyway.
02:05 allison joined
GeJ When is #ps starting? 02:05
02:05 jimmy joined
dalek r35475 | Whiteknight++ | branches/removing_stm (6 files): 02:06
: [removing_stm] a few misc changes to get Parrot to build again, and update BC_COMPAT
review: www.parrotvm.org/svn/parrot/revision?rev=35475
GeJ Whiteknight: I thought stm was a good thing. Like it allowed us to have continuations...
Whiteknight GeJ, you're about 17 hours early
GeJ, STM isn't being used currently
and what it does is allows inter-thread communication
GeJ oh yes... your tuesday afternoon... sorry. 02:07
well, then rip off the damn thing, I say. 02:09
Whiteknight don't worry, I am ripping the hell out of it 02:14
this poor subsystem is going to wish it was never written
cotto GeJ, I'm just trying to streamline the process. 02:20
The CLAs usually take a while to arrive and get processed (i.e. 1-2 weeks).
Whiteknight yeah, mine took a while 02:21
cotto It's not urgent, though. I'll happily continue applying your patches.
Whiteknight and they tried to speed it up for GSOC
GeJ cotto: Ok, I'll send mine first thing tomorrow. 02:22
Whiteknight GeJ++ 02:25
dalek r35476 | Whiteknight++ | branches/removing_stm/t/steps: 02:37
: [removing_stm] remove another stm mention in the tests
review: www.parrotvm.org/svn/parrot/revision?rev=35476
Whiteknight okay, I'm making fulltest right now, and assuming the log comes back clean (or mostly clean) I'm going to merge the branch back in 02:43
02:45 kid51 joined
kid51 Good evening. 02:45
Coke ping 02:46
Whiteknight good evening kid51 02:47
kid51 Hello. Am wondering: Does anyone have anything to say about trac.parrot.org/parrot/ticket/160 (the code stuff, not the SVN stuff)? 02:48
GeJ heya kid51 02:49
kid51 Hi. Did you have any more concerns about that patch -- particularly the Unicode part.
02:51 AndyA joined
nopaste "GeJ" at 202.22.229.231 pasted "t/library/data_escape.t: Unicode tests, Perl vs. PIR" (70 lines) at nopaste.snit.ch/15291 02:58
Whiteknight goodnight gentlement 02:59
GeJ This is a before-after view of the test for anyone interested.
I think the logic is good (I should have added comments), but I may be wrong. 03:00
jimmy Gej++ 03:01
karam GeJ 03:02
GeJ guesses 22
karma me 03:03
purl gej has karma of 24
03:05 tewk joined 03:06 SunConure joined
SunConure I wish Ovid's blog post at use.perl.org/~Ovid/journal/38271 was a tad more descriptive. :\\ 03:07
An example would perhaps make it ever so slightly more digestable. :-)
Because I'm left, at the very least, interested in seeing if I can make a contribution there. But there's insufficient specificity for me to make that determination. 03:09
jimmy SunConure++
Tene pmichaud: ping?
jimmy SunConure: here trac.parrot.org/parrot/wiki/NewPar...loperGuide 03:10
SunConure I guess I'm wondering if, from just knowing Perl 5 well, can I "Then all you need to do is fix one of the Pugs tests listed in the parrot repository's languages/perl6/t/spectest.data" ?
cotto pugs stuff happens more in #perl6 on Freenode, although lots of Rakudo stuff happens here 03:11
SunConure jimmy: thanks - I'll check it out momentarilly
GeJ SunConure: if you're looking at working on the tests specs, you may want to have a look at #perl6 on Efnet.
s/Efnet/Freenode/
cotto++
SunConure Can someone that just know Perl 5 (well) contribute effectively to parrot itself? 03:12
s/know/knows/
jimmy did I mistake?
Tene purl: msg pmichaud I'm getting test failures with my patch to move rakudo to .HLL 'perl6' because of null pmc errors with string arguments to isa 03:13
purl Message for pmichaud stored.
Tene purl: msg pmichaud isa element, 'Perl6Scalar'
purl Message for pmichaud stored.
kid51 SunConure: Well, I fall into that category and I've been committing since late Dec 2006.
cotto SunConure, we have a lot of perl5 infrastructure
SunConure :O
I'm *very* interested in helping there. I do Perl infrastructure work at work :-) 03:14
nopaste "tene" at 166.70.38.237 pasted "Preliminary .HLL 'perl6' patch for pmichaud++" (307 lines) at nopaste.snit.ch/15292
kid51 SunConure: OTOH, that infrastructure is simultaneously (a) well-developed and (b) scheduled to be gradually replaced so that Parrot is no longer dependent on Perl 5.
SunConure That makes sense.
Parrot will eventually want to be self-hosting :-) 03:15
cotto it'll be a while, though
kid51 There are a couple of RT tickets that have not been touched that could be pure Perl 5: rt.perl.org/rt3/Ticket/Display.html?id=58990 03:16
rt.perl.org/rt3/Ticket/Display.html?id=58990 03:17
SunConure So, the infrastructure stuff in Perl 5 seems to be in tools/* ?
cotto and lib
kid51 It's all over. Configure.pl is Perl 5. tools/ -- yes. And, corresponding thereto, t/tools/* 03:18
lib/Parrot/ as well
SunConure ah yes
Coke kid51: pong.
kid51 Coke: Are you satisfied with trac.parrot.org/parrot/ticket/160 ? 03:19
Coke I haven't reviewed the changes, no. 03:20
Doesn't need to be me, though. 03:21
sorry.
kid51 Well, I don't know enough PIR to absolutely confirm that GeJ did the transformation correctly. And GeJ had some concerns about the Unicode tests. 03:22
Coke Yup. I'd just mark the ticket as needed someone to review it, give it away, and leave it open.
kid51 (The tests are all passing.)
ok, that's what I did. 03:24
Coke +1 03:25
purl 1
03:26 jdv79 joined
jdv79 does rakudo build the STD grammar? 03:26
03:27 Fayland joined 03:32 particle1 joined
kid51 I'm seeing this too often: Oopsļæ½ 03:33
Trac detected an internal error:
OperationalError: database is locked
purl i already had it that way, kid51.
jimmy hadn't seen that yet. 03:34
03:37 Fayland joined
GeJ kid51: I just added one last comment to the ticket for posterity (and summerizing my concerns) 03:37
kid51 Thx. If no one else has anything to say, I'll close the ticket in a couple of days. 03:38
kid51 must sleep
purl $kid51->sleep(8 * 3600);
GeJ good night
cotto jdv79, Rakudo's (decreasingly divergent) STD.pm is in languages/perl6/src/parser/grammar.pg 03:47
04:01 elmex_ joined 04:04 ask_ joined 04:18 silug joined 05:08 AndyA joined 05:18 chromatic joined
Coke_away chromatic: hello. 05:23
Coke realizes a bit late that he had meant to dial back into work this evening.
chromatic ahoy 05:24
GeJ what time is it on the west coast? 10:30 PM ? 05:30
purl GeJ: It's just about half past five in the morning where I am.
05:33 leto joined
dalek r35477 | tewk++ | trunk/languages/ecmascript (7 files): 05:36
: [js] moved to a .HLL, removed incorrect methods Set and Get from Array, add get_string to Array
review: www.parrotvm.org/svn/parrot/revision?rev=35477
Coke time? 05:41
purl time is probably 05:41:20 2009 and (did you mean "clock"?) or flowing like a river
Coke clock?
purl Coke: LAX: Mon 9:41pm PST / CHI: Mon 11:41pm CST / NYC: Tue 12:41am EST / LON: Tue 5:41am GMT / BER: Tue 6:41am CET / IND: Tue 11:11am IST / TOK: Tue 2:41pm JST / SYD: Tue 4:41pm EST /
Coke GeJ: there ya go.
time? clock?
05:44 particle joined 05:47 allison joined
Coke particle: you still working remotely for $DAYJOB ? 05:55
06:03 jimmy joined
Coke allison: you around? 06:13
allison Coke: yup
GeJ Coke: shiny... thanks :) 06:21
dalek r35478 | coke++ | trunk: 06:24
: add another missing ticket #.
review: www.parrotvm.org/svn/parrot/revision?rev=35478
06:32 AndyA joined 06:49 jsut|work joined
Coke allison: hurm. 06:51
I think I summed it up in mail to list. 06:52
"DEPRECATION Roundup, 0.9.0"
I have no idea how to fix the last remaining issue for remove_pic
chromatic Allison or I will have to do that probably.
Coke that's and commenting on #47011 would probably be the biggest bang for the buck from allison. 06:53
I can probably do 58410 and 48014 if time permits.
be nice if 60626 could be administratively resolved.
ISTR chromatic has found that function cannot be removed by mortal men.
I can only hope that 60586 is a SMO-merging. 06:54
(but I doubt it)
I have an idea on 36283. 06:55
no, that won't work (have another :flag to mark something as amenable to assign syntax. still a PITA, not more help than just having an initial OUT param.) 06:56
06:59 pdcawley joined
chromatic mmd_cvt_to_types isn't going to go away any time soon. 07:03
allison chromatic: what's using it? 07:04
purl it has been said that using it is asking for troubles, it will cause more work than doing right from the beginning
Coke my vote? resolve the ticket, but note that one remaining function.
chromatic All MMD dispatch uses it.
07:04 Fayland joined
allison chromatic: ah, right that's down in the low-level MMD guts that I left as-is 07:05
chromatic *Something* has to convert a type name to a type number.
allison chromatic: we could at least rename it to something sane
chromatic: yup
chromatic I'll think of a better name.
allison I would consider a sensible rename "done" on that one. 07:06
Coke hurm. does trac have an option to enable dates for tickets? 07:07
(start, end, estimated completion)
(not for parrot)
allison Coke: I don't see a way of adding arbitrary fields. But, if we had them, would people use them? I've never used that stuff in RT. 07:11
Coke (not for parrot)
allison Coke: ah, I see, I was having trouble parsing that phrase into the whole
Coke: there might be a plugin for track 07:12
trac
there
there's a plugin for just about everything
trac-hacks.org/ 07:13
jimmy 3could Timeline be more temporary at trac? it keeps a month. 07:15
allison Coke: on RT #47011, does "blech, I'm sick of revisiting this stupid ticket, make it go away" count as a comment? No, I suppose not. :) I'll make a more useful comment. 07:16
Coke allison: ISTR we had a conversation about this, which I don't seem to have recorded on ticket. My apologies. 07:17
I must go to bed. too much dayjob, and it starts again in 5 hours or so. 07:18
allison oh, not a bad on you, mainly just a frustration with old code that depends on broken old code :)
night
Coke: and, I'm trying to remember the details of the conversation too 07:19
07:40 namenlos joined 07:51 iblechbot joined 07:56 Fayland_ joined 07:58 ask_ joined
dalek r35479 | allison++ | branches/pcc_invoke_convert/src/pmc: 08:00
: [calling_conventions] Re-removing unnecessary mark from CPointer PMC.
review: www.parrotvm.org/svn/parrot/revision?rev=35479
r35480 | allison++ | branches/pcc_invoke_convert/src:
: [calling conventions] Some conversions of Parrot_PCCINVOKE that pass all tests.
review: www.parrotvm.org/svn/parrot/revision?rev=35480
08:04 particle1 joined
moritz pmichaud: spectest-progress.csv says that we dropped from 6172 passing tests (2009-01-09) to 6141 (2009-01-10) - do you know what happened? did we really regress on 30 tests? if so, we should have a ticket for those IMHO 08:10
08:36 Hinrik joined 08:38 jimmy joined
dalek r35481 | simon++ | branches/strings/docs/pdds: 09:05
: A raw "C string" doesn't mean anything any more, must have encoding and
: charset. No default can be assumed because only the producer knows what it is.
review: www.parrotvm.org/svn/parrot/revision?rev=35481
09:18 masak joined 09:27 donaldh joined 09:35 jimmy left 09:36 jimmy joined 09:41 riffraff joined 09:43 namenlos joined 09:52 jimmy left 10:11 barney joined 10:18 jimmy joined 10:22 cotto joined 10:24 particle joined 10:35 cotto joined
cotto clearwire-- 10:36
(at least it works eventually) 10:37
10:39 alvar joined 10:54 elmex joined
dalek r35482 | fperrad++ | trunk/languages/lua/src/lib: 10:59
: [Lua] markdown
: - update Markdown interface (see r35437)
review: www.parrotvm.org/svn/parrot/revision?rev=35482
r35483 | fperrad++ | trunk/languages/lua/src/lib (2 files): 11:04
: [Lua] traceback
: - refactor
review: www.parrotvm.org/svn/parrot/revision?rev=35483
r35484 | fperrad++ | trunk/languages/lua/t: 11:05
: [Lua]
: - fix test : now, hash seed is randomized
purl i already had it that way, dalek.
dalek review: www.parrotvm.org/svn/parrot/revision?rev=35484
cotto purl, ignore dalek
purl cotto: sorry...
cotto purl, make me a sandwich 11:06
purl cotto: sorry...
cotto purl, apologize
purl cotto: i'm not following you...
cotto purl, act confused
purl cotto: i'm not following you...
masak purl's nemesis? 11:08
purl, purl's nemesis is masak.
purl OK, masak.
masak purl's nemesis?
purl it has been said that my nemesis is masak.
11:12 rdice joined
cotto purl, your nemesis? 11:16
purl purl's nemesis is masak.
cotto purl, your mom?
purl pemungkah's mother is a saint! or capable of multiplexed I/O.
cotto purl, your mom?
purl pemungkah's mother is a saint! or capable of multiplexed I/O.
11:27 Hunger joined
jonathan hi all 12:06
12:25 tetragon joined
dalek r35485 | jonathan++ | trunk/languages/perl6/src/parser: 12:31
: [rakudo] For now, when we add_type we will put it not in the current blocks's symbol table, but instead we will search outwards to find the first one representing a package and put it there. This deals with the final regression test failure that enabling the type registration brings up. It does mean lexical classes etc would be too visible - aside from the fact that we don't actually implement them yet, so we'll work out a solution to this when we get there
review: www.parrotvm.org/svn/parrot/revision?rev=35485
r35486 | jonathan++ | trunk/languages/perl6/src/parser: 12:32
: [rakudo] Enable the type registry. Causes no regressions in the spectests or sanity tests.
review: www.parrotvm.org/svn/parrot/revision?rev=35486
12:33 particle1 joined
jonathan waves at particle1 12:34
particle1: Type registry for Rakudo just went in. 12:35
lathos gets sidetracked from Parrot hacking by wedding list aggregator hacking. 12:36
dalek r35487 | jonathan++ | trunk/languages/perl6/src/parser: 13:06
: [rakudo] Parse parametric role declarations. (Note - trying to write one will just cause a runtime explosion; this is just the parsing!)
review: www.parrotvm.org/svn/parrot/revision?rev=35487
13:31 iblechbot joined 13:35 AndyA joined 13:38 riffraff joined 13:40 Lorn joined 13:41 kj joined 13:47 mberends joined 13:58 Whiteknight joined 14:05 gryphon joined 14:47 mj41_ joined
dalek r35488 | kjs++ | trunk/docs/pdds: 14:52
: [pdd19] Be more explicit; if you don't specify a :main flag, execution starts at the top.
review: www.parrotvm.org/svn/parrot/revision?rev=35488
14:54 particle joined, bacek joined
barney Has anybody else seen intermittent segfaults in cat t/op/arithmetics_27.pir ? 15:01
moritz I haven't 15:03
nopaste "barney" at 84.154.53.164 pasted "intermittent failures in t/op/arithmetics_27.pir under Linux" (83 lines) at nopaste.snit.ch/15293 15:08
15:10 ruoso joined
dalek r35489 | pmichaud++ | trunk/languages/perl6/docs: 15:11
: [rakudo]: spectest-progress.csv update: 282 files, 6233 passing, 0 failing
review: www.parrotvm.org/svn/parrot/revision?rev=35489
jonathan wonders if we can hit 7,000 by the end of the month 15:12
moritz (7000-6233)/(13-13) 15:14
purl: where are you when I need your service?
purl moritz: i don't know
moritz rakudo: (7000-6233)/(13-13)
polyglotbot OUTPUT[Divide by zeroā¤current instr.: 'infix:/' pc 19791 (src/builtins/op.pir:186)ā¤called from Sub '_block14' pc 109 (EVAL_15:48)ā¤called from Sub '!UNIT_START' pc 17124 (src/builtins/guts.pir:320)ā¤called from Sub 'parrot;PCT;HLLCompiler;eval' pc 950 (src/PCT/HLLCompiler.pir:527)ā¤called from Sub
..'parrot;PCT;HLLCompiler;evalfiles' pc 1275 (src/PCT...
moritz (7000-6233)/(31-13)
purl 42.6111111111111
moritz that's quite a lot 15:15
pmichaud trying to get to 7k by the end of jan?
moritz 43 nes passing tests per day
pmichaud that'd be a bit.
moritz implementing infix:<also> would get us quite some tests from rx.t 15:16
pmichaud do we still need .does ?
jonathan tires to remember what infix:also is 15:17
pmichaud jonathan: it's a fancy form of 'and'
sorry, 'all'
jonathan ah, ok
moritz it's "all" with ordered execution
pmichaud it's either trivial for us to implement, or it's not :-)
it may just be another Junction type for now.
moritz yes.
jonathan is working on parametric roles. 15:18
pmichaud jonathan: why does registering a new type have to be in a package block?
jonathan { class Foo::Bar { ... } } 15:19
{ Foo::Bar.new() }
Foo::Bar goes into the package.
If we just register it in the current block, it isn't visible in the next block down.
pmichaud hmmm. 15:20
jonathan (Actually doens't matter at all that I wrote Foo::Bar, I coulda written just Foo...)
pmichaud okay, I can live with that (for now :-)
jonathan Right. I know we'll have to re-visit it at some point.
pmichaud also, I'm not sure about the term fixups for package things... but we'll clean that up later also
I don't think we should be processing <name> in the term rule. 15:21
jonathan It is a little verbose.
lot of similar-ish code
pmichaud I'll put it on my list of things to look at. 15:22
jonathan Cool
pmichaud Other than those two items, everything looks fantastic. Great job.
jonathan Thanks. :-)
Thanks for all the input on it too.
pmichaud did you close/mark the relevant rt tickets? ;-)
jonathan No, saving that for later on.
pmichaud okay.
jonathan My brain is vaguely functional at the moment (not been sleeping great and bad stomach are a distraction), so trying to get a first cut of parametric roles in place. 15:23
pmichaud yeah, I've been fighting something for the last few days also. :-|
PerlJam it seems to be going around.
yesterday we had about 6 people absent from work due to illness of one form or another. 15:24
jonathan I think partly the cold here is wearing me down a bit.
PerlJam jonathan: what's the temp where you're at?
jonathan It hasn't been above freezing so far this year.
pmichaud 1 degC here :-)
jonathan -6C at the moment. 15:25
Which isn't *so* bad - if you're used to it.
PerlJam doesn't feel so bad about the ~7degC to ~ 18degC weather here now :)
jonathan Trouble is, I lived in south Spain last winter, and nasty as the UK weather can be, it's not this persistently cold. 15:26
15:31 Theory joined 15:54 rhr joined
dalek r35490 | bernhard++ | trunk/languages/pipp/docs (2 files): 15:55
: [Pipp] Update docs WRT namespaces.
review: www.parrotvm.org/svn/parrot/revision?rev=35490
donaldh any thoughts on how the rvar refactor might have broken NCI? 15:58
I'm trying to look into it, but frankly don't really know where to start.
suggestions for how to break down the problem would be greatly appreciated. 15:59
Infinoid donaldh: for lack of any better suggestions, maybe see if it happens in rvar/rvar2 branch checkouts, and if so, bisect it there? 16:03
donaldh Infinoid: fair suggestion. 16:04
I guess I'll trip over broken iterations some of the time, but it could help. 16:05
Infinoid got a test case?
16:05 alvar joined 16:19 allison joined
Casan lathos: hi, have you heard anything from Tim Bunce or anyone else about DBDI for parrot recently? 16:29
donaldh Infinoid: rt.perl.org/rt3/Ticket/Display.html?id=62244 16:31
Casan has anyone? :)
moritz depends on what you mean by "recently"
Casan moritz: last mention I have is around dec 6th 08, and then I heard it was not far from something which could actually be tested in some basic form. 16:32
moritz but that wasn't Tim's effort, afaict 16:33
donaldh Casan: there's some prototype SQLite3 DBDI code in ext/SQLite3
It breaks rakudo just now, see RT#62244 16:34
Casan ohh, an overview of who's actively working on this important feature could help the initiative.
pmichaud I just responded to #62244 a short while ago.
moritz probably "none" atm
donaldh is trying to use it. 16:36
Casan heh, ok its cooking slowly, ok I'll take a look at the trunks 16:37
donaldh pmichaud: reading it now.
pmichaud: so Rakudo will be given a Parrot Pointer PMC and will create an ObjectRef PMC, but will then pass the ObjectRef PMC into the next NCI call ? 16:44
pmichaud donaldh: I don't know what specifically happens for Pointer PMCs. 16:45
but yes, it tends to want to put things into types it understands.
that's why I need a small test case to work from :-)
we could potentially make Rakudo understand the Pointer PMC such that it doesn't create ObjectRef. 16:46
donaldh Okay, that could explain the SQLITE_MISUSE error I get.
The simple test case was mentioned in the ticket.
pmichaud I didn't see it -- checking again.
donaldh ext/SQLite3/t/test.p6 16:47
pmichaud oh, sorry, that doesn't fit my idea of "simple" :-)
donaldh k. 16:48
pmichaud yes, it's simple from the p6 end, but it means I have to go figure out everything that DBDI is doing.
donaldh I can try and remove the DBDI stuff.
It's admittedly more complex than it needs to be.
I'll see if I can produce a single file test case using the q:PIR syntax. 16:49
pmichaud that's fine. Even knowing that we're dealing with Pointer PMCs is a lot more than I knew before. 16:50
16:51 davidfetter joined
pmichaud where does the Pointer PMC end up? 16:51
(being stored)
looks like it's stored in $dbHandle; 16:52
donaldh In a data member.
Yes, you beat me to it.
pmichaud yes, I'm guessing the problem is that $dbHandle is being wrapped in an ObjectRef. 16:53
or, more precisely, a Perl6Scalar PMC
and so that Scalar is what gets passed to the other functions. 16:54
donaldh That makes sense. And NCI just gets a pointer from the Scalar. 16:56
pmichaud well, NCI actually gets the Scalar. I suspect that confuses it, as it's expecting a Pointer.
lathos Casan: My next stage is to rewrite DBDI using NCI.pm 16:58
Unfortunately it's all going on in dev.simon-cozens.org/6pan/browser/trunk right now.
pmichaud #parrotsketch in 90 17:00
lathos Doubly unfortunately, as usual, I am trying to do too many things at once, and thrashing.
17:01 desertmax joined
donaldh pmichaud: thanks, I'll poke around later. 17:01
pmichaud the short-term fix would probably be to get Rakudo to recognize Pointer PMCs as scalar objects. 17:02
although that wouldn't help in the case of Pointer-as-parameter, since those get wrapped in ObjectRef also.
the real answer may be to get Rakudo's dispatcher to unwrap any ObjectRefs from arguments prior to invoking non-Rakudo-generated subs. 17:04
(including self)
jonathan is probably the only one who knows how hard/easy that would be. 17:05
jonathan please easy please easy!
jonathan reads scrollback
pmichaud jonathan: I can summarize 17:06
jonathan It's OK, I get it.
pmichaud okay.
non-Rakudo-generated subs would tend to be "subs without !signature"
jonathan That check becomes cheap if we have .HLL because then we easily know if we're calling out of Rakudo.
Casan lathos: thanks for the update. It will also be an important piece of parrot when ready.
pmichaud or that
jonathan Oh, we can pick it on that too.
pmichaud :-)
jonathan Anyway, I need to deal with dispatchy things soon anyway. 17:07
lathos Right now my priority is Unicode.
jonathan To do auto-threading.
pmichaud we already have the !signature (or lack of one) from doing dispatch anyway. But I suspect HLL is a better approach.
well, maybe not.
anyway, would it be hard to deref the objectref arguments?
jonathan It's fragile if another language has it's own $!signature 17:08
Well, I guess really it's just detecting if the call is going "outside"
And if so doing deref on each of them.
pmichaud right
jonathan We need to add a layer of indirection to dispatch anyway for the auto-threading.
pmichaud "if language has its own signature" isn't so bad -- we should probably check that $!signature is a Rakudo Signature anyway.
(if we're worried about that case) 17:09
jonathan Right, but by the time we did that, we coulda probably more cheaply checked the HLL. But as a first cut, $!signature is fine.
pmichaud we can always do $!rakudo_signature instead of $!signature
jonathan True.
pmichaud I figure we'll have the signature anyway.
jonathan *nod*
pmichaud anyway, that's probably the better long-term approach to this.
jonathan I think that once we get the extra level of indirection into the dispatch, this won't be too hard. 17:10
I know where to put it for method dispatch.
(metaclass' dispatch method)
Not so sure on sub calls yet. 17:11
pmichaud well, we have to autothread subcalls also, so it ought to be similar.
jonathan Aye.
pmichaud anyway, I have to take Paula to lunch.
bbiaw.
purl somebody said bbiaw was something silly for Coke.
jonathan Well, I did have ponderings about a way to do it without having to add a level of indirection.
OK, after lunch. :-)
jonathan goes back to trying to get a first cut of parametric roles 17:12
17:14 particle1 joined
dalek r35491 | fperrad++ | trunk/languages/befunge (2 files): 17:15
: [befunge]
: - improve exception handling
review: www.parrotvm.org/svn/parrot/revision?rev=35491
17:18 tomyan left 17:25 wknight8111 joined 17:29 jhorwitz joined 17:36 andrew_ joined
dalek r35492 | bernhard++ | trunk/languages/pipp (4 files): 17:40
: [Pipp] Support only the bracketed syntax for namespaces.
: Update the tests. 17:41
: Remove the test for case insensitive namespaces,
: as namespaces are case sensitive.
review: www.parrotvm.org/svn/parrot/revision?rev=35492
17:46 cognominal joined
particle1 coke: i have been out of work since 28nov2008, so no, i'm not telecommuting anymore 17:47
Whiteknight particle1: that's lousy, I'm sorry to hear that. 17:49
PerlJam particle1: did you apply for this job? news.yahoo.com/s/afp/20090113/od_af...0113032028 17:50
shorten PerlJam's url is at xrl.us/becerh
jhorwitz thinks that job would actually suit particle
17:53 chromatic joined
kj Casan: I'll ask Tim tomorrow about the status, I'll be meeting him then 17:54
particle1 kj: give him my greetings 17:56
kj particle1: will do. 17:58
chromatic particle1, any news on the Bugs? 17:59
18:01 masak joined
kj I preposted my #ps report, have to leave now :-( 18:01
particle1 chromatic: no news yet, but they just went through ces, and i'm sure had a busy time there. i'll check into it 18:03
chromatic I have a G1 coming soon to poke at.
18:04 pdcawley joined
particle dev? 18:04
purl rumour has it dev is leaving
masak oh noes, dev is leaving! 18:05
chromatic Yes, that's the plan.
Whiteknight irclog? 18:07
purl i guess irclog is irclog.perlgeek.de/parrot/today or see also: infrared clogs
jonathan infrared clogs? 18:08
purl infrared clogs are like ruby slippers, but more so
Coke_away particle: did I already ask you if you knew cold fusion? =-) 18:09
(though we just hired a FT onshore for that.)
jq what is the official home of STD.pm? 18:10
chromatic That's cruel and unusual employment.
PerlJam jq: the pugs repo
jq: did you want the exact url? 18:11
jq no, that's enough information
thanks
coke chromatic: pays the bills. sort of. well, some of them. :|
18:12 gryphon joined
coke is forwarded a link to a link to the request for grant proposals from TPF, with a hint I might want to apply. 18:12
18:14 rurban joined
coke responds with a kind word and a link to the list of grant committe members. 18:15
PerlJam coke: maybe it was a suggestion that you resign? ;) 18:17
coke :P 18:18
18:19 hercynium joined
particle coke: i'm sure you could help me doctor my resume to look like i know cf, but doesn't that hurt to do every day? 18:20
nopaste "particle" at 76.121.106.245 pasted "chromatic: non-functional stab at inf/nan" (68 lines) at nopaste.snit.ch/15294 18:22
particle chromatic: i spent 10m on this a few days ago, but ran into problems with headerizer and lost track of what i was doing 18:23
i can pick it up later, but thought you should have a look
18:23 riffraff joined
chromatic Looking. 18:24
coke particle: given I used to get paid to write /java/, this is a step up.
riffraff hi
coke (though I'm tempted to use groovy instead.)
chromatic I once made money maintaining an ASP/VB app.
coke see, we're all whores.
I just happen to have a decent health plan and gym membership. 18:25
chromatic No, I mean on one day.
particle, Andy's likely to suggest you use the STREQ macro.
coke dillitente.
particle STREQ++
chromatic If divide_by_zero() is portable across compilers, I think it's a good start.
#ps in 5 18:26
particle i am getting a warning about STRING * to const char *
lathos class Foo { ... }; # If this meant to work? 18:27
coke particle: you want CONST_STRING
lathos rakudo: class Foo { ... };
polyglotbot OUTPUT[No exception handler and no messageā¤current instr.: 'return' pc 15424 (src/builtins/control.pir:39)ā¤called from Sub 'fail' pc 15496 (src/builtins/control.pir:71)ā¤called from Sub '...' pc 15712 (src/builtins/control.pir:177)ā¤called from Sub 'parrot;Foo;_block20' pc 93 (EVAL_19:49)ā¤called from
..Sub 'parrot;PCT;HLLCompiler;evalpmc' pc 888 (sr...
coke you can't just cast a c string into a parrot string.
particle is ... implemented?
coke chromatic: can you run #ps today? I have an actual physical coworker who may interrupt me.
particle coke, interrupted. 18:28
there's a scary image.
coke i'm cuter than winona.
chromatic I have several actual physical coworkers who may interrupt me, but I'll do my best. 18:29
particle, I keep planning to implement ... and compare that to my work implementing it in Perl 5.
I just haven't done it yet.
#ps in 1
jonathan lathos: Don't think we handle subts yet.
lathos sub Foo { ... }; works 18:31
Um. By which I mean it doesn't cause a syntax error.
coke particle: maybe you could get an grant to work on ack plugins for TPF. 18:38
particle coke: any ideas for plugins? i could certainly do that. 18:41
coke did andy ever get the first one done?
jonathan lathos: I don't think it's a syntax error. 18:42
It's because it's failing within the class body by running ...
Rather than recognizing it as a stub.
lathos Oh, I see. That makes sense.
coke zip/[jwc].ar support would be a good first one if it's not already done.
dalek r35493 | bernhard++ | trunk/languages/pipp/docs: 18:43
: [codingstd] shorten long line
review: www.parrotvm.org/svn/parrot/revision?rev=35493
coke chromatic, Whiteknight: I'm willing to help out with the GC stuff (perhaps with scut work, since IANACP) 18:44
Whiteknight any help always appreciated 18:45
pmichaud I wonder if we should put a return handler on class bodies. 18:47
Then < ... > would almost dtrt. 18:48
jonathan Hmm, yes.
That could work.
pmichaud but we'll probably need to do something special anyway, as we really don't want class Foo { ... } to declare an empty class. 18:49
(or worse, overwrite an existing declaration)
jonathan Aye.
chromatic Do we support ... now? I kinda wanted to add that. 18:50
particle class Foo {...}; class Foo is also { method bar {1} }; # should work
purl okay, particle.
particle as well as 'is instead'
dalek r35494 | bernhard++ | trunk/languages/ecmascript (2 files):
: [codingstd] satisfy trailing_space.t
review: www.parrotvm.org/svn/parrot/revision?rev=35494
particle should be made to work, that is.
coke class Foo {...}; class Foo? 18:52
(bad purl, no donut.)
pmichaud perl6: sub foo { ... }; foo(); 18:53
polyglotbot RESULT[undef]
pmichaud looks like it's "supported" but "incorrect".
perl6: sub foo { ... }; say foo(); 18:54
polyglotbot OUTPUT[Attempt to execute stub code (...)ā¤ā¤]
pmichaud looks like it's support and "correct" :-)
*supported
jonathan pmichaud: I thought it was meant to return a failure object?
That explodes if you try to use it...
pmichaud it did return a failure object
jonathan Right. And exploded when you tried to use it. :-)
pmichaud perl6: sub foo { ... }; say foo().PARROT;
polyglotbot OUTPUT[Failureā¤]
pmichaud right 18:55
so it looks correct-ish to me.
jonathan So it's correct or "correct"? ;-)
chromatic That looks correct.
pmichaud (1) I'm not sure that's the correct failure message, (2) it doesn't explain what lathos' example was doing earlier.
in the case of subs -- yes, it appears to be working. 18:56
jonathan pmichaud: It does a "fail" which throws a return exception, I think.
pmichaud also, I wonder about:
jonathan But the class doesn't have a handler, as you pointed out.
chromatic The failure message is wrong.
pmichaud perl6: sub foo { if 1 { ... }; say 'ok'; }; say foo();
polyglotbot OUTPUT[Attempt to execute stub code (...)ā¤ā¤]
pmichaud should the stub code exit the sub? Or just the block? 18:57
jonathan chromatic: That's a bikeshed. 18:58
pmichaud: I think it's meant to do what fail() does.
chromatic jonathan, I thought one of the Synopses specified the error message. If not, then it's for sure a bikeshed and I don't have a brush in this painting party. 18:59
pmichaud jonathan: so do I, but it's not entirely the same as fail(). I was just curious, at any rate.
jonathan chromatic: I haven't seen such a synopses.
chromatic: Though I would be very happy to see it exist and follow it. :-)
Whiteknight chromatic, do you have any information about context-threaded runcores that an interested coder could peruse? 19:00
dalek r35495 | simon++ | branches/strings/docs/pdds:
: Removed Parrot_string_new, since it's exactly equivalent to new_from_cstring.
review: www.parrotvm.org/svn/parrot/revision?rev=35495
coke lathos: should we have kept the other one? 19:01
(to encourage sane naming conventions?)
(RT#35495)
chromatic Whiteknight, citeseerx.ist.psu.edu/viewdoc/summa....1.59.1271 19:02
shorten chromatic's url is at xrl.us/bebjy9
chromatic I found some other papers elsewhere. 19:03
lathos coke: I think the whole string creation stuff is very badly specced for two reasons. First, it talks about "default" encoding and character set, and I believe this is impossible. Second, pretty much all of them do the same job in initializing data from a C string. So currently I'm just slashing out API functions until the API is sane again.
Whiteknight chromatic: thanks
19:04 pdcawley joined
chromatic lathos, I had the same thought as coke, but this approach sounds sane to me. 19:04
lathos Also note I'm doing all this in my own little sandbox branch.
coke oh, yah, Just kibbitizing.
19:05 allison joined
lathos new_init and new_from_cstring are the same too. 19:06
Maybe that's the one to keep.
(new_init I mean)
19:06 elmex joined
dalek r35496 | simon++ | branches/strings/docs/pdds: 19:09
: Hey, all these functions do the same thing. Let's only have one of them.
review: www.parrotvm.org/svn/parrot/revision?rev=35496
chromatic lathos, we might need to throw some of those function names into DEPRECATED.pod, but only if they had PARROT_EXPORT. 19:10
allison chromatic: most of what lathos is doing now won't be merged until after 1.0 19:11
lathos Yeah, uh, there's going to be rather a *lot* of dust to settle once I've finished changing EVERYTHING THAT PARROT DOES THAT TOUCHES STRINGS.
DEPRECATED.pod is kind of the least of my worries right now. :)
chromatic Strings is post-1.0? 19:12
allison chromatic: but once I've reviewed the changes, we'll enter them in DEPRECATED.pod
coke i hate trac.
Trac detected an internal error:
OperationalError: database is locked
purl i already had it that way, coke.
allison chromatic: some of strings is pre 1.0 (I'm working on that part), some is post 1.0
chromatic I like the idea of removing duplicate functions soonerish though. 19:13
lathos I haven't checked if these functions have been implemented yet or just specced.
dalek r35497 | Whiteknight++ | branches (50 files):
: [removing_stm] updating branch to trunk from r35463
review: www.parrotvm.org/svn/parrot/revision?rev=35497
allison chromatic: yeah, I'll cherry-pick from lathos' changes and pull some of them into the first refactor 19:14
Whiteknight allison, any problems with me ripping out STM right now?
lathos Oh, that reminds me, how *do* I update a branch from a trunk? 19:15
particle lathos: svn 1.5 or earlier?
lathos 1.4.4 at the moment, but that can be fixed...
allison Whiteknight: in a branch, sure
particle lathos: the branch had to be created with 1.5 for the new method 19:16
allison lathos: it's 'svn merge'
particle see docs/project/committer_guide.pod
lathos Ah, bummer.
pmichaud pm's approach: create a new branch from trunk, merge old branch into new branch.
Whiteknight allison, it's already ripped out in th branch. I want to merge the branch into trunk
particle meeting &
allison svn merge -r<old rev>:HEAD svn.perl.org/parrot/trunk/
(run from within your branch root directory) 19:17
Whiteknight: passes all tests and all language tests?
Whiteknight: tested on mac/win/lin?
Whiteknight oh, i haven't run all the languages tests yet. nevermind
coke (languages used stm?)
chromatic Can't believe that they did.
lathos Urgh.
% svn log -v --stop-on-copy svn.perl.org/parrot/branches/parrot-strings
svn: REPORT request failed on '/parrot/!svn/bc/35497/branches/parrot-strings'
svn: '/parrot/!svn/bc/35497/branches/parrot-strings' path not found
coke i think a simple 'ack -i stm' will assuage your concerns.
Whiteknight i'm pretty convinced that nobody uses STM, but I'll test it anway 19:18
allison Coke: I'm not sure, but there's one sure way to find out
lathos Oh, I'm an idiot.
chromatic ack STM languages/
coke i already had it that way, lathos.
allison chromatic: aye, but ripping out subsystems has occasionally had unexpected side-effects, especially in the older code 19:19
coke given that many languages are still not fixed from things ripped out since 0.8.2, it might not help. 19:20
but certainly doesn't hurt to check if you have the time.
pmichaud it hurts a lot less to check and then merge instead of merge and wish you had checked first. 19:21
chromatic allison, RT #46819. 19:24
allison chromatic: looking... 19:28
coke wonders why particle just sent that email. 19:29
coke ahs, as he sees something buried in the forward. 19:30
allison chromatic RT #46819 patch looks good, go ahead and apply it and add the optional warnings for intify and numify 19:33
chromatic Does ./parrot -j t/op/trans_13.pasm pass for anyone else on 32-bit x86 Linux?
I don't *think* I broke it.
dalek r35498 | Whiteknight++ | branches/removing_stm: 19:34
: [removing_stm] since I'm monkeying with it anyway, update PBC_COMPAT to remove a duplicate line TT#169
review: www.parrotvm.org/svn/parrot/revision?rev=35498
r35499 | bernhard++ | trunk/languages/pipp/t/php:
: [Pipp] check the namespace tests against PHP 5.3 and fix them
: Variables are not namespaced, so don't test that.
review: www.parrotvm.org/svn/parrot/revision?rev=35499
coke commiters, please comment on tts 106, 107, 108, 109, 120
Whiteknight we got any mac users around here willing to test the removing_stm branch? 19:35
coke Whiteknight: sure.
barney Whiteknight++ , jimmy++
cotto chromatic, appears to work for me
19:36 particle1 joined
chromatic Hmm. Somehow I've broken atan then. 19:37
Whiteknight atan2 has been broken
chromatic This is normal atan, not the sequel.
Whiteknight oh, is atan2 still broken for you then?
chromatic Only ./parrot t/op/trans_13.pasm fails for me. 19:38
I'm trying to remove a magic number related to cpu_ret.
coke pmichaud: the only email related ticket I see is regarding perl6-internals. 19:39
19:40 ask_ joined
pmichaud coke: I think that's the only one at present. 19:40
I simply know that the question has arisen on channel a few times.
and I'm not sure where to direct things. 19:41
dalek r35500 | coke++ | trunk: 19:42
: Mention TT# 6
review: www.parrotvm.org/svn/parrot/revision?rev=35500
19:46 Andy joined
nopaste "tene" at 67.137.148.182 pasted "rakudo HLL patch for pmichaud++ (has a few spectest failures)" (364 lines) at nopaste.snit.ch/15295 19:46
dalek r35501 | bernhard++ | trunk/languages/pipp/docs: 19:47
: [Pipp] Delete some obsolete SEE ALSO links
review: www.parrotvm.org/svn/parrot/revision?rev=35501 19:48
pmichaud Tene: cool! Okay if I take a crack at it later tonight? 19:49
lathos Is this something I've done, or is it everywher? t/library/mime_base64.t 1 256 550 1100 200.00% 1-550 19:54
chromatic Did you change PBC compatibility?
lathos invoke() not implemented in class 'Integer' 19:55
current instr.: 'parrot;JSON;value' pc 3968 (JSON/grammar.pir:1470)
chromatic That looks like languages/json needs rebuilding.
lathos Still no joy, I'll try a make clean, make test.
chromatic I think you can get by with make clean in languages/json and rebuild. 19:56
lathos Better safe, etc.
chromatic I'm lazy.
19:57 rurban_ joined
Tene pmichaud: whenever is fine. 19:57
dalek r35502 | chromatic++ | trunk/src/jit/i386 (2 files): 20:01
: [JIT] Turned a magic number representing the cpu_ret opcode into
: PARROT_OP_cpu_ret, which is somewhat more clear.
review: www.parrotvm.org/svn/parrot/revision?rev=35502
riffraff shall I report compiler warnings? if so, where? 20:05
chromatic mailing list or nopaste, please. 20:06
coke Whiteknight: passes all tests on os x/85 20:08
+1
purl 1
barney lathos: t/library/mime_base64.t looks sane here
coke pmichaud: that's a perl.org mailing list, so i sent it to the admins.
(@ perl.org) 20:09
closd a few tickets that appear to have resolved spontaneously in re: website/mailing also.
riffraff ok
Whiteknight thanks coke
lathos barney: Yep, got it working.
Whiteknight so this branch is passing all tests on mac/win/lin now, and all language tests I've tried 20:10
I think tht's good enough, unless there are any other objections
allison: Good for a merge then?
allison Whiteknight: thumbs up!
purl thumbs up is at www.friedmanarchives.com/China/Web/...%20dpi.jpg
allison heh, thanks purl :)
coke shorten that 20:12
purl That URL is at xrl.us/becfak [parrotvm.org]
coke ... stupid purl.
Whiteknight allison, have you take a look at the pcc_invoke_convert branch yet? All the function calls I tried changing over caused weird test errors or segfaults 20:13
20:14 donaldh joined
chromatic Looks like a no. Sorry, guy. 20:17
dalek r35503 | simon++ | branches/strings (268 files): 20:27
: Merging in from trunk.
review: www.parrotvm.org/svn/parrot/revision?rev=35503
r35504 | Whiteknight++ | (53 files):
: [removing_stm] merging the removing_stm branch into trunk from r35498. This removes the STM subsystem from Parrot.
review: www.parrotvm.org/svn/parrot/revision?rev=35504
jhorwitz hm, mod_parrot errors out with " '' is not a valid vtable function name. " when loading perl6.pbc. trying to reproduce outside of mod_parrot without much luck. from gdb it looks like it's unpacking constant PMCs. any ideas? 20:33
GeJ Good morning everyone. 20:34
thanks for the trust. 20:35
jhorwitz good afternoon
chromatic That sounds like someone's returning a CONST_STRING which someone modifies.
Whiteknight GeJ, you get your CLA mailed yet?
coke seen obra/ 20:36
purl I haven't seen 'obra/', coke
coke seen obra?
purl obra was last seen on #p5p 1 days, 15 minutes and 16 seconds ago, saying: nothingmuch: that's from git-as-of-a-few-months-ago and deals with repo bogosity. but it taught me what I needed to know about migrating branches to local from the svn "remote" [Jan 12 20:21:11 2009]
GeJ Whiteknight: downloaded and filled. I was wondering if emailing it would be faster. If not, I'll snail-mail it this morning. 20:37
coke email should be ok.
parrot cla?
cla?
purl somebody said cla was Contributor License Agreement or www.perlfoundation.org/contributor_..._agreement or www.parrot.org/foundation/legal
jhorwitz chromatic: at the very top of the trace, it looks like it's trying to store a nameless sub in a namespace (and failing miserably)
i can paste if you'd like to see it 20:38
chromatic I'll take a look, but my brainpower is low at the moment... something to do with blood sugar.
coke duh-duh-duh-duh-diabeetus 20:39
jhorwitz sends sugar packets
chromatic More like the lack of lunch. 20:40
Whiteknight diebeetus sounds scary. We should call it live-beetus
Tene paints a pentagram on the floor with a mixture of equal parts sugar and blood. 20:41
nopaste "jhorwitz" at 96.245.16.64 pasted "chromatic: tasty backtrace" (80 lines) at nopaste.snit.ch/15296 20:42
chromatic What are the contents of key and value in stack frame 2? 20:45
jhorwitz key is a NULL string (strstart=0x0). value is a Sub. 20:46
dalek r35505 | Whiteknight++ | branches: 20:51
: deleting the removing_stm branch because it's already been merged into trunk and is no longer needed
review: www.parrotvm.org/svn/parrot/revision?rev=35505
20:51 allison joined
chromatic Hm, so the Sub's ns_entry_name is a NULL string. 20:52
Something froze a Sub with no ns_entry_name.
That's okay, but it can't be a vtable override, because we don't know which vtable entry it overrides. 20:53
Whiteknight does that happen for an :anon sub?
jhorwitz and why does it happen under mod_parrot but not from the command line....
all interesting questions. :) 20:54
Whiteknight and why do bad things happen to good people?
chromatic An :anon sub should probably store its vtable override.
... and that does get frozen. 20:55
I don't see anything immediately applicable. What if you rebuild the PBC? 20:57
jhorwitz done that many times.
20:57 Cantacuzenus joined
chromatic That was my next best guess; invalid PBC. 20:59
jhorwitz nah, the perl6.pbc is good.
passes tests, etc.
runs fine on the command line
chromatic Is that the PBC you're trying to load from mod_parrot? 21:00
jhorwitz mod_parrot loads a bootstrap PBC for mod_perl6. that contains a :load sub that does a load_bytecode "languages/perl6/perl6.pbc" 21:01
so yes, in the end, that's the same file
chromatic I'm stumped for now.
jhorwitz ok. i'll keep plugging away. 21:05
21:05 elmex joined
chromatic I'll skim the code for misbehaving const_strings. 21:06
Hm, nothing in Rakudo that looks wrong. 21:08
pmichaud is it a vtable sub in Rakudo itself? 21:15
21:16 pdcawley joined
jhorwitz it happens while loading perl6.pbc, so i'm inclined to say yes. but since it has no name it's tough to tell exactly what it is. 21:17
pmichaud many of rakudo's :vtable subs have names of ''
jonathan pmichaud: You were right as usual. I too hate enum_declarator now. :-)
pmichaud jonathan: :-)
jhorwitz pmichaud: yes, i saw that. but it works on the command line, so something is screwy. 21:18
pmichaud jhorwitz: I wonder if you're running into the same bug that Tene++ is seeing
jhorwitz ah?
purl rumour has it ah is much more useful
pmichaud no, I guess that would be different.
jhorwitz purl forget ah
purl jhorwitz: I forgot ah
pmichaud but Tene runs into issues with running load_bytecode 'perl6.pbc' because of the .load_library directive.
jhorwitz is there a ticket for that? 21:19
on the off chance that it's related
pmichaud yes -- looking.
(trac.parrot.org is becoming as slow as rt :-( ) 21:20
TT #150
jhorwitz danke
pmichaud filename=0x95c1f98 "/home/jeff/build/mod_parrot/trunk/lib/ModParrot/HLL/perl6.pbc") at src/packfile.c:4372 21:22
is that the correct path for the .pbc ?
jhorwitz that's the mod_perl6 bootstrap file
pmichaud okay.
jhorwitz it has a :load sub that loads rakudo's perl6.pbc
pmichaud what's the value of *name in Parrot_Class_add_vtable_override 21:25
?
lathos *boggle*
svn: Commit failed (details follow):
svn: MERGE of '/parrot/branches/strings': 200 OK (svn.perl.org)
jonathan lathos: It happens now and then. :-S
Generally if you svn up it sorts itself out. 21:26
coke internets are slow today.
pmichaud lathos: we've never been able to track down why that happens (on svn.perl.org)
or I should say that Robert/Ask haven't been able to do it.
the rest of us are just clueless :-)
lathos Now I have no idea what state my working copy is in. :)
Ah, looks like it's OK. 21:27
jonathan lathos: svn up can help.
coke svn migration coming up soon.
jonathan lathos: That normally got it OK for me.
lathos Yep, svn up cleared things.
jonathan ...apart from one time when I did it with svn 1.5 and it trashed my working copy...
chromatic shhhhhh!
jonathan :-)
chromatic Don't spoil the surprise.
jonathan lol 21:28
jonathan does make spectest, hoping to be down to relatively few failures
21:29 purl joined
jhorwitz pmichaud: sorry, got distracted.... 21:29
pmichaud jhorwitz: no problem.
I think chromatic's question supercedes mine anyway.
jhorwitz name is a STRING with a NULL strstart 21:30
pmichaud yeah, that's bogus.
chromatic What's its length? 21:32
jhorwitz 0
strlen = 0
pmichaud why is it using ns_entry_name in the first place?
chromatic In theory, it's a vtable override. 21:33
A NULL strstart seems okay with strlen and buflen of 0.
jhorwitz ah, but the error is an assertion failure
pmichaud sure, but what about .sub :anon :vtable('get_integer')
?
do we store that with ns_entry of 'get_integer'?
jhorwitz oh wait, no it isn't -- never mind my last comment.
chromatic Should, yes. 21:34
dalek r35506 | jonathan++ | trunk/languages/perl6/src/pmc:
: [rakudo] Get the signature in Perl6MultiSub straight from the property, rathre than via a method. Bit faster, and solves an ordering issue in the forthcoming role changes.
pmichaud but it shouldn't be entered in the namespace (because of the :anon)
dalek review: www.parrotvm.org/svn/parrot/revision?rev=35506
jonathan Gah, svn hung for me too!
pmichaud jonathan: once it starts hanging for one person, it often starts hanging for many. 21:35
I thought that ns_entry was only for subs that were to be stored in the namespace itself. 21:36
jonathan The best bit is it doesn't respond to ^C...
chromatic set_pmc_keyed_str on the NameSpace relies on ns_entry for the vtable entry. 21:39
Which seems a little wrong.
pmichaud Agreed.
chromatic See src/pmc/namespace.pmc:75
coke jonathan: ^\\ ? ^/ ? 21:41
jonathan coke: ? 21:46
dalek r35507 | jonathan++ | trunk/languages/perl6 (12 files):
: [rakudo] Initial refactor of roles to work towards parametric role support. Now we have a Perl6Role object installed in the namespace. It knows how to produce a Parrot-level role when given parameters, by doing a multi-dispatch on them. There's a bunch of nasty things to make enums keep working (that code is over-ripe for a big refactor soon), plus many comments of things that remain to be done. Happily, discounting enum stuff which will shrink a lot soon,
review: www.parrotvm.org/svn/parrot/revision?rev=35507
21:47 Tene joined
jonathan OK. That was a painful patch to produce. :-) 21:47
Tomorrow is the fun stuff: actually building on it so we can do parametric roles rather than just declaring them. :-) 21:48
moritz how about writing some tests first? ;-) 21:49
jonathan moritz: Believe it or not, that patch doesn't actually really add any new functionality. :-)
moritz we have one file with tests for parameteric roles, but that test seems complicated, and needs review really badly
jonathan moritz: But I'll certainly be writing a bunch of tests tomorrow when I get to adding functionality on top of this.
Where is it? 21:50
purl it's it!
moritz jonathan: I meant before tomorrows patch
jonathan In t/spec?
moritz t/spec/S12-role/parameterized.t
chromatic That should get us a few passing tests....
moritz jonathan: I happened to have moved it over christmas ;-)
jonathan moritz: Yes, it wants review. 21:52
moritz: And some simpler cases too. ;-)
moritz jonathan: that's what I thought, yes
jonathan moritz: But thanks for litting me know it's there.
*letting 21:53
moritz jonathan: but since I don't understand them (yet), I couldn't do very much with reasonable effort
donaldh pmichaud: I've constructed an 'interesting' test case for the NCI problems I'm having.
jonathan moritz: That's OK, I'm still wroking on the understanding part too. ;-)
nopaste "donaldh" at 213.123.171.12 pasted "test case for pmichaud++" (20 lines) at nopaste.snit.ch/15298
21:56 particle joined
jhorwitz chromatic/pmichaud: found the offending code in mod_parrot, but not sure why it causes a problem 21:59
donaldh say sqlite3_libversion().PARROT; # String
say SQLite3::libversion().PARROT; # List
jhorwitz the class in question is P6protoobject
mod_parrot loads P6object.pbc so it can register some classes. when i comment that block out, the error goes away (but breaks other things obviously) 22:00
chromatic Now we need to trace that down to what part of P6object calls which parts of Parrot in C.... 22:01
jhorwitz tries again to write a test case 22:04
yay
22:05 Whiteknight joined
jhorwitz 4 line test case! 22:05
moritz do they fit on 80 columns each? ;-) 22:06
jhorwitz indeed!
nopaste "jhorwitz" at 96.245.16.64 pasted "vtable override test case" (4 lines) at nopaste.snit.ch/15300 22:07
jhorwitz i'll open a ticket too
22:07 kj joined
jhorwitz hm, parrot ticket or rakudo ticket? 22:08
donaldh jhorwitz: is perl6_group needing loaded?
22:09 Theory joined
jhorwitz donaldh: for P6object? never had a problem with it before, but i could try forcing it. 22:09
donaldh I think this was mentioned a couple of days ago. load_bytecode per6.pbc failing to load perl6_group 22:10
jhorwitz nope, not the issue
perl6.pbc itself works fine
coke we have leo sign!
jhorwitz but i just tried throwing a loadlib in there for good measure, and it still broke
dalek r35508 | kjs++ | trunk/compilers/pirc/src (4 files):
: [pirc] fix implementation of find_outer_sub. This is kind of a hacky solution, but that's fine for now, Let's Just Make This Work. (and it does!)
review: www.parrotvm.org/svn/parrot/revision?rev=35508
chromatic I'll work to track that down. 22:11
jhorwitz chromatic: i guess it makes sense to open a parrot ticket for this rather than rakudo 22:12
chromatic jhorwitz, you might add debugging printfs to Sub's thaw to see if something goes wrong somewhere.
jhorwitz will try 22:14
chromatic I need to catch an airplane in a bit.
The first thing I'd check is that we're not somehow reusing a STRING (especially a CONST_STRING).
Whiteknight I don't think i've ever seen a live communication from Leo
chromatic That tends to make it wonky. 22:15
jhorwitz afk & 22:16
dalek r35509 | kjs++ | trunk/compilers/pirc/src: 22:24
: [pirc] complete create_lexinfo() function.
review: www.parrotvm.org/svn/parrot/revision?rev=35509
r35510 | jkeenan++ | trunk/config/gen/makefiles: 22:31
: Correction in 'help' message re 'make rpms'. Cf.: trac.parrot.org/parrot/ticket/171. Patch contributed by Gerd Pokorra.
review: www.parrotvm.org/svn/parrot/revision?rev=35510
r35511 | simon++ | branches/strings (4 files): 22:36
: This is where I'm up to in speccing out the Parrot strings API in Perl 6.
review: www.parrotvm.org/svn/parrot/revision?rev=35511
r35512 | kjs++ | trunk/compilers/pirc/src: 22:45
: [pirc] fix a bit in create_lexinfo. Unfortunately, doesn't work still. How are .lex statements handled?
review: www.parrotvm.org/svn/parrot/revision?rev=35512
22:50 HG` joined
jonathan lathos: Are you running the strings code under Rakudo? 22:50
Or aiming to etc?
lathos I am, yes.
jonathan Cool! :-)
I hope. ;-)
lathos Yes, it works. 22:51
jonathan Nice.
dalek r35513 | kjs++ | trunk/compilers/pirc/t: 22:59
: [pirc] some tests.
review: www.parrotvm.org/svn/parrot/revision?rev=35513
23:04 TiMBuS joined
dalek r35514 | kjs++ | trunk/compilers/pirc/t: 23:20
: [pirc] add some tests for indexing arrays and hashes.
review: www.parrotvm.org/svn/parrot/revision?rev=35514
kj jonathan: ping 23:37
jonathan kj: pong 23:42
kj hi, do you happen to know anything about how .lexicals are stored? 23:43
eh, in bytecode 23:44
jonathan Something. :-)
So .lex 'foo', <register>
kj yes, I invoke a method declare_lex_preg on lexinfo PMC, with "foo", and <register>
jonathan There's two bits. A LexPad and a LexInfo. 23:45
LexPad exists at compile time, LexPad at runtime.
So a LexPad PMC needs to be created and frozen. 23:46
And referred to from the Sub PMC.
IIRc.
kj LexInfo exists at compile time? or a Lexpad?
jonathan Erm, one or the other. ;-)
jonathan checks the PDD
kj ok, I tihnk its' the lexinfo
jonathan ah
LexInfo PMC is compile time
Note also :lex, which declares that a sub will need a LexPad but doesn't know about it's lexicals at compile time. 23:47
kj yes, I found that 23:48
jonathan But you still should, according to the PDD anyway, create a LexInfo
kj yes, that's all taken care of. I just wonder whether invoking declare_lex_preg method on the LexInfo sub for each lexical is enough 23:49
jonathan Is it not working?
I think the name => register mapping is the main thing, though. 23:50
kj ok, well that's happening, so it must be something else then
thanks anyway :-) 23:53
23:56 Limbic_Region joined

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!