parrot.org/ - clean up those smolders for the release!
Set by moderator on 20 October 2008.
00:00 rdice joined 00:09 AndyA joined 00:29 Theory joined
dalek r32353 | Whiteknight++ | calling_conventions: 00:41
: [calling_conventions] renamed function Parrot_pcc_invoke_helper to Parrot_pcc_invoke_from_sig_object on suggestion from Allison++
diff: www.parrotvm.org/svn/parrot/revision?rev=32353
01:00 magnachef joined
dalek r32354 | Whiteknight++ | calling_conventions: 01:09
: [calling_conventions] killed function Parrot_pcc_invoke_sub_from_sig_object in favor of the new (and identical) Parrot_pcc_invoke_from_sig_object.
diff: www.parrotvm.org/svn/parrot/revision?rev=32354
r32355 | coke++ | trunk: 01:20
: [t] Fix skip count so test passes.
diff: www.parrotvm.org/svn/parrot/revision?rev=32355
01:28 petdance joined
dalek r32356 | coke++ | trunk: 01:47
: Remove .pragma n_operators, closing out RT #57438.
: This leaves no .pragma remaining, so remove .pragma support.
diff: www.parrotvm.org/svn/parrot/revision?rev=32356
Coke ... sure, now that robrt closed out my svn ticket, NOW i start getting commit issues again. 01:50
Whiteknight: I stole your .pragma ticket. 01:56
Whiteknight presses charges 01:57
that's fine, I probably didn't have time for it yet
chromatic Whiteknight: research.microsoft.com/~simonpj/pap.../index.htm 01:59
Whiteknight I think I may have read this one already, but I'll take a double look 02:00
chromatic I saw it in a discussion of GHC 6.10. 02:01
02:17 kid51 joined 02:32 davidfetter joined 02:49 Ademan joined 03:19 magnachef__ joined
szabgab @messages 03:33
@messages[0] 03:34
@messages[0].say
@messages.pop
Tene try /msg purl messages
szabgab thanks, I thought it is perl6 code.. 03:35
Coke: thanks, File-ShareDir-PAR-0.02 also failed for me, I should investigate 03:37
have you forced it?
03:43 sjansen joined 03:48 Psyche^ joined
cotto win! 04:05
04:25 japhb joined 04:28 magnachef joined
Coke (forced) nope. 04:32
04:32 nopaste joined 04:44 TonyC joined
dalek r32357 | coke++ | trunk: 05:19
: Remove some [DEPRECATED] (and unused) mmd functions.
diff: www.parrotvm.org/svn/parrot/revision?rev=32357
chromatic Sweet.
Coke ? 05:26
chromatic Removing code is good. 05:28
Coke mmm 05:29
Those were the easy ones. 05:30
chromatic And now they're gone. 05:35
Big win. 05:36
purl big win is when you pretend that http isn't stateless
06:14 sjansen joined
Coke tries to remove find_global and has a ton of tests fail. 06:19
chromatic HEY I was going to do that. 06:20
Patch stealer!
Coke I'm in the middle of an EPIC FAIL.
chromatic Really? Mine's working. 06:27
Coke chromatic++ 06:28
I thought it was going well, but I tried temporarily removing the opcodes, doing a build, updating everythign I found. I then reverted the opcode change, and did a make test. "boom". 06:29
(so just updating runtime/ was not working out for me.)
if you have something working, awesome.
06:39 Zaba joined
Coke -> 06:41
07:24 japhb joined
sjn gets a "Null PMC access in get_string()" error when running latest perl6 "Hello World" 07:26
moritz sjn: define "latest". svn? or latest release?
sjn svn 07:27
At revision 32357.
moritz sjn: does it work with ../../parrot perl6.pbc -e 'say "hellow world"' 07:28
sjn same error 07:29
pastebin? 07:30
purl rumour has it pastebin is see nopaste. or see shadowpaste
07:30 Zaba_ joined
sjn nopaste? 07:30
clunker3 pasta.test-smoke.org/ or paste.husk.org/ or nopaste.snit.ch:8001/ or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/
purl nopaste is, like, at nopaste.snit.ch/ (ask TonyC for new channels) or rafb.net/paste or poundperl.pastebin.com/ or paste.scsys.co.uk/ or App::Nopaste or tools/dev/nopaste.pl or at www.extpaste.com/
moritz weird... did you try a 'make realclean' first?
nopaste "sjn" at 84.215.55.4 pasted "Hello World fails" (7 lines) at nopaste.snit.ch/14482 07:32
sjn yes
had a few warnings during build though 07:33
07:33 uniejo joined
sjn none of them look serious though 07:35
sjn tries again 07:36
moritz well, I have no clue. Could you please open a ticket? mailto:rakudobug@perl.org (including information about your system)
sjn mm 07:37
if the error is still there after a rebuild, I'll look into it
07:54 iblechbot joined
sjn seems it's ok now :-) 08:01
moritz so it didn't work after the first 'make realclean' + rebuild, but worked fine after the second? 08:06
chromatic Wow, an exception handler crash in multi-threaded code. 08:19
Easy to explain though. 08:20
08:20 UltraDM joined
chromatic If your thread throws an exception and you don't have a handler, Parrot won't exit. 08:20
find_exception_handler will return NULL.
But Parrot_ex_throw_from_c doesn't expect that and tries to dereference the pointer. 08:21
08:29 particle1 joined, Zaba joined
chromatic ... and fixed. 08:31
dalek r32358 | chromatic++ | trunk: 08:32
: [exceptions] Fixed a segfault when a thread threw an exception without an
: active handler. The handler finding code now detaches the thread. The main
: interpreter exits in this situation, but previously the handler finding code
: returned NULL, which the handler invoking code happily dereferenced. BOOM.
: Per RT #45917, this check should probably move into Parrot_exit, but one thing
: at a time.
diff: www.parrotvm.org/svn/parrot/revision?rev=32358
r32359 | chromatic++ | trunk: 08:37
: [t] Replaced all uses of deprecated find_global opcode with get_hll_global and
: get_global opcodes.
diff: www.parrotvm.org/svn/parrot/revision?rev=32359
09:16 cosimo joined 10:02 magnachef__ joined 10:13 mj41 joined 10:15 masak joined 11:04 tomyan joined
bacek g'night everyone 11:05
masak bacek: good night. 11:06
purl sleep well too
bacek masak: I've readsubmit my revised junction patch directly to #60168... My mail didn't hit RT in more than 16 hours... 11:07
masak: so, you can try it :)
masak bacek: goodie.
I will when time permits.
bacek masak: fair enough
11:10 kj joined 12:02 gmansi joined 12:16 magnachef joined 13:11 ruoso joined 13:53 jhorwitz joined 14:00 gryphon joined 14:24 PacoLinux joined 14:30 masak joined 14:49 magnachef joined
Coke yawns. 15:04
Patterner stuffs a durian cookie on Coke's open mouth
Coke vomits. 15:23
. o O (Sorry. :|)
bug going around the haus.
anyone remember what '.global' was? was it like '.local'?
(can't seem to use it anymore, but it's still taking up space in imcc.[yl]) 15:24
kj .global was never implemented 15:25
the keyword was there, but you'd get some message I believe, saying it's not implemented
I think it was *supposed* to be like .local, but then... ehm. well globally scoped :-) 15:26
Coke ok. then I think it can be safely removed. 15:28
(unless someone is relying on the error message you get when you try to use it..) 15:29
particle1 blame the line and search the log
kj yes I think it was removed already from the parser
PerlJam remove it and see who screams
kj it never has worked, AFAIK 15:30
particle yes, i agree with kj, but it's better to check first than remove without warning
Coke if it hasn't worked, then removing it won't matter. 15:32
particle *sigh*. we need more tests. 15:33
kj I suspect that ack \\.global won't give you anything, except perhaps in imcc itself
Coke it's only used in an .s file, which isn't PIR/PASM.
kj what *is* an .s file?
Coke particle: as long as they are real tests, sure! =-)
jonathan Assembly, ain't it?
particle yes, .s is assembly 15:34
kj that's a N/A then
well, the *wrong* assembly :-) 15:37
particle ok, how do i start X? 15:38
kubuntu 7.10
moritz startx 15:39
or do you want /etc/init.d/kdm start?
15:39 davidfetter joined
particle actually, now i have a console process that i can't interrupt with ^C ^D or ^Z 15:40
so i can't even try to start x 15:41
is there some three-finger salute i should try?
moritz go to another terminal? Ctrl+Alt+F2 15:42
Coke future events? 15:52
Zaba particle, try ^\\ 15:53
Coke wonders why his 'make test' is suddenly so verbose on feather.
rant: I just ran a very long test cycle, but my changes to imcc.y and .l didn't cause a rebuild, even with --maintainer. 16:11
kj so bison and flex weren't invoked then.. 16:20
Coke nope. timestamp trouble. 16:21
particle: any serious objection to removing the last remnant of .global? 16:25
(which isn't used, all tests pass, couldn't possibly have worked.)
particle remove it!
Coke k. letting 'codetest' finish, will commit then. 16:26
masak Coke++ 16:32
dalek r32360 | coke++ | trunk:
: remove last remnant of '.global' from IMCC.
: - the token is recognized, but has no function; it's not used, and is going away as part of the 'global' cleanup mentioned in DEPRECATED.pod.
diff: www.parrotvm.org/svn/parrot/revision?rev=32360
16:47 Theory joined 16:55 ruoso joined 16:59 gryphon joined 17:18 hercynium joined 17:22 johbar joined 17:28 magnachef joined 17:32 magnachef__ joined 17:46 rdice joined 18:23 sjansen joined
dalek r32361 | coke++ | trunk: 18:24
: [DEPRECATED] these ops have been deprecated for some time; mark them so that "./parrot -w" reports them.
diff: www.parrotvm.org/svn/parrot/revision?rev=32361
sjn hello 18:38
Coke sjn: hio
sjn thought of looking at some of the perl6 spectests
just to get into it 18:39
perhaps look for tests that don't need to be skipped anymore? :)
any suggestions on where to start? :)
moritz sjn: no, that's automated ;)
sjn <- completely newbie 18:40
moritz sjn: t/TASKS in the pugs repository contains a list of things that need to be done in the test suite
sjn moritz: ok, what else would be good for a neophyte like myself to dig into? :)
moritz tests, tests, tests ;) 18:41
sjn I'll look there
oh, Pugs isn't part of the parrot tree anymore
moritz it never was. 18:43
(there was some pugs related stuff in the parrot repo, but that wasn't pugs itself, afaict)
svn co svn.pugscode.org/pugs/
sjn ook 18:44
sjn is checking out pugs 18:45
jonathan btw, if you just do make spectest, you can get a checkout just of the spectests too, under t/spec. I have done that, then I commit from that copy. :-) 18:47
sjn jonathan: now you tell me :) 18:48
moritz jonathan: but then you don't have t/TASKS, the README, HOWTO, the smartlinks checker etc.
sjn would it be an idea to check out those too, when checking out the spectest files? 18:49
particle no, unless they're moved to t/spec/ 18:50
jonathan moritz: Oh, that's a good point. :-)
particle but, you could just co pugs/t/ if all you want is the tests
jonathan Will be having second Rakudo day this week, tomorrow. 18:51
jonathan has some to catch up on after his vacation
Plus it's more fun than most of the other stuff I do...
sjn yay \\o/ 18:55
jonathan: so, how long until the next time you're in Oslo? :)
jonathan sjn: Not sure - I'm planning to be at the next Nordic Perl Workshop. 18:56
sjn sweet :)
jonathan I like Oslo. Until I go to buy a beer. :-) 18:57
sjn wanna come to the hackathon too? :)
jonathan Ooh, that could be fun.
The last Oslo hackathon was good, so I'd happily attend another one. 18:58
sjn sweet
moritz next year I'll come to Oslo as well (if it works out), to visit some of my girlfriend's family (and vacation, of course ;) 18:59
jonathan Ah, 2 day hackathon after 2 day workshop. And on a weekend too. 19:00
Looks good.
purl O_O
Tene argh. Work just scheduled me for a flight from SLC to ATL on the 16th
Leaving at 1:30 PM.
I sent them mail. They're supposedly coming up with a solution. 19:01
sjn www.perlfoundation.org/perl6/index....katon_2009
jonathan OK, food time...back later. 19:02
19:06 chromatic joined
krunen hi, all. Should I do anything if I find a spec test that passes for rakudo, but is not in t/spectest.data? 19:09
moritz krunen: depends on the file... 19:10
krunen: which one is it?
S29-math/trig.t isn't in t/spectest.data because the test file is wrong
connect.t isn't in it because it doesn't run any test by default 19:11
moritz should learn how to not scare off people ;) 19:13
krunen moritz: ok, trying to find it again... I lost my history :( 19:17
moritz krunen: irclog.perlgeek.de/parrot/today helps a lot ;)
particle i think he means shell history
moritz oh. 19:18
chromatic twitter.com/kroah then
krunen here it was, sorry: S02-literals/char-by-name.t 19:25
dalek r32362 | moritz++ | trunk:
: [rakudo] improved documentation of tools/update_passing_test_data.pl, masak++
diff: www.parrotvm.org/svn/parrot/revision?rev=32362
moritz dang, the local svn hangs again 19:27
Coke chromatic: so; as long we do incremental updates where all tests pass in between, there's no reason not to do the type ids work in trunk, izzer?
moritz: been seeing those for the past few days.
chromatic: I'm thinking we could put in the first patch from the branch that removed '.Integer' and friends. 19:28
chromatic +1 from me. 19:31
19:33 davidfetter joined 19:35 bacek joined
dalek r32363 | coke++ | trunk: 19:38
: [docs] update filename so the perldoc hint actually works.
diff: www.parrotvm.org/svn/parrot/revision?rev=32363
jonathan For the Parrot summit - do I just book a room at the hotel myself? That's for me to organize, right? Also, anyone wanting to share? 19:39
Coke chromatic: I tried but it requires some loving care I can't afford atm.
I'll see if I can do it later.
chromatic I managed to fix some of the find_global stuff last night. 19:40
Found and fixed a bug in threads and exceptions at the same time.
particle jonathan: go ahead and book a room, i've reserved 6 double-doubles and three single-kings for now 19:44
jonathan particle: OK, should I book a double-double and assume someone will want to share? 19:45
jhorwitz particle: any super secret password we should give to get the rates? :) 19:46
particle i'm just sending an email to jeanine about that now, jhorwitz
jhorwitz coolio
particle jonathan: sure, go ahead with that plan 19:47
jonathan OK
particle: I'm arriving on the Thursday rather than Friday. But I think I may, since I'm flying into SFO which is nearer the city than the conference hotel/venue, stay in the city the night before and then on the Friday come to the conference hotel. 19:48
jhorwitz: I see you also arrive into SFO. How were you planning on doing the airport to the hotel? 19:49
jhorwitz either rental, cab, or quantum teleportation
jonathan Fancy meeting at SFO and heading there together, and sharing the cost? 19:50
jhorwitz i think particle has dibs on the last one though. ;-)
when are you getting in?
particle jonathan: wise move (hotel)
jhorwitz oh i see
jonathan jhorwitz: Well, I arrive the previous day, but plan to stay the night in the city.
jhorwitz yeah, let's do that.
jonathan From what I can see, the airport is in the correct direction anyway. 19:51
OK
jhorwitz done and done!
jonathan So you're in at 1:30pm?
jhorwitz no....checking....
davidfetter who's going to beijing perl workshop?
jonathan davidfetter: Are you?
davidfetter <-- heading to beijing in a couple of hours
yep :)
particle jonathan, jhorwitz, there's also caltran. google transit is your friend
jhorwitz have to check that out 19:52
jonathan: i get in 6:58 PM
davidfetter if i didn't love traveling so much, i'd think it was my turn in the barrel this year
jonathan I spotted the caltran, but it doesn't go all the way.
Coke chromatic: saw that. I followed up with a removal of the old .global stub.
jonathan davidfetter: Beijing is nice - have a good time! :-)
davidfetter so i've heard
jonathan jhorwitz: Ah, OK. I was looking at the departures page rather than arrivals. ;-) 19:53
davidfetter wondering whether they do beijing duck any better there than elsewhere
not that i've had beijing duck much...
jhorwitz jonathan: does that still work for you?
particle jonathan: i have a minivan, and arrive at sjc early afternoon
also, we'll need a second car, so maybe i'll get one for you from sfo, jhorwitz
i get an excellent discount at avis
jhorwitz particle: that works. esp. if anyone else is coming in early evening 19:54
to SFO
jonathan Ooh, do I have multiple offers here... :-)
I think pmichaud and I were going to try and do some Rakudo stuff on Friday afternoon after arriving...so earlier may work a bit better. 19:55
jhorwitz is lonely
i'll check the arrivals page for potential companions... :) 19:56
jonathan particle: How is SJC to get to?
TimToady phone in 5
davidfetter well, you have to be in SJ to get there, so not so hot 19:57
particle jonathan: ~$40 cab ride 19:58
...from the hotel
dalek r32364 | coke++ | trunk:
: [cage] fix typo in diagnostic
diff: www.parrotvm.org/svn/parrot/revision?rev=32364
particle jonathan: when you book your room, you can use the 'parrot' group name. if they don't know what you're talking about, don't worry, they will soon enough. 19:59
Tene Aw, work chose to cancel my class instead of paying for my flight out of CA. 20:00
particle that's... silly.
jonathan particle: No, I meant from center of SFO to SJC
erm
Center of San Francisco to SJC...
grr
jonathan thinks there's gotta be public transport 20:01
particle jonathan: you'll have to look at google transit or maps
jhorwitz um, gcc ouch: perl6.ops:106: error: unable to find a register to spill in class 'SIREG'
TimToady phone!!!!
purl phone is always interrupting
davidfetter jonathan, caltrain can get you most of the way
Coke seen allison? 20:03
purl allison was last seen on #parrot 1 days, 52 minutes and 20 seconds ago, saying: heads to airport [Nov 4 19:11:15 2008]
Coke ... how about now? 20:04
jonathan davidfetter: thanks 20:05
Tene particle: not cancel, actually, just have someone else teach it. I said 'cancel' because I was reading the trip cancelation email from expedia then.
davidfetter jonathan, check out 511.org/ 20:06
re: beijing, anybody have an idea whether there's an easy way to get a local SIM card?
particle anyone reserving a room... you can email jeanine conforti directly if you wish at jconforti@jdvhotels.com. mention 'parrot', and she'll help you along. 20:08
20:11 mberends joined
jhorwitz jonathan: ping 20:54
jonathan jhorwitz: pong 20:58
jhorwitz the memcpys you added to perl6.ops recently are causing register issues with gcc on x86. silly gcc memcpy optimization. 20:59
perl6.ops:106: error: unable to find a register to spill in class 'SIREG'
particle your fault for using such a flaky compiler, i guess
jonathan wtf...
jhorwitz stomps on particle
chromatic Use memmove instead.
jhorwitz yes
jonathan Is there a pragma to disable it?
jhorwitz chromatic's solution is mine as well 21:00
jonathan fn
jhorwitz coolio
jonathan Feel free to just change 'em and commit, or I'll do it soon (tomorrow latest)
jhorwitz wishes for more registers on x86
i'll go ahead and commit.
Coke chromatic: for RT#
chromatic wishes for 32-bit pointers AND the use of his 64-bit machine's extra registers. 21:01
Coke?
purl hmmm... Coke is mailto:will@coleda.com
Coke 48014, it would be helpful (IMO) to have a patch that ripped out the union val and replaced it with whatever was going to replace it; then folks could apply that and fix build failures.
21:01 rob joined
Coke neh? 21:02
chromatic Mostly we just undefine PMC_int_val, PMC_pmc_val, PMC_struct_val, etc.
21:03 masak joined
masak hooray, :: works in class names again! 21:03
Coke what about int_val2?
masak hugs jonathan
moritz hugs November
chromatic Yeah, that too.
I just posted confederacy.patch to that ticket. 21:04
particle we definitely need cached ints for efficient array push/pop/shift/unshift
Coke if we're moving to attributes, we lose that, neh? 21:05
21:05 Andy joined
particle there's still room for cached values iirc 21:05
Coke skips that ticket. 21:06
chromatic: temporary_pmc_free doesn't return. 21:07
chromatic Yeah, that should be void. Thanks!
particle to be clear, you should name that 'furlough' instead of 'escape' 21:09
jonathan What does furlough mean?
chromatic I think of GC as a prison, not a resort.
jonathan :-)
chromatic Hm, is SVN slow because I'm also downloading an interview, or are we cursed? 21:10
dalek r32365 | chromatic++ | trunk:
: [src] Fixed return signature of temporary_pmc_free() (it's void), reported by
: Will Coleda.
diff: www.parrotvm.org/svn/parrot/revision?rev=32365
Coke we're cursed. 21:12
It's been hanging on commit for some time.
jhorwitz hm, i just committed and it went fine 21:13
dalek r32366 | jhorwitz++ | trunk:
: [rakudo] use memmove instead of memcpy to avoid gcc register spilling failures on x86
diff: www.parrotvm.org/svn/parrot/revision?rev=32366
chromatic It was just slow, but it didn't hang.
Interesting.
Coke -> 21:16
particle .
bacek $ ../../parrot perl6.pbc -e 'my $j=(Bool::True&Bool::False); say ?$j == Bool::False; say ?(Bool::True&Bool::False) == Bool::False; say true($j); say true($j) == Bool::False;' 21:24
1
0
bacek cries
chromatic You're in the wrong universe for the last two tests. 21:25
pmichaud say true($j) == Bool::False # all(Bool::True, Bool::False) 21:27
oh, perhaps not.
because true($j) imposes boolean context.
bacek pmichaud: its bug in parsing (looks like) 21:28
$ ../../parrot perl6.pbc --target=pir -e 'my $j=(Bool::True&Bool::False); say true($j).WHAT'
pmichaud I suspect it's a bug in the true() implementation.
bacek $P20 = $P19."WHAT"() 21:29
$P21 = "true"($P20)
$P22 = "say"($P21)
pmichaud yes, we currently have some difficulties with method calls on results of named unaries
bacek find_lex $P19, "$j" was few lines befoore
pmichaud that's a known bug.
bacek ah, ok...
"true" implementation looks good for me... 21:30
pmichaud try (try($j)).WHAT instead
er,
(true($j)).WHAT
or even
(true $j).WHAT
chromatic (true $^a).DAT 21:31
There, the "FORREALZ?" macro.
bacek Bool...
pmichaud so yes, it's a (known) parsing issue. 21:32
moritz it seems that parsing Perl 6 is non-trivial. 21:34
masak *gasp*
pmichaud how many lines of code is STD.pm ? ;-) 21:35
moritz 4k
jonathan How many lines of code does it take to implement something that *runs* STD.pm? ;-)
moritz a few more ;)
particle use v6;
you didn't specify what language to use :P 21:36
jonathan :P 21:38
bacek $ ../../parrot perl6.pbc -e 'use Test; plan 1; $a="a"; $b="b"; $c="c"; ok(($a & "b" & "c") eq ("a" | $b| $c), "Passed")' 21:40
No handler to delete.
current instr.: 'parrot;PAST;Compiler;as_post' pc 6052 (src/PAST/Compiler.pir:1651)
called from Sub 'parrot;PAST;Compiler;post_children' pc 1779 (src/PAST/Compiler.pir:367)
yak...
afk # kids, school, work... 21:41
pmichaud "No handler to delete" means that there's one-too-many pop_eh's somewhere.
bacek pmichaud: I didn't touch PCT :)
moritz bacek: when you write tests like this, take care that ok() doesn't autothread over its first argument
pmichaud I don't think that rakudo is autothreading at all yet.
moritz that's not an issue yet because autothreading is NYI
pmichaud but yes, in Perl 6 that would tend to autothread.
bacek moritz: it's from S03-junctions/misc.t
moritz but I don't want to see that in the test suite 21:42
:(
dalek r32367 | kjs++ | trunk: 22:00
: [pirc] fix symbol declarations and parameters. Parameters segfaulted, but this is fixed now.Refactor/redo some code.
diff: www.parrotvm.org/svn/parrot/revision?rev=32367
22:02 kj joined 22:04 Whiteknight joined
kj thought he'd broken pirc badly, but luckily it's still fine... 22:07
Whiteknight (not breaking things badly)++ 22:16
particle *cough* GC *cough* 22:17
;)
dalek r32368 | kjs++ | trunk: 22:32
: [pirc] add support for braced arguments in macro expansion.
diff: www.parrotvm.org/svn/parrot/revision?rev=32368
tewk_ Whiteknight: you coming to the summit? 22:42
Whiteknight Not sure yet, still working out the details
i would very much like to
tewk_ I'm starting to work with GCs in PLTScheme, I'd love to talk 22:43
Is your GC work based off suns G1 at allt?
Whiteknight very distant relations, if any 22:51
i mostly made the algorithm up by myself, based on a few references I read 22:56
a little too ambitious, as it turned out
tewk_ So does the WKGC use conservative tracing of the stack. 22:57
?
PLTScheme does some clever pre_processing of .c files to allow precise tracing of the stack. 22:58
The G1 paper looks impressive. complex but impressive. 22:59
jonathan Wow...that's...scary.
tewk_: Does that pre-processing claim to work generically on C, or have they impsoed some restrictions etc? 23:00
particle i'm sure "clever" means "quite limited in scope" and "vm-specific" 23:01
chromatic I'm sure it's also difficult to reason about dynops and dynpmcs too. 23:02
tewk_ jonathan: the only thing it requires is to unnest functioncalls, and that is just a formality. other tools like CIL can do that automatically. 23:03
b = x( a() ) => t = a(); b = b(t)
dynops and dynpmcs would work just fine as long as they are preprocessed. 23:06
jonathan ponders reading the paper out of curiosity 23:09
23:10 TiMBuS joined 23:11 clunker3_ joined
tewk_ G1 is different from what 23:11
PLTScheme does
Unfortunately I don't have a good single paper reference 23:13
jonathan I'm sure PLTScheme have a paper about their approach too.
ah
tewk_ I've got a prelim paper plus a dissertation I could point you to.
jonathan "Our guestrooms come equipped with a Yo-Yo, Etch A Sketch and Rubik's Cube" - so I read this and my first thought is, "and what about wifi?" 23:14
tewk_: Sure - I can always look at it if I get time. 23:15
chromatic Can you do concurrent stack scanning?
tewk_ sure as long as you stop the world. 23:17
portal.acm.org/citation.cfm?id=5124...N=69693195
I can send you the pdf if you don't have acm 23:18
www.cs.utah.edu/~awick/awick-dissertation.pdf
particle jonathan: yes, the yo-yo has wifi. i'll have to check on the rubik's cube.
chromatic I meant incremental.
jonathan particle: Thanks...I've heard the wifi on the yo-yo tends to go up and down a lot... 23:20
tewk_ incremental stack scanning? 23:25
jonathan chromatic: You want to scan the stack of one thread, from another thread? 23:26
chromatic: Without having to suspend the thread whose stack you're scanning?
chromatic Yeah, I meant heap too.
Perhaps I should not multitask and discuss concurrent, incremental GC.
I get lost.
tewk_ It basically creates a tunnel through the stack containing only pointe values..
pointer values. 23:27
Right now its stop the world.
23:29 bacek_ joined
jonathan is happy to see that California is a tad warmer than where he is right now 23:30
dalek r32369 | kjs++ | trunk: 23:50
: [pirc] More Tricks with Bison and Flex.
: * override Flex' memory allocation function, which now use Parrot's mem_sys_allocate and friends.
: * add an option to the .y file which generates the header file; no more bison -d pir.y; just bison pir.y
: * some other neat trick with the grammar to Do The Right Thing in case of unless a relop b goto L; in case of "unless", don't invert "gt" and "ge" ops, as they then will be inverted AGAIN, resulting in, you guess it: "gt" and "ge", they do not exist in Parrot.
diff: www.parrotvm.org/svn/parrot/revision?rev=32369
23:54 Andy joined

Powered by Cro and the Raku® Programming Language.

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

Thank you!