Parrot 0.6.0 "P&P" released | Please mentor for SoC | parrotcode.org/
Set by moderator on 7 April 2008.
tetragon accidentally confirms that she still needs her config patch 00:00
Coke kid51: (hurm. that actually does give you a bit more than you want, but is easily processable.)
"rt show -f content ticket/52710/attachments/174160" is closer. 00:01
diakopter strawberry and cygwin smokers up and running. now what? 00:16
<sigh> I suppose MSVC is worth a try 00:22
particle msvc compiles fine, and is easy to set up 00:23
64bit however, hates msvc
diakopter particle: so which would you prefer I smoke? XPx64 or 32 00:28
particle we know xp x64 is broken, but it's the build, not the tests
32 would be helpful, just to have consistent, frequent smokes
wknight8111 i'll run an xp smoke tonight. I don't think i can set up a consistent one though 00:30
dalek r26920 | chromatic++ | trunk: 00:53
: [OO] Optimized isa() vtable entry for Class. Instead of delegating most of its
: work to isa_pmc(), this entry now performs its work directly. This is slightly
: faster and avoids creating a temporary PMC.
: The Perl 6 build process is now 19.78% faster with this patch, and all tests
: still pass.
diff: www.parrotvm.org/svn/parrot/revision?rev=26920
01:06 Theory joined, mire__ joined
kid51 We're in bad shape: I cannot complete 'make' on either Linux or Darwin. 01:12
See: rt.perl.org/rt3/Ticket/Display.html?id=52726 01:13
See: rt.perl.org/rt3/Ticket/Display.html?id=52680
pmichaud kid51: what svn revision are you at? 01:17
kid51 26917
pmichaud hmmm 01:18
r26910 introduced a bug in building PGE, but I thought I fixed it
kid51 and, needless to say, I do 'make realclean' religiously. In fact, the sandbox on Parrot is brand-new.
I think the Linux failure (TGE) occurred at 26896. 01:19
pmichaud which one is linux and which is darwin?
aha 01:21
I fixed the build problem in r26918... you probably need to svn up
kid51 The Darwin failure occurred sometime after r26892: my last successful build attempt on Darwin.
pmichaud: At 26920, 'make' completed on Linux. 01:33
It will take me a while to re-test Darwin. 01:34
pmichaud kid51: okay, good. So we should be able to re-close 52726, yes? (since the problem you observed was unrelated to the one that originated the ticket)
oh, sorry, the other one (52680)
got them backwards :-)
01:38 particl1 joined
tetragon kid51: Make succeeded on my laptop 01:38
At least svn does
I'm currently on r26919 01:39
kid51 All tests succeeded on Linux at 26920 01:42
tetragon So far I've only seen the usual crashes 01:44
kid51 pmichaud: Can you post in rt.perl.org/rt3/Ticket/Display.html?id=52680 what action you just took -- and then re-close it?
tetragon: What do you mean: the "usual crashes"? How are you running 'make' such that you see crashes? 01:46
tetragon There are seven tests that usually end up with a segfault 01:47
(One of the seven occasionally doesn't crash) 01:48
And as for how I am running make, I use it in the usual manner 01:54
kid51 But what tells you that they segfault?
tetragon The crash reporter
kid51 And what is that? 01:55
purl well, that is what he is using :)
tetragon And then manual confirmation with parrot
kid51 purl he is a she
purl ...but he is <reply>Never heard of the guy....
tetragon OS X has a crash reporting daemon that grabs stack traces
kid51 What is it called? Run from GUI or CL?
tetragon It's run by default 01:56
kid51 (Not that I would know what to do with a stack trace ...)
tetragon Things that crash with my UID get a GUI window about it, things that don't are merely logged
kid51 BTW 'make' succeeded on Darwin for me
tetragon make succeeds for me if I apply my config patch 01:58
It's just in the tests
Coke it's todo tests that fail with a segfault. 01:59
neh?
tetragon Some
Coke kid51: your build failure was fixed by pmichaud, wasn't it? 02:00
Coke catches up and sees this conversation happened. 02:01
tetragon t/dynoplibs/myops.t test 3 is a non-todo test that both succeeds and crashes
Coke ah. presumably printing all the output it needs before doing so. 02:02
Coke needs to reboot 02:03
kid51 tetragon: t/dynpmc/gdbmhash.t is still hanging for me, preventing me from completing make test 02:04
tetragon kid51: That test passes without hanging for me 02:05
kid51: I can only test on 10.5 (ppc) right now
02:05 particle joined
pmichaud kid51: (close 52680) sure thing, will do 02:12
02:22 jenny20 joined, wknight8111 joined
kid51 needs to reboot 02:29
dalek r26921 | pmichaud++ | trunk:
: [pct]:
: * Add symbol_defaults method to PAST::Block, to simply setting
: default 'scope' attribute for PAST::Var nodes.
diff: www.parrotvm.org/svn/parrot/revision?rev=26921
02:47 Theory joined 03:06 afbach joined
dalek r26922 | pmichaud++ | trunk: 03:06
: * I think r26670 was a little over-eager in adding an 'elems' method
: to ResizablePMCArray (along with shift/push/pop/unshift), as
: 'elems' is fairly Perl 6 specific. So, I'm removing that
: method.
diff: www.parrotvm.org/svn/parrot/revision?rev=26922
03:08 konobi joined
dalek r26923 | pmichaud++ | trunk: 03:12
: * Forgot to commit the test change as well.
diff: www.parrotvm.org/svn/parrot/revision?rev=26923
r26924 | pmichaud++ | trunk:
: * Eliminate trailing space in source.
diff: www.parrotvm.org/svn/parrot/revision?rev=26924
r26925 | rgrjr++ | trunk: 03:17
: * examples/pir/hanoi.pir:
: + Improve tower representation, add symbolic names, use lexical
: scoping more fully, update doc. This now serves as an example of
: stack ops, and is the only (so far) example of lexical scoping.
diff: www.parrotvm.org/svn/parrot/revision?rev=26925
r26926 | rgrjr++ | trunk: 03:22
: + Oops; forgot to check for trailing whitespace.
diff: www.parrotvm.org/svn/parrot/revision?rev=26926
konobi whats available to do for novices? 03:27
Infinoid build it, play around, write a simple perl6 program perhaps 03:41
konobi well... i terms of contributing i suppose 03:43
PerlJam konobi: tests, documentation (checking,writing,etc.), coding (what languages do you know?) 03:45
konobi perl, c, c++, plenty inbetween 03:46
PerlJam you could check the rt queue and see what piques your interest 03:47
Infinoid what platform are you on?
konobi OSX
Infinoid well, one thing you could do is help to make the Parrot sources c++-compilable 03:48
there are several places where g++ barfs on it, at present.
konobi could have a quick jookie... git++ 03:49
4 minutes to clone all of parrot
Infinoid :) 03:50
PerlJam git++ indeed
Infinoid to attempt a build with g++: perl Configure.pl --cc=g++ --link=g++ 03:51
04:01 Theory joined
konobi meh... parrot's currently using a bunch of deprecated dynamic loader functions on OSX, so that seems to be a reasonable stumbling block 04:18
05:04 Psyche^ joined 05:35 Psyche^ joined 05:48 iblechbot joined 06:16 UltraDM joined 06:22 uniejo joined
mj41 taptinder is back, tt.perl6.cz/ .. need port watchdog to linux ... some test hang in r26917 07:16
07:39 purl joined 07:55 c9s joined
spinclad pmichaud, jonathan: re pugs feature roadmap: see svn.pugscode.org/pugs/docs/01Overview.kwid, /roadmap/. (the version # asymptotic to 2pi thing.) 08:20
09:23 iblechbot joined 09:30 wknight8111 joined 10:16 tetragon joined 10:54 ruoso joined 10:55 cognominal joined 11:23 askie joined 11:32 wknight8111 joined
dalek r26927 | fperrad++ | trunk: 11:40
: [docs]
: since r26888, tools/build/revision_c.pl was removed
diff: www.parrotvm.org/svn/parrot/revision?rev=26927
11:47 Coke joined
pmichaud spinclad++ # pugs roadmap, thanks! 11:59
12:00 wknight8111 joined
Coke boggles at pmichaud's cheery 7 am send. 12:01
pmichaud I'm running way ahead of schedule this morning on getting kids to school :-) 12:18
12:38 wknight8111 joined
Coke Tested tcl's timings again today. optimized build, run of t/cmd_expr.t is down to 41.47s today from 52.547s on 3/12. about 21% improvement. 12:40
this with virtually no tcl changes in that time. 12:41
(which sounds good, but we're down to only 1382 TIMES slower than tclsh. =-)
so, keep prodding chromatic. =-)
pmichaud will do. 12:54
13:00 rotty` joined 13:07 gryphon joined 13:11 Ademan_ joined 13:20 afbach joined 13:21 ambs joined
ambs is there any way to shut up Piper? 13:21
:-S
Coke ignore him.
or patch him. =-)
piper source?
piper code?
piper/ 13:22
ambs Coke: forget it :D
purl ambs, I didn't have anything matching it :d
Coke piper?
purl i heard piper was the logbot for #parrot
particle i think robrt posted the code somewhere....
ambs I'll try to make colloquy ignore it 13:23
particle damn, the airlines just keep folding 13:25
pmichaud frontier? 13:30
purl hmmm... frontier is a Mac/Win scripting system. or costs, like, money now. or Frontier Communications, a rather large telecom provider at frontieronline.com/ or Frontier::RPC
13:31 skids joined 13:33 jan joined
particle yep, frontier went ch11 13:35
betheny's currently in denver, and needs to fly them home tomorrow
pmichaud cnn's article makes it sound as though they did it mainly for cash flow protection, and not because they're having any operational difficulties
particle yep. 13:36
Coke were they the ones whose credit card company said "hah, we're going to slow down your cash flow"
particle unlike sunjet, aloha, and ata
pmichaud coke: yes
13:37 lichtkind joined
pmichaud the american airlines groundings have really hit a lot of folks I know 13:37
particle i bet. big hub in your neighborhood
lichtkind seen chromatic 13:38
purl chromatic was last seen on #parrot 2 days and 8 hours ago, saying: I'll take temporary victories until we can fix the real things. [Apr 8 22:18:34 2008]
13:47 rdice joined 14:04 Andy joined
Infinoid racing sailboats & 14:21
14:36 IllvilJa joined 15:37 sjansen joined 15:42 Senaka joined 15:44 Senaka left 16:08 Theory joined 16:16 iblechbot joined 16:24 AndyA joined 16:25 wknight8111 joined 16:59 cotto_work joined 17:00 rdice joined
rotty seen barney? 17:13
purl barney was last seen on #parrot 20 hours and 37 minutes ago, saying: Why isn't there simple a reserved hash in the Parrot namespace ?
17:26 davidfetter joined
wknight8111 "make clean" is terminating for me with errors 17:41
pmichaud you may need to realclean and/or rebuild the makefiles 17:43
there have been a number of makefile/config changes lately
wknight8111 okay
i wont submit a bug about it then
pmichaud I generally do realclean before committing to a bug ticket
just to make sure 17:44
wknight8111 i didn't want to have to make realclean, configure, make, and make clean to test it, i wanted to ask first
pmichaud makes sense. :-)
Coke having just done that tango, I see no errors. 17:45
particle make sense 17:46
make: *** No rule to make target `sense'. Stop making sense.
pmichaud misses the "make love" target.
I still remember SVR4 coming back with "Not war?" in response to "make love" :-) 17:47
17:56 peeps[work] joined
cognominal I thought that was some early bsd replying "Not love?" to "make war" 17:57
davidfetter iirc it was how pmichaud recalls 17:59
pmichaud I'm sure I was on SVR4 at the time. One of the old AT&T 3B2 machines
jonathan Well, I guess we have make xmas in Rakudo. :-)
cognominal yes, apparently it was inpired by the standard answer : make war Make: Don't know how to make war.
particle oh, sigh. 3b2 18:00
rotty karma rotty 18:02
purl rotty has karma of 2
cognominal gnole++ 18:04
btw, lambda ultimate gives an interesting link : research.swtch.com/2008/04/computin...-labs.html 18:14
shorten cognominal's url is at xrl.us/bi99j 18:15
Coke seen ambs 18:19
purl ambs was last seen on #parrot 4 hours and 56 minutes ago, saying: I'll try to make colloquy ignore it
18:27 barney joined 18:32 ambs joined
dalek r26928 | chromatic++ | trunk: 18:53
: [PMC] Cached fully-qualified name in Class PMC for all PMCs that have a
: fully-qualified name -- that is, Class PMCs in a namespace. This improves the
: Rakudo-building benchmark by 17.53%. Amusingly, it speeds up the isa vtable
: entry in Class by much more.
diff: www.parrotvm.org/svn/parrot/revision?rev=26928
rotty does parrot do automatic stack save/restore for $P<N> registers? where can I read about this issue? 18:54
pmichaud depends on what you mean by "stack save/restore" 18:55
if you mean "when calling another sub", then yes -- the registers in a called sub are different from the ones in the calling sub
rotty ok 18:56
barney: did you ask about the licensing issue yet? 18:58
or perhaps someone else knows wether BSD-licensed code can go into parrot? 19:01
Coke he just asked on list. 19:04
Allison is pretty good about answer the licensing stuff pretty quick.
Coke wonders about the tcl code he has checked in under languages/tcl. 19:05
rotty barney: ignore my question, just saw your post 19:07
dalek r26929 | bernhard++ | trunk: 19:08
: $(RM_F) uses $(PERL), so there must be a $(PERL).
: cd compilers/tge && make clean should work again.
diff: www.parrotvm.org/svn/parrot/revision?rev=26929
19:09 sjansen joined
barney actually seem to be more 'public domain' licensed, than BSD-licensed 19:10
$_ is riaxpander
rotty indeed 19:11
rotty should have taken a closer look in the first place 19:12
dalek r26930 | bernhard++ | trunk: 19:21
: #52720: [TODO] Update documentation of the compreg opcode
: Courtesy of Patrick Michaud.
diff: www.parrotvm.org/svn/parrot/revision?rev=26930
r26931 | bernhard++ | trunk: 19:24
: #52720: [TODO] Update documentation of the compreg opcode
: Point to PCT tutorial instead of giving outdated info about 'compreg'.
diff: www.parrotvm.org/svn/parrot/revision?rev=26931
19:24 gryphon joined
pmichaud picasaweb.google.com/patrick.r.michaud/Lego # what Matthew and I built for Christmas 2007 19:26
barney rotty: Artistic License 2.0 licensed code can relicensed as GPL, but GPL-only-licensed code is not compatible with Artistic 2.0 19:31
19:40 Dave joined
rotty barney: ack 19:44
jonathan wow, that's quite some lego...
rotty how would one implement an optimization (before emitting PAST) that lifts some constants (e.g. for true/false, which are global constants in eclectus ATM) into registers in a sub? or will parrot (eventually) implement such optimizations itself? 19:46
barney Does anybody know the difference between ChangeLog and compilers/imcc/ChangeLog ? 19:50
19:50 ambs joined
rotty (is that even possible at the PAST level?) 19:53
ambs purl: seen merlyn 19:57
purl merlyn was last seen on #moose 21 days and 22 hours ago, saying: ... methodsandmessages.vox.com/library/...ltalk.html [Mar 20 15:09:20 2008]
19:57 ambs left, ambs joined, ambs left, ambs joined
jonathan rotty: You mean, only do one lookup of the constant to use many times in the sub, rather than looking it up each time true and false are used? 19:57
chromatic++ # visible difference in time it takes to run the rakudo tests 19:59
rotty jonathan: exactly 20:01
pmichaud rotty: at present PAST doesn't have a way to do that, although I'm working on ways to get there 20:02
jonathan: (re message to p-p about super)... why would the 'assign' opcode we discussed need super? 20:03
ambs wonders how gettext is detected in the configure script
jonathan pmichaud: Take an Int, for example. Normally you call the PMC's vtable assign_pmc method. 20:04
That stashes the value.
Oh, hang on a minute...
Yeah, why would it. :-)
It all made sense. Last night. 20:05
pmichaud :-)
rotty: (constants) Whatever we do should probably be applicable to things like 'self', and I think it may also be related to handling parameters in for/while loops. 20:08
ambs t/op/string_cclass.t 2 512 11 2 7 9 20:19
t/op/string_cs.t 15 3840 50 15 31-34 36-46
t/op/stringu.t 4 1024 25 4 20-22 25
failing tests
purl failing tests are the early warning long before you experience other breakage. or things that expose the behaviour you want
ambs kind of strange 20:22
pmichaud ambs: what platform? 20:35
ambs pmichaud: linux
pmichaud ambs: do you have icu loaded?
ambs configure detected it
version 3.8 20:36
pmichaud I suspect it's icu-related then
I'll load icu on my system... one moment
jonathan has, to show for his efforts to get assign and copy working so far, a bunch of failing tests and a Parrot segfault to show for it 20:37
pmichaud progress++ 20:38
ambs lol
jonathan Actually, they are the same ones I think that failed when I tried to use the copy op before... 20:39
pmichaud ambs: hmmm... with icu on my box the string* tests all pass. So perhaps that's not it. 20:40
ambs pmichaud: what icu version, btw?
pmichaud 3.8
jonathan Ah. set_pmc in FixedPMCArray assumes that if the thing it's passed to assign does array, it can treat it as if it's an instance of FixedPMCArray. Which...it ain't.
ambs ok, probably not icu 20:41
or a bad icu on my system 20:42
(make realclean)
20:47 Theory_ joined
ambs All tests successful, 23 tests and 570 subtests skipped. 20:53
ambs needs to get used to make realclean before complaining
Coke ... or we should get used to fixing the build deps. =-)
ambs hehehe
how do I build rakudo, btw? 20:54
jonathan cd languages/perl6 && make
ambs jonathan: thanks. I though there was a rule on the top-level makefile 20:55
jonathan oh, make perl6
ambs *shame*
jonathan I'm not used to doing that one. :-) 20:56
ambs jonathan: shame on me :D 20:57
all tests successful 20:58
purl Time for a well deserved vacation, ambs!
ambs purl: thanks
purl de rien ambs
Coke bueller? 21:02
purl Um, he's sick. Coke's best friend's sister's boyfriend's brother's girlfriend heard from this guy who knows this kid who's going with the girl who saw Ferris pass out at 31 Flavors last night. I guess it's pretty serious.
ambs :-/
tcl: t/cmd_lsort.t 1 256 21 0 ?? 21:03
Coke ambs: known issue.
ambs k 21:04
Coke caused by the inferior runloop issue.
konobi who's hacking parrot on OSX?
Coke me, kid51, more.
konobi anyone on 10.5?
ambs just tests parrot on OSX
Coke not I.
this about the deprecation warnings?
ambs tests on 10.5.2, bw 21:05
konobi dydl stuff?
Coke mm.
there's an open ticket about that, I think.
ambs is runing the test suite right now
Coke: any other language I should test?
Coke "make test" in languages? 21:06
ambs ok
Coke (that gets quite a few.)
(but not perl6 or... pheme?)
ambs perl6 was tested manually already (all tests successful)
konobi Coke: developer.apple.com/releasenotes/De...index.html 21:07
shorten konobi's url is at xrl.us/bjah2
jonathan pmichaud: Well, this is a tad odd. I just changed :pasttype('copy') to :pirop('assign') and without putting any of the assign vtable overrides in place, all the sanity tests pass. 21:09
That's with a local patch to FixedPMCArray 21:11
Coke jonathan: are you sure rebuilt? 21:12
"it"
konobi: danke.
jonathan Yup. I'm removing my patch to fixed PMC array and seeing a bunch of failures in the sanity tests.
Then putting it back in and they pass.
konobi Coke: bitte schone 21:13
ambs konobi: btw, my make test result under 10.5.2
Test Summary Report
-------------------
t/examples/library.t (Wstat: 256 Tests: 4 Failed: 1)
Failed test: 3
Non-zero exit status: 1
Files=568, Tests=11308, 490 wallclock secs ( 4.27 usr 3.69 sys + 196.94 cusr 78.19 csys = 283.09 CPU)
konobi i was just looking at compiling with g++ last night for obvious stuff and the dydl stuff is beyond my knowledge realm 21:14
Coke konobi: added that to the ticket. 21:15
(40826: Mac OS X and Dylib Funcs
ambs probably that library.t failure is related to that dylib ticket. 21:18
konobi if compiling with g++ it just bails compilation at that point 21:19
ambs is compiling with the default compiler detected by parrot 21:22
nopaste "ambs" at 77.54.92.255 pasted "Summary of errors on languages make test (linux)" (31 lines) at nopaste.snit.ch/12679 21:24
dalek r26932 | jonathan++ | trunk:
: [core] Fix segfault in FixedPMCArray's set_pmc vtable method. It would assume that something that does array was either a FixedPMCArray or something with the same structure; a bad assumption if you subclassed it with a high level object.
diff: www.parrotvm.org/svn/parrot/revision?rev=26932
konobi git++ # 3.5 mins to get full history clone 21:25
ambs smokes 21:26
ambs smokes on a mac as well 21:30
jonathan Aha. The reason some things fail with copy, is because they don't clone themselves properly... 21:42
pmichaud: Got a full set of passing tests using copy instead of morph/assign now. 21:57
Lots of cleaning up to do.
pmichaud jonathan++ 22:35
jonathan pmichaud: Just about to check in, switching us over to using copy. 22:38
Not 100% sure about it, but it doesn't break anything, and certainly makes .WHAT print the correct results.
22:39 skids joined
jonathan Well, it's in, see what you think. 22:39
dalek r26933 | jonathan++ | trunk: 22:40
: [rakudo] Switch us over to using the copy opcode rather than the morph/assign sequence we had been doing up to now. This makes .WHAT give the correct answers.
diff: www.parrotvm.org/svn/parrot/revision?rev=26933
pmichaud if copy doesn't work, we'll file tickets to get someone to make it work :-) 23:03
afk, dinner
jonathan ;-)
I'll let the dust settle from this before looking into moving over to overriding the assign vtable. :-) 23:04
time for sleep for me now
pmichaud sounds great, excellent work (as always!) 23:12
23:25 darbelo joined 23:37 tetragon 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!