Parrot 0.6.0 "P&P" released | Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Mar 31 | tinyurl.com/2pmnlq
Set by moderator on 18 March 2008.
chromatic I'm less convinced. 00:00
The justification for s/does/provides/ in PMCs is so as not to confuse the syntax for applying a role to a PMC with the syntax for declaring that the PMC fulfills the same duties as if you had applied that role to the PMC. 00:01
Infinoid so, which one does the current op naming suggest? and which one should it suggest? 00:06
(I'm just a lowly cage cleaner.)
chromatic The *opcode* name asks "Do you perform this role?" 00:07
The does *keyword* in PMCs says "Compose in this role at compilation time."
The provides keyword in PMCs says "I fulfill this role without composition."
wknight8111 how does one become a lowly cage cleaner? 00:14
Infinoid you compose the "lowly_cage_cleaner" role, with the "does" keyword. :) 00:16
wknight8111 you don't subclass the committer class with limited attributes? 00:18
ewilhelm I think the committer traits can be added at runtime 00:20
chromatic Subclassing would break Liskov.
wknight8111 my $wknight8111 := new 'Committer', 'cage_cleaner';
haha, i think i mixed PIR and NQP syntax 00:21
wknight8111 hangs his head in shame 00:22
I've been bouncing back and forth between the two all day, trying to write my Octave port 00:27
01:58 slightlyoff joined
dalek r26698 | chromatic++ | trunk: 02:01
: [PDD] Typo fixes and minor formatting nits.
diff: www.parrotvm.org/svn/parrot/revision?rev=26698
Coke sends out his promised type id email. 02:07
... is there anything else I owe parrot today?
particle feed the bird.
Coke heads off to read his new spider robinson book. 02:08
02:09 Juerd_ joined 02:14 Juerd_ joined 02:30 grim_fandango joined 02:38 guru joined 02:50 Andy joined 02:53 Theory joined
tetragon Hrm... I can't view the ticket mentioned in one of the TODO tests that's crashing on me 03:26
Tene Wow, I actually have a running laptop with internet access at my hotel tonight. 03:27
And nothing is broken.
Infinoid Tene: knock on wood
Tene As long as you don't knock on the laptop
;) 03:28
Infinoid or touch it
or look at it
tetragon: rt seems to be working, here. is there something wrong with that particular ticket?
tetragon 46511
Tene But... but... but... svn up! I need my svn up!
tetragon I don't have permission to view it 03:29
Tene tetragon: there are two urls to view tickets, one public and one private.
Infinoid hmm. neither do I.
Tene Or something weird like that.
Infinoid (and I'm logged in and everything)
tetragon I'm also logged in
I set up an account after one of my replies was eaten by some mail server
Infinoid Tene: neither the Ticket/Display.html nor the Public/Bug/Display.html will view it 03:30
whatever permissions it needs, I don't have them
I think I'm set up with basic committer privs
tetragon It's test 14 of t/pmc/threads.t 03:31
Infinoid nice.
# Failed (TODO) test 'CLONE_CODE | CLONE_CLASSES; superclass built-in'
# at t/pmc/threads.t line 664.
# Exited with error code: [SIGNAL 11]
03:32 AndyA joined
tetragon I get SIGNAL 10 03:32
Infinoid sig11 is SIGSEGV on my OS
tetragon My one keeps putting SIGBUS in the crash reports, but depending upon the test I can get 10, 6, and I think I've seen others 03:33
Infinoid well, whatever names those numbers map to, its crashing, and that's bad. :)
tetragon Time to create a new ticket, as I have no idea what the hidden one is about 03:35
I even know which thread crashed 03:36
Infinoid + local $TODO = "vtable overrides aren't properly cloned RT# 46511";
-- from r22492 diff
tetragon Anyway, the one for it I created is 52398 03:42
Infinoid cool. 03:49
04:39 tetragon joined
chromatic tetragon, I just fixed the segfault in RT #41097. 04:51
tetragon You did? 04:52
chromatic Yep, going to check it in in a second.
dalek r26699 | chromatic++ | trunk: 04:55
: [IMCC] Made IMCC report an error when encountering get_results, get_params,
: set_args, and set_returns opcodes with the wrong number of arguments. Fixes RT
: #41097.
diff: www.parrotvm.org/svn/parrot/revision?rev=26699
tetragon On a non-crash note, test 2 of t/compilers/imcc/syn/regressions.t is failing on me 04:57
It can't find the opcode div_i_ic_ic 05:00
Infinoid ooh. what platform? 05:01
tetragon ppc
Infinoid I reopened #43048 yesterday because of the same thing on amd64
been banging my head on the table, trying to find out what's different about amd64... maybe its something different about x86 instead. 05:02
tetragon I can see about copying a fresh source tree over to my 32-bit Intel Mac, and see what it does there 05:03
Infinoid ok. Coke tested it on x86 linux and saw no problems there 05:04
I spent a bunch of time poking around in gdb yesterday. (single-instruction-stepping through longjmp is fun, by the way.) 05:06
found a difference of behavior in compilers/imcc/parser_util.c line 653: the retval was NULL on both platforms, but the "ok" flag was set to 11 on x86
and it was 0 on amd64.
but that's as far as I got with it 05:07
05:10 wknight8111 joined
tetragon 32-bit intel mac passes the test 05:20
chromatic Which runcore?
tetragon I'm also getting 0 for ok on line 653 on ppc 05:26
And 11 on i386 05:27
Infinoid chromatic: all of them. occurs with -C, -f, -g, -j, -p, -S, -t, or whatever's used by default. 05:30
chromatic That should be easy to reproduce.
Infinoid need an account to poke around on? 05:31
tetragon ok is 11 in its undefined state on i386, but not on ppc
Infinoid I thought 11 looked suspicious, since the function in question only sets it to 1 05:32
tetragon The second time flow goes through there on i386, it starts as 110 05:34
And is then set to 0 by IMCC_subst_constants
Infinoid I wonder if initializing ok to 0 causes it to break on x86. 05:35
Infinoid tries it
tetragon There are some cases in IMCC_subst_constant where ok isn't touched 05:38
A single return statement 05:39
Infinoid awesome, it does break x86.
tetragon I think that return is being hit
i386 passes because ok remains 11 05:40
tetragon sets yet another breakpoint
nopaste "Infinoid" at 76.215.208.106 pasted "Initialize the darned variable." (13 lines) at nopaste.snit.ch/12606
Infinoid I'm tempted to check that in, as-is.
Infinoid likes breaking x86. Stuff gets fixed when its reproducible on x86. :P 05:41
tetragon Yep, it's hitting the return that doesn't modify ok
chromatic Wow. That's not a lovely function at all.
tetragon Line 980 of optimizer.c
That's the return that doesn't touch ok 05:42
chromatic It ought to.
tetragon Well, unlike the other two returns in the function, it doesn't
chromatic With the appropriate modification, do you get the segfault? 05:43
tetragon That case was one that that didn't segfault
It just failed
I haven't tested the 41097 changes yet 05:45
05:48 Ademan joined
Infinoid what does this code need to do, to make the test pass? 05:48
05:49 jan joined
Infinoid I'm not sure whether ok should be set to 0 or 1 for that return, but neither value seems to have an effect on the test. 05:49
05:50 zarchne joined
dalek r26700 | chromatic++ | trunk: 05:51
: [t] Untodo the test for RT #41097, which is passing as of r26699.
diff: www.parrotvm.org/svn/parrot/revision?rev=26700
Infinoid the way I understand it, it tries a constant-folding optimization, fails that, and so does it at runtime. (That much is consistant between platforms.) The weird thing is that x86 wasn't breaking at runtime, I guess because of the uninitialized "ok" variable. 05:52
So does that mean we need the 'div_i_ic_ic' op after all?
chromatic Looks like it to me. 05:53
I tried adding it, but...
Infinoid looks to see what else breaks on x86 when "ok" is initialized to 0 05:54
tetragon chromatic: Your segfault fix works on ppc 05:57
chromatic Good.
nopaste "chromatic" at 63.105.17.30 pasted "Test Patch to Fix IMCC syntax regression" (1428 lines) at nopaste.snit.ch/12607 05:59
Infinoid hooray for make -j, hooray for ccache. 06:00
tetragon wonders about if anything special needs to be done for ppc 06:02
chromatic Hm, it looks like *ok = 0 is wrong.
Infinoid ok 2 - cannot constant fold div by 0
chromatic Probably some of the PPC JIT is wrong in that some of the functions now take interp as their first argument.
That'd be in src/jit/ppc/core.jit
Heavily macro-ized and templated code.
Infinoid chromatic: patch looks good on amd64 06:03
chromatic Something breaks x86 horribly though.
tetragon Feh, *ok = 0 was equivalent to what your amd64 and my ppc were doing
Hrm... the build process crashing generally isn't a good sign 06:05
Infinoid I had to do a realclean before it built cleanly
chromatic Yeah, I invalidated bytecode.
Infinoid hmm, I see no x86 breakage. 06:13
chromatic I don't, if I do the int i = 0; 06:14
tetragon t/compilers/imcc/syn/regressions.t passes ppc 06:16
Infinoid oh, I do see breakage, just not in the test I had originally tried.
t/compilers/imcc/syn/op.t for example
tetragon I also see breakage in op.t
Infinoid breakage is consistent between x86 and amd64
tetragon Tests 16 and 26-31 fail
Infinoid yep 06:17
Infinoid heads off to bed, goodnight 06:18
chromatic only t/compilers/imcc/imcpasm/opt1.t fails for me. 06:19
test 15
tetragon I also get a failure in t/compilers/imcc/syn/const.t 06:20
Test 2
Couldn't find the opcode add_i_ic_ic 06:21
The op.t failures are of the same variety of not finding opcodes
nopaste "Infinoid" at 76.215.208.106 pasted "Test failures in t/compilers/imcc/ from nopaste.snit.ch/12607" (518 lines) at nopaste.snit.ch/12608
Infinoid (that's on amd64) 06:22
*gone*
chromatic Ah, but I added the opcode in my patch.
tetragon Loads of sigbus 06:23
opt1.t has a load of failing tests, opt2.t has three tests that not only fail, but sigbus 06:24
nopaste "chromatic" at 63.105.17.30 pasted "Test Patch for IMCC Regressions, Take Two" (1429 lines) at nopaste.snit.ch/12609
chromatic Is that patch any better?
Revert the old one, and try this one.
If you already did a make realclean for the old one, no need to do it again for this one. 06:25
tetragon Bah, right after I started one
chromatic Here's a chance to plug ccache then. 06:28
dalek r26701 | duff++ | trunk: 06:29
: [rakudo] Add loop statement
diff: www.parrotvm.org/svn/parrot/revision?rev=26701
tetragon All tests under t/compilers/imcc/reg and syn pass 06:38
But test 15 of imcpasm/opt1.t fails 06:39
No sigbus crashes in t/compilers/imcc/* other than syn/macro.t test 32, which is not a new failure 06:40
chromatic I expect opt1 to fail
tetragon No other failures in imcpasm
chromatic Where were you getting the segfault before? 06:41
tetragon the macro.t test 32 one?
It didn't change
not ok 32 - invalid label syntax # TODO RT#47978, RT#51104 06:42
chromatic Oh. Is t/compilers/imcc/syn/regressions.t passing now then? 06:43
tetragon LYes
chromatic So it fixed that much anyway.
purl, messages
msg Coke Does nopaste.snit.ch/12609 make sense for fixing t/compilers/imcc/syn/regressions.t? 06:44
purl Message for coke stored.
07:18 darbelo joined 07:20 iblechbot joined 07:57 dalek joined
dalek r26702 | fperrad++ | trunk: 08:03
: [docs]
: - add README_cygwin.pod
diff: www.parrotvm.org/svn/parrot/revision?rev=26702
08:15 darbelo left 08:27 askie joined 08:57 contingencyplan joined 09:38 nopaste joined 09:45 ruoso joined 10:48 divot_powell joined 11:40 IllvilJa joined 11:44 iblechbot joined
dalek r26703 | kjs++ | trunk: 12:41
: [pdd29] give pdd29 a first swing
diff: www.parrotvm.org/svn/parrot/revision?rev=26703
13:03 muixirt joined 13:16 skids joined 13:21 ruz_ joined 13:32 gryphon joined 13:38 gryphon joined
Coke (regressions.t) feel free to re-todo that test until I fix it. 13:39
13:42 rdice joined 13:44 jhorwitz joined 13:47 gryphon joined
Coke msg chromatic : regarding your nopaste: yup, make sense, but there are other ops we shouldn't skip then: anything that causes an exception to be raised during constant folding. (like div_n_nc_nc) 13:52
purl Message for chromatic stored.
Coke will try to apply that when he gets home tonight if someone doesn't beat him to it.
should also add a test for the floating point version of that test.
jhorwitz particle: ping 13:53
13:54 ruz_ joined 13:55 gryphon joined
particle jhorwitz: pong 14:06
jhorwitz hey there particle
particle hi-o jhorwitz
jhorwitz see the oscon schedule? they put mod_parrot and rakudo in the same slot! 14:07
14:07 iblechbot joined
particle uh. no. rats. 14:08
link?
jhorwitz en.oreilly.com/oscon2008/public/sch...2008-07-23
shorten jhorwitz's url is at xrl.us/bisig
particle i'll tickle allison's ear about it
jhorwitz yeah, it would suck to split our audiences
particle yep, that would be four upset people. 14:09
jhorwitz and i want to see your talk, anyway. :)
jhorwitz makes five
particle planning on yapc::na? 14:10
jhorwitz if my talk is accepted, i'll obviously go. i think i'll go anyway, since i haven't been to chicago 14:11
particle you're going. we're hacking the weekend before. i'll likely buy my tickets later this week. 14:13
shower &=
14:13 DarkWolf84 joined
Coke will be there the weekend before as well. 14:16
jhorwitz Coke: i seem to remember meeting you briefly in toronto in 2005. 14:20
14:26 davidfetter joined
particle i think i introduced you two 14:27
jhorwitz are you typing from the shower?
avar g/w 24 14:30
14:33 Andy joined
Coke jhorwitz: very likey. 14:36
"likely"
Andy Morning all. 14:37
purl evening, Andy
Andy Is Parrot still continuing?
particle no, audrey used to do that
Andy Even though I haven't done any work on it in weeks?
particle until she fried her laptop while doing so
...perhaps 'drowned' is a better word :)
andy. we couldn't continue without you, so we destroyed the repo. 14:38
we're all working on ack now.
Andy You are? 14:39
jhorwitz mod_ack is a pain...
Andy Coulda fooled me!
I hope there's a backup of the repo. I'd like to do some Parrot again.
Just no time right now.
Of course, I have to write slides about how to write C.
14:40 ruz_ joined 14:42 kj joined 14:49 jan joined
pmichaud looks at booking tickets for yapc::na and oscon 14:58
Debolaz should visit one of these places one day. 15:00
Infinoid too.
pmichaud are we hackathoning or anything around oscon? 15:01
particle yes, weekend before
i'm flying in friday night
pmichaud oscon, not yapc
particle ah. righto.
i'm free either weekend 15:02
pmichaud I'm thinking I may want to attend Damian's vim tutorial
particle start a mail, or bring it up at sixperl
pmichaud and perhaps ubuntu live
15:02 jjore_ joined
particle 60*36 15:08
purl 2160
particle 2160-1917
purl 243
particle 243/36
purl 6.75
15:09 rhandom joined
Coke ... try #balanceMyCheckbook =-) 15:19
particle i'm pricing a new laptop 15:23
$1917. $2160 at $60/mo for 36mo lease means it costs $6.75/mo in interest. 15:24
not worth it.
pmichaud yes, I may be in the market for a new laptop soon 15:25
I can no longer extend the insurance on my current laptop (which has been a dream), so when/if it breaks I'll need a new one on short notice :-)
particle i really like the look and specs for the lenovo x61. i'd love to see one in person before buying 15:26
Infinoid *hearts* his t61
particle i have a t43, wish it was a t61 15:28
but it's not mine, it belongs to a client
pmichaud I have a Sharpie pen -- maybe you can switch the number? ;-)
Andy Rakudo blogging++ reddit.com/info/6e3o0/comments/?alr...itted=true 15:30
shorten Andy's url is at xrl.us/bisof
15:30 Theory joined 15:31 ruz_ joined
Tene particle: what do you think of the x300? 15:34
15:34 divot_powell joined
pmichaud heh 15:36
my laptop is a (Dell) x300 :-P
Tene Heh.
particle i really want to like the x300, but it's not powerful enough for me
pmichaud I think it's a terrific laptop. I wish I could extend the coverage on it. :-)
I'm seriously considering that my next laptop will be a mac
pmichaud decides to check out Dell's website, just in case 15:37
Tene I'm probably buying a T61 soon.
particle i considered that, too. then i found somebody who installed osx on a dell.
Tene I get an employee discount, too.
pmichaud well, I've been very happy with kubuntu lately -- 8.04 beta runs extremely well for me on my systems
Tene Well, employee friends and family.
Hm, any parroters around Philadelphia next week? 15:38
pmichaud particle: are you planning to stay in the on-campus accommodations at yapc::na ?
particle yes. there's nothing else convenient 15:39
divot_powell Hi, I'm a student at the University of Edinburgh considering applying for the summer of code program. I have a background ibn compiling and I've been looking at the proposals online and am quite interested in the implementation of perl6 in perl6 and the AST optimisation for parrot. Is there anyone here that can give me a little more background on either of these projects? Such as skills you recommend for them or whe
*rakudo 15:40
particle divot_powell: i think the ast optimization project is a good one, because you can learn from what gcc did with GIMPLE and Tree-SSA
...and it would benefit all languages running on parrot 15:41
perl 6 in perl 6 is a goal (maybe) but not important at this point in rakudo development
it's much more important that we extend rakudo to something that can be used generally before we worry about bootstrapping 15:42
divot_powell the AST did catch my eye as I have been learning alot about AST's recently on my course. Do you have any idea as to a timescale of when I should have done certain things by etc? The reason I ask is due to the amount of exams I have coming up in the next month and a bit 15:43
particle the google soc calendar is available on google's site 15:47
lemme find a link 15:48
code.google.com/opensource/gsoc/200...1_timeline 15:49
shorten particle's url is at xrl.us/bispn
divot_powell excellent, thanks... 15:50
Sorry there's just a fair amount of info online and it can be a bit daunting at times...
particle read the faq. it'll answer a lot of questions 15:51
pmichaud the yapc::na wiki says hackathon will be June 19-20 (?!) 15:58
particle you'd better fix it. 16:05
the parrot hackathon will be the 14-15
pmichaud will we have access to on-campus accommodations then? 16:06
(I'm composing a reply to the list)
particle i believe so. jmcadams is the one to poke about that 16:07
iirc there was a thread about that already
however perhaps it was off-list
pmichaud right, I'm replying to the thread.
particle ah. perfect.
did audrey previously attend sixperl? 16:08
pmichaud no
particle i think it may be time. i'll mention it today 16:09
pmichaud iirc, she was invited but declined
she didn't want to be involved in meetings that weren't "completely open"
but I could definitely be mis-remembering that. it was long ago. 16:10
particle that does sound like her style :) 16:14
dalek r26704 | infinoid++ | trunk: 16:28
: [imcc, raduko]
: Minor fixups to pass t/codingstd/c_parens.t and t/codingstd/cuddled_else.t.
diff: www.parrotvm.org/svn/parrot/revision?rev=26704
16:58 peeps[work] joined
particle heh: perl.org/six 17:06
17:06 Psyche^ joined 17:12 sjansen joined
Coke I wonder who set that up. :p 17:16
kj Back to the 80s 17:17
17:19 grim_fandango joined
Infinoid guess you had to be there. 17:20
kj oh yaa! :-)
cotto_work particle++ for finding that 17:22
17:26 cotto_work joined 17:52 barney joined
barney abc 1.3.0 plnews.org/posts/abc_130_released_2...61600.html 18:11
shorten barney's url is at xrl.us/biszx
18:28 cotto_work joined
dalek r26705 | bernhard++ | trunk: 18:36
: [HQ9+]
: Fix whitespace handling. Whitspace is not required between commands.
diff: www.parrotvm.org/svn/parrot/revision?rev=26705
19:00 peepsalot joined
dalek r26706 | kjs++ | trunk: 19:14
: [NEWS] mention the new pdd29.
diff: www.parrotvm.org/svn/parrot/revision?rev=26706
19:21 ruoso joined 19:23 kj joined 19:29 barney joined
TimToady phone 19:59
pmichaud can't find my phone :-( 20:00
btiab
20:03 IllvilJa joined 20:09 cotto_work joined
cotto_work once type ids are eliminated, will the proper way to check types within PMCs be via VTABLE_isa and VTABLE_does? 20:48
Coke does doesn't tell you the specific type, and isa only tells you if you've guessed right. 20:54
20:55 slightlyoff joined
Coke VTABLE_name and VTABLE_getclass, more likely. 20:55
... it would be nice if the .ops files didn't have to specify "goto NEXT()" but instead had that as a keyword. 20:57
cotto_work ok 21:00
that's going to be less than a fun transition 21:01
Coke cotto_work: (less than fun) hence the branch. 21:04
cotto_work working out-of-tree FTL
Coke are you anti-branch? heretic. =-) 21:07
Infinoid Coke: I tried doing a s/goto NEXT()/next/g and putting a "#define next goto NEXT()" at the top. But it turns out "next" is also fairly common as a variable name, so that didn't build. any other choices? 21:09
Infinoid is a bit bored at the moment, and therefore, up for trying weird stuff like this. 21:10
Coke goto NEXT() is already sugar.
I'd add a flag like "normal-flow-control" except shorter and saner as one of the flags. If that flag is present, act as if "goto NEXT()" was the last line of the tag body. 21:11
Infinoid darn, that sounds *so* much less hackish. 21:12
Coke have at it if you're bored. 21:14
Infinoid stage 1, figure out how to add flags in .ops files. stage 3, profit. 21:15
dalek r26707 | kjs++ | trunk: 21:17
: [pdd19] add some more proposals to pdd19; it will be reviewed shortly (I heard), so this is the chance to put in ideas.
diff: www.parrotvm.org/svn/parrot/revision?rev=26707
21:21 wknight8111 joined
pmichaud use.perl.org/~pmichaud/journal/36035 21:26
cotto_work Coke, I'm still waiting for the legal fun to clear. I'm reasonably sure you were joking, but I'm definitely eager to get my code back in svn. 21:27
Infinoid is there any visible progress on that, by the way? 21:32
21:43 sjansen joined
cotto_work Infinoid, none that I can see, but my boss is getting back from vacation on Monday, at which point I'll get an update. 21:43
Infinoid great, hoping it all goes through smoothly and quickly :) 21:44
cotto_work it's neither as far as I can tell, but I'll be content with it just going through
fwiw, I think it's quite likely to happen 21:47
Coke cotto_work: I was talking to Infinoid. 21:54
I know you're hamstrung atm.
cotto_work and now I know that you know 22:07
(watch out for the recursion)
22:08 grim_fandango joined
Infinoid I know you know I know you know, but you didn't know I knew that you knew I knew you knew, until I just now told you. 22:11
cotto_work my stack broke 22:12
Tene picks up the pieces. 22:14
cotto_work hey! I need those.
22:18 pjcj joined 22:20 pjcj joined 22:21 pjcj joined 22:26 kid51 joined 22:27 jan joined 22:54 Debolaz joined 23:00 TonyC joined 23:01 tetragon joined 23:04 Limbic_Region joined 23:16 Theory joined 23:34 nopaste joined 23:39 liona29 joined 23:40 TonyC joined 23:47 jan joined 23:51 skids 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!