Parrot 1.3.0 "Andean Swift" released | parrot.org
Set by moderator on 23 June 2009.
dalek rrot: r39955 | jkeenan++ | trunk (14 files):
Use most traditional form of File::Path::mkpath(), as it seems to work best across all versions of Perl and File::Path.
 00:02
pmichaud bacek_at_work: I'd want to see a patch for it, first. Overall I'm not sure I see the necessity for them. 00:05
bacek_at_work pmichaud: ops2c uses them heavily. 00:08
pmichaud the tricky part is parsing them correctly.
bacek_at_work pmichaud: (about patch) I'm thinking about "borrowing" quote_expression.pir from Rakudo and simplify it
pmichaud well, I've also been thinking about putting regexes directly into NQP 00:09
so at that point you pretty much want all of quote_expression.pir :-)
bacek_at_work s/and simplify it//
:)
kid51 Trac is very slow; possibly timing out on Submits.
00:10 Limbic_Region joined
bacek_at_work pmichaud: currently NQP isn't very good at string munching. And afaiu we are going to replace Perl5 based tools with PCT based. Adding string interpolation will help this transition. 00:11
pmichaud bacek_at_work: I think I'm agreeing with you now 00:12
but I would just say to take quote_expression.pir as-is
it should be pretty
easy to install
and there's not a whole lot of benefit to simplifying it
00:12 Limbic_Region joined, Infinoid joined, kid51 joined, s1n joined, spinclad joined, davidius joined, solarion joined, jevin joined, Khisanth joined, Tene joined, japhb joined, Coke_afk joined, TonyC joined, dukeleto joined, preflex joined, silug joined, purl joined, rhr joined, rblackwe joined, workbench joined, jjore joined, TimToady joined, nnunley joined, mmpf joined
bacek_at_work bah, netsplit 00:13
pmichaud: Agreed. I'll borrow it as-is.
pmichaud: another question/suggestion. 00:14
trac.parrot.org/parrot/browser/bra...?rev=39948
cotto Yay! Easier string manipulation in nqp will make me happy.
kid51 Trac is currently unresponsive to Submits. I have one that got as far as the first place where it needed to resolve a hyperlink. It's been hanging for > 5 minutes. 00:15
bacek_at_work There is ".sub '' :anon..." which is workaround for absence of "class Foo is Bar". Can we add it to NQP as well?
cotto: btw, we don't need sub "new" in op/op.pir 00:16
cotto I just cargo-culted it over from pmcc. 00:17
If you can excise it without breaking anything, go for it.
bacek_at_work starring at "at_work" suffix... 00:18
00:18 skids joined
pmichaud NQP doesn't currently support "is Bar"? 00:18
I thought it did.
you're right, it doesn't 00:19
I'm open to looking at patches to add "is Bar"
but I might be a bit picky there.
bacek_at_work pmichaud: ok. I'll "borrow" some more code from Rakudo :) 00:20
pmichaud well, Rakudo's handling of classes is likely to not fit well with NQP
it's certainly not as straightforward as quote_expression.pir 00:21
I'm thinking NQP will want its own new design
kid51 Hmm, my recent Macports installation provided me with an updated 'svn' client without my being aware of it. 00:23
Now, if you get a conflict while doing 'svn up', you get a message like this: 00:24
Conflict discovered in 'MANIFEST'.
Select: (p) postpone, (df) diff-full, (e) edit,
(mc) mine-conflict, (tc) theirs-conflict,
(s) show all options: p
C MANIFEST
... forcing you to take some action (if only hitting a single key) immediately.
00:56 patspam joined
dalek rrot: r39956 | jkeenan++ | branches/darwinhints/t/configure/049-options_test.t:
Beginning to get this test file back into shape; much to do.
 01:00
01:02 kurahaupo joined 01:25 hercynium joined 01:32 darbelo left 01:51 mikehh_ joined 02:15 brbrooks joined 02:16 brbrooks left, brbrooks joined 02:20 Ryan52 joined
Ryan52 does Allison Randal IRC? 02:20
rg yes, most often on tuesdays for #parrotsketch. 02:22
Ryan52 ok, I'll just email then. 02:24
jdv79 any idea when sub traits will be available? 02:28
02:29 brooksbp joined
dalek rrot: r39957 | jkeenan++ | branches/darwinhints/t/configure/049-options_test.t:
Restore this test file to good health.
 02:34
rrot: r39958 | jkeenan++ | branches/darwinhints (1 files):
Make name of test file more self-documenting.
02:36 Theory joined 02:41 janus joined 02:43 Andy joined 02:53 Andy joined 02:54 amuck joined
Austin purl: seen pmichaud? 03:03
purl pmichaud was last seen on #parrot 2 hours, 42 minutes and 7 seconds ago, saying: I'm thinking NQP will want its own new design
03:19 dukeleto joined 03:20 donaldh joined 04:01 Coke joined, cognominal joined 04:10 Coke joined
Coke hopes he didn't just brick his iphone. 04:10
brbrooks by... 04:11
Coke msg kid51 messaging me on irc is not a good way to get svn support.
purl Message for kid51 stored.
Coke msg kid51 I'd say open a ticket, but if trac is also giving you crap, an email to the -dev list is the way to go.
purl Message for kid51 stored.
Coke brbrooks: no clue; I docked it, it didn't show up. Found a site that suggested I put it into a "please recover me" mode. Did so, recovered... got an error. rebooted the machine, trying again. 04:12
I've been using it all day, no clue why it's acting up now.
brbrooks what gen do you have? 04:13
Coke 1.
it's the 4GB model.
brbrooks any dead pixels?
errr "colorful" pixels 04:14
Coke some... not dead, but one patch that's glowy.
very small. presumably from one of the drops.
brbrooks yeah can't stand it
Coke this one will eventually go to the wife. =-)
brbrooks iphone still functionally responsive? Just doesn't show up in itunes? 04:16
Coke it was responsive until I put it into recovery mode, yes.
just wasn't showing up as a device. (showed up fine once it was in recov mode.)
brbrooks nice 04:19
i hate how iphone os 3.0 supports MMS, but AT&T doesn't yet... or at least for the iphone 04:20
Coke scams upon scams. 04:21
still a very shiny phone.
brbrooks Pretty useful... excellent UI for basic phone stuff, best I've ever played with... and I use Maps all the time 04:22
the whole idea/concept of the App Store is pure scam 04:23
except for Shazam... that's a damn fine app 04:24
04:27 tetragon joined 04:30 dukeleto joined 04:32 eternaleye_ joined 04:39 eternaleye joined
cotto pmichaud, ping 04:41
seen pmichaud
purl pmichaud was last seen on #parrot 4 hours, 19 minutes and 48 seconds ago, saying: I'm thinking NQP will want its own new design
05:23 Theory joined 05:33 tetragon joined 06:08 uniejo joined, barney joined
mikehh All tests PASS (pre/post config, smolder, fulltest) at r39958 - Ubuntu 9.04 amd64 06:19
cotto msg bacek opsc needs to split ops into different variants before sticking them all into the past. If you need something to do, you could figure that out. 06:25
purl Message for bacek stored.
cotto time for sleep 06:30
(I hope)
bacek_at_work cotto: ping 06:36
cotto: I don't like idea of "Op.set_body" called from Actions. 06:37
06:39 tetragon joined 06:40 Theory joined 06:47 iblechbot joined 07:16 eternaleye joined 07:21 donaldh joined 07:24 tetragon joined
cotto bacek_at_work, how would you do it? 07:31
(If I can't sleep, this is the next best thing.) 07:32
There needs to be some code somewhere to analyze the body and set the appropriate jump flags. 07:34
clock? 07:44
purl cotto: LAX: Thu 12:44am PDT / CHI: Thu 2:44am CDT / NYC: Thu 3:44am EDT / LON: Thu 8:44am BST / BER: Thu 9:44am CEST / IND: Thu 1:14pm IST / TOK: Thu 4:44pm JST / SYD: Thu 5:44pm EST /
bacek_at_work cotto: body analyze is emitter's job. 07:46
PAST nodes should be similar for ops and L1ops 07:47
cotto I can buy that. The functionality in op/op.pm would simply be pushed into runcore/base.pm. 07:48
bacek_at_work indeed. 07:50
afk. # not going home yet :/ 07:51
cotto bacek_at_work, feel free to implement that. I'd really rather be asleep, so no coding for me tonight. 07:52
I'll put it on my todo list either way. I like having other people shorten my todo list. 07:55
dalek rrot: r39959 | cotto++ | branches/ops_pct/compilers/opsc/runcore/runcore.pm:
[opsc] delete runcore.pm; base.pm is the base class
 07:58
08:07 brooksbp joined
brooksbp false fire alarm in apt building 08:08
no worries, people don't have to be at work in the morning or anything like that
dalek kudo: e7a7999 | moritz++ | docs/ChangeLog:
[docs] ChangeLog updates
 08:32
08:36 MoC joined 08:47 Zak joined 09:07 whoppix joined 09:19 pjcj joined 09:21 tetragon joined, bacek joined 10:19 Ryan52 joined 10:23 eternaleye joined 10:26 mokurai joined 10:29 AndyA joined 10:46 tetragon joined 11:20 donaldh joined
dalek rrot: r39960 | bacek++ | branches/ops_pct (5 files):
[nqp] Carefully steal quote_expression from Rakudo.

  - Switch "PIR q" to use quote_expression directly
  - Add stubs for interpolations.		 11:41
rrot: r39961 | bacek++ | branches/ops_pct/compilers/nqp (2 files):
[nqp] Implment simple scalar interpolation.
 12:12
bacek msg cotto Look at r39961! It will make you happy :) 12:13
purl Message for cotto stored.
bacek msg pmichaud Can you review trac.parrot.org/parrot/changeset/39960/ please? It's stol^W borrowed quote_expression from Rakudo planted into NQP.
purl Message for pmichaud stored.
bacek msg pmichaud Yak. We'll need deprecation notice about string interpolation in NQP. Even Rakudo can't built as-is. Or don't interpolate "quoted" string. Interpolate qq<> only. 12:23
purl Message for pmichaud stored.
nopaste "bacek" at 122.110.37.140 pasted "Small fixes of Rakudo's actions.pm related to string interpolation for pmichaud++" (52 lines) at nopaste.snit.ch/17179 12:29
moritz bacek: uhm wait, you're not going to do string interpolation is single quoted strings, are you? 12:33
s/is/in/
bacek moritz: of course not.
moritz: check nopaste
moritz bacek: I did.
bacek and...? 12:34
purl somebody said and was the register frame something like the Sparc's?
moritz bacek: you substituted '$!' with '\\$!'
bacek in "quoted string"
moritz ah, sorry
bacek no worries :) 12:35
I just stea^W borrowed code from Rakudo almost as-is
12:47 Andy joined 12:49 Whiteknight joined
nopaste "bacek" at 122.110.37.140 pasted "Simple "class Foo is Bar" implementation for pmichaud++ for review." (43 lines) at nopaste.snit.ch/17180 12:50
12:52 mokurai left 12:53 skids joined
Whiteknight good morning #parrot 12:54
bacek Whiteknight: good evening! 12:55
Whiteknight good evening bacek!
12:55 Coke joined
Coke ~~ 12:55
moritz wanders what Coke smart-matched against :-) 12:56
bacek Coke matched to... Empty string???
Coke is it.
is still stuck in perl 5.8, apparently, as he doesn't use ~~ to mean that.
13:07 masak joined
dalek tracwiki: v6 | coke++ | SmokingParrot 13:08
tracwiki: trailing &amp;&amp; in shell command is not boss.
tracwiki: trac.parrot.org/parrot/wiki/Smokin...ction=diff
cotto bacek: <3 13:11
bacek cotto: :) good morning
dalek TT #822 created by coke++: Using wrong libparrot. 13:20
Whiteknight Installing stuff is TEH SUX 13:26
cotto Yay for me! Interview time! 13:29
moritz cotto: good luck! 13:30
purl You'll need it.
bacek cotto: good luck!
purl You'll need it.
cotto "I can haz job plzkthxbye?"
Coke cotto: break a leg.
bacek And it explains why you appear on #parrot in so unsual time :) 13:31
cotto yeah 13:32
moritz (amusingly in German you even wish "break neck and leg") 13:33
cotto It figures, since that's where germs were invented. ;) 13:36
13:37 ruoso joined
bacek must sleep 13:39
purl $bacek->sleep(8 * 3600);
bacek see you tomorrow
13:39 mj41 joined 13:40 ruoso joined
cotto night, bacek 13:41
13:42 ruoso joined 13:44 ruoso joined 13:46 ruoso joined 13:49 ruoso joined 13:50 ruoso joined 13:51 slavorg joined
cotto wheeeeee! 13:52
purl i guess wheeeeee is wee or Wii
whoppix Whiteknight, you got around to make that list yet? 14:07
14:09 hercynium joined
Whiteknight what list? 14:10
purl list is groups.google.com/group/parrot-dev or take that, moose-heads
Whiteknight is far too forgetful
14:16 jdv79 joined 14:26 jdv79 joined
Coke Alias++ # cpandb & friends. 14:30
14:48 iblechbot joined
dalek TT #115 closed by doughera++: [todo] make test with already installed parrot 14:49
14:56 Theory joined
Coke doughera++ 15:30
15:35 Theory joined 15:40 dukeleto joined
dalek TT #823 created by coke++: packfile pmc tests failing 15:45
Coke who do the packfile pmcs belong to?
15:53 estrabd joined 15:54 estrabd joined
dalek n: d64c0bb | Jarrod++ | (5 files):
Fixed the build to work with the new parrot. Seems fine..
 15:57
16:05 jdv79 joined 16:11 jrtayloriv joined 16:13 jdv79_ joined 16:21 brbrooks joined 16:22 Psyche^ joined 16:31 register joined
register I just read the news about L1 on Wk blog 16:36
isn't this new assembly language going to slow down the release of parrot considerably
? 16:37
16:40 jrtayloriv joined
Whiteknight register: no, I don't think so 16:44
I think the transition will be very gradual
Coke it will have no impact on the /frequency/ of releases, anyway.
those happen the 3rd tuesday of every month, for good or ill. 16:45
dalek a: 60eb87f | fperrad++ | t/function.t:
ambiguous function call : add a TODO test
 16:51
brbrooks "All object-like things that get allocated during the execution of parrot bytecode are managed from the arena_base member of the interpreter structure" 16:55
Where is the code that allocates these objects to the arenas?
Whiteknight src/gc/* somewhere 16:57
I think api.c or mark_sweep.c
brbrooks so it's a hook into the gc api?
Whiteknight yes
brbrooks like function callback or somethin
Whiteknight no, not a callback, just a regular API function 16:58
brbrooks k
are these PObjs just objects that I would normally think of as allocated on the heap?
Tene We should just put all of parrot's code in src/gc/, so that could be the answer to every question! 17:01
17:02 Theory joined
Coke Can someone else eyeball the patch in TT #822 - worksforme, go ahead and apply it if you like it. 17:02
17:04 chromatic joined
Coke Can someone else eyeball the patch in TT #822 - worksforme, go ahead and apply it if you like it. (resend for chromatic, as he's been known to apply things.) 17:05
chromatic I wish I could see the color highlighting in Trac for patches. 17:08
Coke color blind? 17:09
chromatic Yep.
Coke we can probably adjust the colors. 17:10
17:10 japhb joined
chromatic Something other than very light red and very light green would help. 17:10
NotFound Yeah, it lacks contrast in luminance.
Coke chromatic: I don't appear to have sudo privs on that box, so I don't think I can fix it the hard way. 17:13
chromatic What's the easy way, a modified retrovirus to grow more cones?
NotFound chromatic: a private stylesheet 17:14
Coke looks like there's a PatchRenderer plugin...
ah, looking in the wrong place. need the trac setup, not the apache docroot. 17:18
dalek rrot: r39962 | chromatic++ | trunk/src/pmc/default.pmc:
[PMC] Added the PObj_is_special flag to PMCs when they get new properties. The

perform it.
17:23 Austin joined 17:24 flh joined 17:27 cghene joined
dalek rrot: r39963 | chromatic++ | trunk/src (2 files):
[GC] Moved the setting of the special PMC flag into Parrot_gc_add_pmc_ext(),

flag on the PMC.		 17:28
chromatic I have a good feeling about that. 17:29
particle is adding pmc_ext to a PMC the *only* reason to set that flag? 17:30
17:31 hercynium joined
chromatic No, but it's a sufficient reason. 17:31
particle sure, i'm not arguing that it isn't.
i want to make sure the documentation for that flag is accurate 17:32
or, for the macro to set that flag
Coke chromatic: I'll see what I can do about the colors. I think I need a trac plugin.
chromatic particle, that part of the flags and the GC is messy; it could use cleanup for sure. 17:33
particle coke: colorschemedesigner.com/ has a very helpful color picker 17:35
Coke particle: I can't even pick "white" atm, but thanks. 17:37
17:39 japhb joined
NotFound chromatic: @-moz-document url-prefix(trac.parrot.org/) { mod .l { background-color: red ! important; } } 17:40
17:41 HG` joined
dalek rrot: r39964 | chromatic++ | trunk (2 files):
[PCC] Humanized error reporting for argument overflow/underflow. Now Parrot

See Rakudo RT #67358. There's more discussion there about the wording change, but the pluralization change is a clear benefit right now.		 17:41
rrot: r39965 | chromatic++ | trunk/src/hash.c:
[src] Tidied code; no functional changes, I hope.
NotFound Put this on ./mozilla/firefox/.....default/chrome/userContent.css and the trac page will look as ugly as you may want. 17:42
chromatic Trying that, thank you. 17:43
NotFound Don't forget to restart firefox. 17:44
particle r39964 stinks for i18n
Coke not like we're doing well in that area anyway. 17:45
particle but we can address it when we get to i18n
NotFound Uh, there was a dot before mod that the paste has eated.
Whiteknight chromatic++ on that flag thingy
those damn PObj flags are a pain in the ass
particle i laid the groundwork for i18n, but tuits are expensive and hard to find these days
chromatic At least r39964 makes pluralization for i18n more obvious. 17:46
Let's see how Rakudo does with that PObj flag now.
NotFound particle: it stinks more without that patch.
Whiteknight that could be what bacek was seeing where some hash keys weren't being marked 17:48
NotFound chromatic: this is the correct version: @-moz-document url-prefix(trac.parrot.org/) { .mod .l { background-color: red ! important; } }
Whiteknight if the property hash wasn't getting marked, and it contained the only references to certain keys, that would cause what he was seeing
chromatic That's my hope. 17:49
17:54 jrtayloriv joined
Whiteknight that's a good case where the documentation for the flag being more clear might have alerted us to a problem much sooner 17:57
chromatic I still get Method 'succ' not found for invocant of class 'Method' for t/01-sanity/07-try.t though. 17:58
particle Whiteknight: that's why you want to update those docs now :) 18:01
chromatic Looking at p6opaque's clone() now. 18:03
Whiteknight particle: I want to do a lot of stuff. Good luck finding me some tuits for it 18:05
what I need to do is get fired from my job. Then I can work on parrot all day long
...while eating PB&J sandwiches every day, and not being able to afford to drive anywhere or wash my laundry 18:06
NotFound Whiteknight: find a job in a pet shop
chromatic You can live like a king if you don't mind living in a tent, eating soup made from discarded ketchup packets.
18:07 jdv79 joined
Whiteknight yes, exactly like a king 18:09
chromatic The Fisher King, perhaps. 18:10
NotFound And if you want to drive anywwhere, you can trade the kingdom for a horse. 18:11
particle well, then, it's settled. let's get whiteknight fired. 18:13
brbrooks Eating PB&J sandwiches every day ain't so bad. Not being able to wash your laundry... different story... 18:22
chromatic Move somewhere that drizzles. 18:23
particle 's fine, if you're telecommuting
Coke chromatic: I hope that mozilla fix works for you, because trac admin is not something I am good at. =-) 18:33
we now have ThemePlugin installed.looks like I have to /write a plugin/ to customize it now. whee.
Whiteknight I did the PB&J diet for about a month and a half freshman year, and ended up in the campus clinic with a severe vitamin deficiency problem 18:34
nurse was using words like "pre-scurvy" and gave me vitamin C
szbalint heh, no wonder
this is why I order pizza with pineapple on it 18:35
Whiteknight oh, and B complex. Doesn't matter how many sandwiches you eat if your body cant metabolize them
18:35 jdv79 joined, jrtayloriv joined, japhb joined, hercynium joined, Austin joined, chromatic joined, Theory joined, brbrooks joined, estrabd joined, dukeleto joined, ruoso joined, skids joined, Whiteknight joined, eternaleye joined, Ryan52 joined, Zak joined, COke_afk joined, mmpf joined, nnunley joined, TimToady joined, jjore joined, workbench joined, rblackwe joined, rhr joined, purl joined, silug joined, preflex joined, Tene joined, Khisanth joined, jevin joined, solarion joined, davidius joined, spinclad joined, s1n joined, Infinoid joined, confound joined, hachi joined
Whiteknight netsplit-- 18:36
szbalint talking about pizza: dev.perl.hu/static/IMG_0010.JPG - the next batch, looking similarly should arrive soon 18:37
Austin So that's what? Corn, onions, and ham? 18:38
szbalint and oregano and pineapples
Whiteknight is that ...corn?
Austin Okay, I get the ham+pineapple -- we call that a Hawaiian. I totally don't understand the corn. 18:39
Also, someone is having roundness problems.
Whiteknight yeah, a lot of guys do 18:40
Austin szbalint: FYI, WhiteKnight and I live not too far apart, in a part of the US that is noted for having a pizza place about every 0.5 km.
szbalint that pizza is from Vienna
and corn is a great match :)
Whiteknight I've got 3 pizza places in walking distance from my work 18:41
more if you count the lunch trucks at the university down the road
Austin My house, for example, is within a two mile radius of 22 (!) pizza shops that deliver, plus 2 that don't, not counting whatever may be on the other side of the river (less than 2 miles away).
Whiteknight !!!
szbalint quantity < quality :)
Austin So we're going to be pizza snobs.
Whiteknight corn < quality :)
chromatic My house is within walking distance of 37 Starbucks, and I live a mile away from two shopping centers. 18:42
Austin See if you can get some shrimp, crab meat, and asparagus on a pizza. (Yowza.)
Chromatic: that's an interesting hell. Who did you piss off?
chromatic I'm within 250 miles of Seattle. It's inevitable. 18:43
Whiteknight doesn't even drink coffee
Austin (Although I suppose it's a great place to be if you're a teenager looking for work.)
chromatic I think there's only one SB within walking distance, though there are spots downtown where you're within a mile of at least a dozen.
Austin :)
As long as you're always within walking distance of a starbucks, their mission is accomplished. 18:44
szbalint I prefer tea :) 18:45
Austin I prefer beer.
Coke Houston had a similar concentration of SB.
szbalint that too
Austin But I'll settle for Coke in the mornings. (I don't drink coffee, either.)
Coke BOOYAH.
Coke , perhaps ironically, is a big coffee drinker. 18:46
Austin Houston is wierd.
pmichaud Looks like four Starbucks within 2 miles of my house
Austin *weird
No zoning laws.
pmichaud but if I moved 2 miles west of here, there'd be eight
chromatic They're all hoping that a great old one will arise from the Gulf and devour Houston again.
Austin I remember driving past a roadside mall that contained a tobacco shop, a baptist church, a Subway, and a strip club. 18:47
I thought that was perhaps the greatest commercial plaza ever. Everything you really need, in one place.
pmichaud They're just missing a liquor store.
Coke ia ia c'thulhu fthouston.
chromatic Add a drive-through liquor store and you have Texas in a nutshell.
Austin It's Texas - I think all four of them sold package goods. 18:48
Coke houston is not like the rest of texas.
IM(L)E.
Austin True.
pmichaud Thankfully.
Austin It's better than a lot of the rest of Texas, for sure.
pmichaud I don't know about "a lot". Some.
chromatic Galveston...
purl somebody said galveston was my best bet for seeing "America's Ocean" :-)
pmichaud Beaumont 18:49
purl Beaumont is oil town
Coke galveston is also the big fat rabbit from gaiman's _the day I swapped my dad for two goldfish_
purl okay, Coke.
chromatic Oh yeah, the whole golden triangle there. Beaumont is a special kind of awful.
pmichaud or "the Golden Triangle" in general (Beaumont/Orange/Port something)
I used to drive through the golden triangle every couple of weeks. Reminded me of the mississippi delta. 18:50
chromatic Port Arthur
Orange is the least awful.
pmichaud Primarily because compares favorably with what is to the east :-) 18:51
*it compares
chromatic It has a nice bridge.
pmichaud Port Arthur's level of industrialization was... spooky. On par with Freeport. 18:52
chromatic That whole thing is one oil refinery.
pmichaud "Somewhere among all of these pipes, stacks, and flares there's a city with actual people living(?) in it." 18:53
particle Metropolis
purl i heard Metropolis was GORGEOUS
pmichaud When I hear "Metropolis" I always think "skyscrapers"
Port Arthur doesn't qualify.
unless we count refinery flare towers as "skyscrapers". 18:54
chromatic If you love pipes, it's like Disneyland.
particle precisely, and the working class feeds them, working in the basements
szbalint pipes should be in walls or computers. I really dislike industrial landscape heh. 18:55
particle then why do you like perl 5? 18:57
everything is exposed.
szbalint perl 5 is in computers 18:58
jrtayloriv I'm getting the following message when running 'make test': pastebin.com/m4d61def6 ... Is this something that I should post a bug report for? If so what information should I provide along with this? 19:00
(sorry if this question is inappropriate to ask here ... I'm new to this) 19:01
szbalint besides in computer analogies I think of pipes as information flow, instead of a metaphor for exposed internals :)
chromatic jrtayloriv, that's an appropriate question for here. We should debug that. 19:02
jrtayloriv chromatic, should I post a bug report? 19:03
chromatic I was hoping that someone here would have time to debug this interactively with you; that'd be faster. 19:05
NotFound I think there is already a ticket for that, and maybe more than one. 19:06
19:12 jdv79 joined 19:18 HG` joined 19:20 brbrooks joined
NotFound This is related, but not the same: trac.parrot.org/parrot/ticket/406 19:26
The main problem is that the library and the configure probe does different things.
Coke NotFound: that test is skipped unconditionally now. how is he getting the error? 19:29
he shouldn't even be running the test.
ah. because we're testing pcre in 2 places. :|
NotFound One more time, we have examples of what to not do X-) 19:31
19:46 davidius joined
Coke NotFound: right way to check is Parrot::Config, HAS_PRCE flag. 19:54
you want the ticket? if not, I can fix this tonight. 19:56
... easy fix, I'll just do it. 19:58
NotFound Coke: not so easy. That way will give fails when Configure is able to find libcpre but the library and/or the example or the test can't. 20:00
The problem is that the probe checks if a program using libpcre can be compiled, not if it can runtime-linking with it. 20:02
And IMO the checking for libpcre availability must be done by the parrot library itself, not by his examples and tests. 20:04
mikehh All tests PASS (pre/post config, smolder, fulltest) at r39965 - Ubuntu 9.04 amd64 20:14
pmichaud ....intresting data point for gc bug hunters... 20:18
with current parrot head, 64-bit linux fails all of its tests in rakudo
details to come
rakudo e7a799, parrot r39965 20:19
Whiteknight did chromatic's patch have any beneficial effect on Rakudo?
or has nobody completed a spectest run yet? 20:20
pmichaud I think that's my point
now all tests fail
NotFound Same for debian 5.0 amd64
Whiteknight all tests?
purl all tests are green
Whiteknight i'd call that a regression
pmichaud all spectests
I'm starting again to check
NotFound Same as mikehh, I mean
pmichaud most of 32-bit passes, though.
Whiteknight do the failures go away pre r39962? 20:21
pmichaud don't know yet, will try that.
Whiteknight I ask because r39962 is when chromatic fixed the PObj_is_special flag in a few places 20:22
which I still think looks like a good fix, unless that flag is used in other nefarious ways
NotFound rakudo is starting to pass tests here 20:26
pmichaud confirmed, with r39965 all spectests fail under 64-bit
(completely fresh checkout of all components)
NotFound pmichaud: does not fail here, 2.6.26-2-amd64 20:28
Whiteknight okay, so that stupid flag is causing stupid failures that don't make a lot of sense 20:29
pmichaud r39961 fails also. 20:30
Whiteknight oh...well the flag is still stupid
pmichaud (all tests fail with "Method 'succ' not found..."
NotFound At least his name is not very clever
pmichaud checking a few other things
Whiteknight I seem to remember somebody else complaining about that same error the other day 20:33
pmichaud it's common in these gc failures now 20:34
Whiteknight at least, that wording is familar o me
pmichaud that's how most of the failing scripts choose to fail
Whiteknight damnit
chromatic r39965 must be this line: - memset(new_bi + old_size, 0, sizeof (HashBucket *) * old_size); 20:37
+ memset(new_bi + old_size, 0, sizeof (HashBucket *) * (new_size - old_size));
NotFound spectest is running fine, also
pmichaud I'm retrying with r39935, which was my last known good test. 20:38
20:39 mattc joined 20:40 Austin joined
pmichaud r39935 still works, trying r39950 20:44
20:44 estrabd joined
moritz pmichaud: are you bisecting parrot for the GC bugs? 21:05
pmichaud moritz: yes 21:12
r39950 fails
Austin pmichaud: Does the C preprocessor in the c99 language work? 21:13
pmichaud Austin: I have no idea
Austin ok
pmichaud I don't remember who wrote that
Austin Istr you and Jerry talking knowledgeably about it at PVMW. 21:14
pmichaud I remember you and I were discussing handling '#include' in PGE
but I don't remember c99 specifically
maybe it was something else, or maybe something I said prompted particle to bring up that there was a c99 parser written for PGE/Parrot 21:15
Austin That's okay. I was asking in case it prompted you to say something like "Oh, there's a bunch of stuff not done yet." 21:16
pmichaud I'd bet that's the case, but I couldn't say that for certain :-)
Austin :) 21:18
pmichaud r39942 fails
Austin on to ..38? 21:19
pmichaud yes.
21:19 bacek joined 21:21 davidius joined
Austin pmichaud: Is there any kind of global-symbol aliasing for registers, the way there is for .lex? 21:23
pmichaud Austin: no, not yet.
Austin Does that mean its planned, or jus that it's possible?
*just
pmichaud afaik it's not even planned yet.
Austin <whew>
Cool. 21:24
pmichaud I suspect we could potentially get lexicals to map to i/s/n registers instead of just PMCs
for anything else it's likely to need some additional work
r39938 fails, trying r39936 21:25
chromatic Please don't be r39337... 21:28
Austin If you have to pray...
you probably already know how it's going to turn out. 21:29
pmichaud r39936 passes (at least it doesn't instantly fail like r39938)
that pretty much leaves r39337
21:30 Limbic_Region joined
pmichaud note that these fail only on 64-bit, on 32-bit I have the same random set of ~4 failing tests 21:30
chromatic Given that r39938 was a branch commit and r39936 tidied some generated code....
pmichaud right.
trying 39337 now, to confirm.
moritz if you haven't tested r39337, you can't know if r39337 or r39338 caused it
NotFound pmichaud: it doesn't fail in my Debian amd64
pmichaud except that 39338 is a commit to a different branch. 21:31
moritz ok
pmichaud confirmed. r39337 causes all spectests (at least as far as I've let it run) to fail with "Method 'succ' not found for invocant of class 'Method'" 21:34
sorry, 39937, not 39337
that remains true for all test revisions through current head (i.e., 39965) 21:35
*tested
moritz r39337 looked rather harmless
r39937 does not.
pmichaud right, I just mistyped earlier (and kept following the mistype). In my test environment the revisions were all correct. 21:36
NotFound Urgh. I was testing in one machine and looking at the svn info in a ssh to the other %-)
moritz (but both were commits by NotFound++) 21:37
pmichaud at least some of the tests pass when run from the .pbc directly (without -G), but fail when run from the fakecutable 21:38
let's see if I can find any that fail from the .pbc
so far they all work from the .pbc and fail from the fakecutable. 21:42
21:42 Whiteknight joined
Whiteknight what's that status of that bug? 21:42
whoppix Whiteknight, the list about the topics that need illustration.
Whiteknight, so that I can improve the docs with shiny little vector graphics 21:43
Whiteknight ah, right. I forgot about that
let me look into that tonight
21:44 Psyche^ joined
whoppix sure, just drop me a message when you got something. 21:44
Whiteknight sure thing, will do 21:46
moritz I'm just trying a spectest with parrot HEAD + r39937 reverted 21:47
moritz wishes for a faster machine 21:48
bacek working of faster parrot
good morning #parrot
moritz not bad either ;-)
Infinoid morning bacek
pmichaud moritz: good idea, I'll try it as well. 21:49
bacek Infinoid: hi!
moritz t/spec/S03-operators/arith.rakudo .............................. Method 'succ' not found for invocant of class 'Method' 21:52
meh.
pmichaud is that the only failure you're seeing? 21:53
if so, that's different from what I'm seeing in r39965
moritz it's the first
pmichaud right
moritz I'm not through yet
pmichaud with 39965, I see *every* test fail with that message.
moritz the next I'd expect is S04*/fail.t
Whiteknight when did that error start? was it 39965? 21:54
pmichaud which error?
purl which error are you getting?
Whiteknight pmichaud: the error where every test fails
pmichaud It starts with 39937
39936 results in just a few tests failing. 39937 results in every test failing.
but mainly in the fakecutable -- the perl6.pbc version works fine. 21:55
(still a few failing tests with the perl6.pbc, though)
shall I summarize this to a mailing list post, or is irclog sufficient for now? 22:03
22:04 athomason joined
Whiteknight i would put it on the mailing list 22:04
we need to figure out exactly where this error is coming from, and put a test for it into Parrot's suite 22:05
moritz with 39937 reverted on top of current HEAD it fails even more tests (5 vs. 3 or 4) 22:10
22:10 kid51 joined
chromatic How about against r39964 instead of HEAD? 22:11
pmichaud I think I tried that and it failed, but I'll try again. 22:12
oh, that was aimed at moritz
bacek bah. ExceptionHandler.mark dosn't call Continuation.makr
mark
So, we probably loosing continuation user in r39937
pmichaud moritz: with unpatched HEAD you don't see failures on every test? 22:13
bacek (And probably contexts allocated by Continuation as well)
moritz pmichaud: no, just on 3 or 4 files 22:14
pmichaud moritz: 64 bit?
purl 64 bit is much faster for me than 32 bit
moritz pmichaud: 64 bit.
pmichaud weird.
what os/distro?
moritz Debian stable, gcc-4.3 22:15
NotFound bacek: on some test I've been doing this test, looks like the contrary, it doesn't free enough contexts
s/test/days
bacek NotFound: it's likely problem with Sub.invoke
chromatic I'm sure it's Sub.invoke 22:16
nopaste "bacek" at 114.73.168.179 pasted "Patch for pmichaud++" (12 lines) at nopaste.snit.ch/17193
NotFound moritz: same here
bacek pmichaud: can you test with patch from nopaste 17193?
pmichaud bacek: you want me to apply that to r39965 ?
bacek pmichaud: r39965 is HEAD? 22:17
NotFound I've also tested a C++ build, and also doesn't fail
pmichaud bacek: yes.
bacek pmichaud: than yes
pmichaud bacek: that patch doesn't look right to me. 22:19
bacek pmichaud: why?
purl Left field.
pmichaud bacek: none of the other PMCs call SUPER() from their mark() function. 22:20
it's not clear to me why this one should.
I take that back -- Coroutine does.
as does Eval 22:21
okay, I'll try the patch.
22:22 davidius joined
chromatic Some PMCs should. 22:22
pmichaud CallSignature does not (and perhaps should -- it extends Capture) 22:23
chromatic If it marks everything its parent might mark, it doesn't need to. 22:24
bacek chromatic: it doesn't...
chromatic Then there's another bug. 22:25
pmichaud I'm curious about the comment in coroutine.pmc:259
Whiteknight it is probably worthwhile at this point to go through all core PMC types and double check the mark routines
pmichaud Whiteknight: that's what I'm doing right now.
/* co->ctx marked in SUPER(), so do not mark here */ 22:26
bacek chromatic: I take this back.
pmichaud ...how do we know that co->ctx gets marked in SUPER() ?
Whiteknight actually, this might be a good opportunity to hack Pmc2c to automatically generate mark VTABLEs for all GCable ATTRs
pmichaud nm, I know how we know.
the variable names were throwing me off.
FileHandle.mark doesn't call SUPER(), although FileHandle extends Handle. 22:28
Looks like Handle has nothing to mark, though. 22:29
bacek pmichaud: there is no Handle.mark.
ok, time for $dayjob.
pmichaud bacek: with your patch applied, I still get failures in all spectests. 22:31
(r39965)
bacek yak... 22:32
pmichaud I still suspect your patch is needed, though.
bacek Why ExceptionHandler.clone clears custom_mark flag???
Whiteknight ...that doesn't sound right 22:33
but when do exceptionhandlers get cloned?
bacek I have no idea. 22:34
But EH.clone is very suspicious.
pmichaud Proxy.mark doesn't call SUPER but extends Class 22:35
er, PMCProxy.mark, that is 22:36
22:36 rg1 joined
pmichaud PMCProxy.mark and Class.mark look very duplicated -- the only difference is that PMCProxy.mark doesn't mark any longname 22:36
bacek pmichaud: Rakudo's ObjectRef doesn't mark cached_type
pmichaud bacek: looking.
bacek: I'm guessing that's a fossil -- it doesn't appear to me that cached_type is ever used. 22:38
bacek pmichaud: ok.
Looks like whole PMCProxy.mark can be removed. Class.mark will dtrt
pmichaud it's initialized to PMCNULL, but nothing else in the code appears to make use of it.
(cached_type is initialized to PMCNULL)
bacek pmichaud: line 1385 22:39
pmichaud ...line 1385 of ?
bacek objectref.pmc
SET_ATTR_cached_type(INTERP, SELF, type_obj);
pmichaud bacek: yes, that's the line I'm referring to.
It's the only place that it's used at all. 22:40
and type_obj is PMCNULL
bacek ah. ok
pmichaud (line 1377)
bacek found it.
pmichaud in all likelihood I'll remove cached_type from ObjectRef.
oh, wait 22:41
bacek someone sneaking behind and set cached_type?
pmichaud oh, nm. It's just the generated code items.
no, nothing else uses cached_type. If I get rid of it then I suspect everything works fine. 22:42
I'll try it to confirm, though.
bacek ok, definitely time for $dayjob. 22:44
see you soon.
pmichaud so PMCProxy and ExceptionHandler are the only two PMCs that have suspicious marks() 22:45
(that I can find)
Whiteknight I'm sure there are more that are subtly wrong 22:48
but getting rid of obvious lousiness is good too
22:57 mikehh_ joined 22:58 jrtayloriv joined 23:20 brbrooks joined
bacek_at_work hi again 23:22
purl oh, you're back!
mikehh hi bacek_at_work 23:30
dalek ose: r64 | Austin++ | wiki/CloseIntro.wiki:
Edited wiki page through web user interface.
 23:33
mikehh BTW I am failing all 3 subtests in t/01-sanity/07-isa.t in rakudo make test which might be indicative of a lot of the other failures make spectest 23:36
pmichaud are they failing with the weird typenames? 23:37
mikehh it's a very simple sanity test checking for "Str" "Num (TODO) and "Code" 23:39
pmichaud try "make t/01-sanity/07-isa.t" 23:40
I bet you see that it's checking for "Sts", "Nun", and "Codf"
dalek kudo: 544038f | pmichaud++ | src/pmc/objectref_pmc.template:
Remove obsolete (and unmarked) reference to cached_type. bacek++
 23:41
mikehh if $string.isa("Str") { say "ok 1" } else { say "not ok 1" } 23:42
pmichaud right
but what's the output of running the test?
oh, wait, it won't show up there
never mind
mikehh not ok 1
pmichaud right
when it actually gets run, something in Parrot changes the "Str" to be "Sts" 23:43
mikehh the other make test tests pass - just that one fails .isa seems to the problem 23:44
pmichaud right
try this:
mikehh to be
chromatic Sounds like a missing COW for what should be a constant string.
pmichaud parrot/parrot -G perl6.pbc t/01-sanity/07-isa.t
chromatic: whenever I've followed the traces, *something* ends up calling increment on the constant string. 23:46
mikehh all 3 ok
pmichaud that matches what we see with the "Method 'succ' not found ..." messages -- something is calling increment that probably shouldn't be.
mikehh: try again without the -G option
parrot/parrot perl6.pbc t/01-sanity/07-isa.t 23:47
mikehh still all ok
pmichaud mikehh: and try directly from the command line "./perl6 t/01-sanity/07-isa.t"
mikehh all not ok 23:48
fails
pmichaud okay, that matches the same sorts of things I've been seeing. 23:49
kid51 mikehh Were you the person who originally contacted me about File::Path make_path? 23:50
mikehh kid51: Yes I seem to remember that
kid51 Are you okay with the resolution of the TT on that issue? 23:51
mikehh kid51: Yes - it works fine now
kid51 Could you make a note about that in trac.parrot.org/parrot/ticket/755 ? Then, I'll close the ticket. Thanks. 23:52
mikehh kid51: will do

Powered by Cro and the Raku® Programming Language.

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

Thank you!