#parrot Parrot 0.6.1 "Bird of Paradise" Released | parrotcode.org
Set by moderator on 22 April 2008.
Infinoid japhb: would using the HUGEINTVAL type help you? that looks to be declared as "long long" on linux/i686, and "long" on linux/amd64... 64 bits in both cases 00:04
japhb yawns
Infinoid the problem is, NCI will have to return and pass those arguments from the rest of parrot, so we need to use a type that the rest of parrot can handle
japhb Infinoid: that would be good, if NCI were taught to use it
Infinoid so I'm trying to find a suitable type :) 00:05
japhb How is HUGEINTVAL currently used?
Infinoid kinda sporadic 00:07
japhb Still, it does sound like the right typedef, if NCI can be taught to use it. I nominate 'L' for this task. 00:09
Infinoid stares at call_list.txt 00:11
... is this really what I think it is? we have to declare all our function signature types ahead of time? 00:13
00:14 Theory joined
japhb Infinoid: yes, that's exactly what it is. It's ugly, but eventually will go away. I just have no idea how far away that will be. 00:21
Supposedly eventually the NCI thunks will be JIT on demand 00:23
I get the feeling that has a low priority (as in "may not make 1.0")
particle japhb: end of summer, i hope 00:24
japhb particle: oooh, better than I thought! 00:25
Infinoid japhb: well, so far this is an interesting hack (meaning, I had to go as deep as pobj.h). I'm curious whether the rest of parrot is up for this
particle this is the subject of a GSoC project
Infinoid japhb: I might have something for you to test (I don't have a clue how) in half an hour 00:26
particle by a long-time contributer to parrot
japhb Infinoid: "this"? You mean adding HUGEINTVAL to NCI?
Infinoid yeah
with an L signature
japhb ah, thank you!
excellent
Infinoid don't thank me yet, it might not work, and it might not be what you want :)
japhb particle: I remembered seeing something NCI related in the GSoC list, but "end of summer" is too far away to affect my immediate plans. :-) 00:27
Infinoid: well OK then.
particle tewk plans on using languages/c99 to parse headers and generate nci stubs
Infinoid ooo, h2pmc 00:28
or h2pir?
japhb Once you have the parsing correct, the output should be way easier.
particle on a different topic, i'm having a load of failures on ubuntu x64 00:32
112 failed subtests
Infinoid All tests successful on gentoo amd64 00:33
any pattern to your failures?
amazingly enough, nci.c compiles 00:35
particle nothing i can immediately identify
lemme nopaste the test output, before i head off to get cleaned up and make dinner 00:36
Infinoid I'm not so convinced I can add another basic type to the PObj union this easily...
something's gotta break.
nopaste "particle" at 24.19.3.148 pasted "test failure summary for ubuntu 8.04/x64" (12 lines) at nopaste.snit.ch/12839 00:37
japhb Infinoid: and also note that chromatic posted a patch a bit ago that messes with PObj. There will probably end up being a conflict there.
Infinoid probably. but this is just a proof of concept, I don't actually have the foggiest idea what I'm doing :) 00:38
basically I'm just too naive to have learned not to mess with this stuff yet. 00:39
particle just watch the size of the object in the union
it had better be the same as the size of the other objects
shower &
Infinoid sounding good so far. its 64 bits, and the other objects are guaranteed to be at least that big
at least, it won't be any bigger. it may be smaller, but that won't throw off the union size 00:40
"All tests successful." *blink* 00:43
nopaste "Infinoid" at 75.32.77.88 pasted "ugly patch to add HUGEINTVAL types to NCI (and add it to everything else too, I hope)" (173 lines) at nopaste.snit.ch/12840 00:44
Infinoid I seriously doubt the rest of parrot is up to this. Might need a HugeInteger PMC or something to actually manipulate these values in the rest of parrot 00:45
but the NCI side of things might work now :)
japhb Infinoid: in tools/build/nativecall.pl , in the chunk starting with '/L/ && do {', should that specify "long long t_$temp_num;" or "HUGEINTVAL t_$temp_num;" ? 00:52
(I honestly don't know which is supposed to be used there) 00:53
Infinoid I just followed the nomenclature that "l" used 00:54
japhb ah.
japhb trusts the force for now
RT that patch! ;-)
Infinoid it seems to use entries from the "compiler-vague" family of types, on the external side, and entries from the "parrot-vague" family of types, on the parrot side 00:55
japhb interesting
Infinoid and hoping they match up. which they will, for now, because of some other weirdnesses
japhb hmmm, perhaps in the longer term, L == long long, H == HUGEINTVAL
Infinoid neither of those are really fixed 00:56
japhb Sure, but I would guess they cover the vast majority of times you would want a large int type 00:57
For NCI purposes, I only care about L right now ....
Infinoid well, what worries me is, there are 128-bit processes out there, which means "long long" will probably eventually get bigger 00:58
s/processes/processors/
unless they want to get ridiculous with a long long long type
"really really long"
"bond james bond"
japhb Infinoid: that's true, but I would expect that 'long long' => 128 bits on such a processor would actually be a good thing. 00:59
Infinoid I'd rather see an alternate set of platform-indignant fixed size types available
like uint32_t, Parrot_UInt4, that kind of thing
p5's pack() and unpack() are nice that way
japhb I'm all for that, especially since some APIs define and use such fixed width types 01:00
AFK for a bit
Infinoid some do, some don't, and we have to support both
ok, thanks
steak & 01:11
01:13 tewk joined 01:19 wknight8111 joined 01:31 tewk joined 01:41 Theory joined
wknight8111 seen chromatic? 02:10
purl chromatic was last seen on #parrot 4 days and 23 hours ago, saying: I'm tracking down a similar crash in one of the Rakudo tests. [Apr 22 20:13:47 2008]
02:22 contingencyplan joined 02:37 particle joined 03:15 particle joined 03:23 tetragon joined 03:55 Theory joined 04:29 Psyche^ joined 05:35 markus joined 06:11 uniejo joined 07:02 iblechbot joined 07:49 slightlyoff joined 09:03 contingencyplan joined 09:35 ruoso joined 09:52 rotty` joined 10:31 wknight8111 joined 11:35 Zaba joined
dalek allison@perl.org | Concurrency Tasks: 12:32
link: www.perlfoundation.org/parrot/index...ency_tasks
12:34 itz joined 12:36 iblechbot joined
dalek allison@perl.org | Concurrency Tasks: 12:41
link: www.perlfoundation.org/parrot/index...ency_tasks
itz is there anything like a simple tasks for parrot/perl6 n00bs wikipage? 12:42
Coke parrot has the cage tasks. moment. 12:48
itz: www.perlfoundation.org/parrot/index..._to_parrot 12:52
er.
www.perlfoundation.org/parrot/index..._to_parrot
dalek allison@perl.org | Concurrency Tasks: 12:53
link: www.perlfoundation.org/parrot/index...ency_tasks
Coke I can't see the link of a list of cage tasks, will recreate it...
cage?
purl it has been said that cage is <url:www.speech.cs.cmu.edu/~sburke/stuff...3.html>
Coke parrot cage?
many perl.org sites not coming back. 12:54
So, start at the wiki page and we can go from there. =-)
12:55 itz joined 12:57 allison joined, Juerd joined 13:30 paco joined 14:09 Zaba_ joined
Coke this is the only perl.org service I have up at the moment. 14:25
14:29 mire joined
Coke seen ask? 14:32
purl ask was last seen on #perl 14 days and 20 hours ago, saying: hex: :-) [Apr 13 11:51:32 2008]
Coke seen robrt?
purl robrt was last seen on #p5p 11 days and 12 hours ago, saying: xdg: Exactly. Thats what someone was complaining about [Apr 16 19:53:27 2008]
Coke .win 8 14:33
pmichaud Coke: I don't think that putting hackathon work into branches "sends the right message" 14:34
Coke Could you be more specific? 14:35
pmichaud if I'm a hackathon participant, and I do anything of note, I'd like to see my work accepted into the trunk quickly
I don't want to see it shunted into a branch because of an upcoming release (that could conceivably be moved to avoid conflicting with the hackathon) 14:36
particle i've been thinking about changing the release process
so we create branches/QA, test it for a few days, and release from there 14:37
pmichaud I'm looking at it from an "instant" versus "delayed" gratification perspective
particle then trunk becomes current development *only*
and any changes made to branches/QA are rolled into trunk after release
pmichaud particle: why after release? 14:38
I'd rather roll them directly into trunk as I locate them
particle sure, that's fine too
pmichaud in that case I don't need the branch :-)
particle i mean META.yml, etc
14:39 desertmax joined
particle bugfixes found during testing can be rolled directly into trunk 14:40
pmichaud I guess I don't see the advantage of the branch
particle release prep activities shouldn't affect trunk, as trunk is never a release candidate
pmichaud if the release process took a couple of days, I could see it. But that's not generally the case. 14:41
or, put another way, I'm not sure what existing problem you're trying to improve
particle i believe as we approch 1.0, the release process will lengthen
the existing problem is this:
14:41 kj joined
particle i'm in hawaii, and only have access to a win32 laptop 14:41
i want a quality release
i can't test it well enough to ensure quality, but i release anyway 14:42
some days later, i get failure reports from cpan testers.
i don't like that release cycle.
pmichaud are the failure reports likely to be ones that would be caught by a "release candidate" phase?
particle if we have a branch for release candidates, we can even submit developer releases to CPAN for platform testing 14:43
pmichaud hmmm, that might work
having the branch is useful if we do indeed get multiple testers for the candidate
particle yes 14:44
we can have smokers that look for changes to the branch
...and prioritize those requests over trunk smokes
Coke (RC phase) arguably, every release we are doing pre-1.0 doesn't even get as far as RC in terms of stability.
pmichaud: I have gotten no feedback from committers to indicate that branches are considered second-class. 14:45
particle coke: yes, that can be argued. but i argue that the only way to increase velocity is to ensure high stability
Coke have you? or are you trying to address a potential problem?
pmichaud Coke: the people at hackathons aren't committers
(often aren't)
Coke sorry, "contributers"
particle a few small scripts under tools/ could help us manage our branches better 14:46
make it easier for contributors unfamiliar with svn to deal with them, etc
pmichaud coke: I'm looking primarily at new contributors, not existing ones
Coke but do you *know* this is an issue from previous anecdotal evidence on parrot? or are you trying to avoid a problem that you think *might* exist? that's what I'm trying to get a handle on here. 14:47
And i think we can address that level of expectation socially.
pmichaud avoid a problem I think *might* exist, coupled with knowing that past hackathons haven't netted as many new long-term contributors as we might like 14:48
Coke but we don't have feedback from those that didn't come back that using a branch was the issue, right?
pmichaud we weren't using branches then (also, we've never had a hackathon that coincided with a release date)
i.e, the reason we may not have observed this is because hackathons never juxtapositioned with releases as this one will 14:49
particle chicago hackathon in november 2006 was within a week of a release date
Coke I would rather address this by being more socially interactive with our new contributors.
particle many patches were submitted to rt and applied after the release
pmichaud well, I don't want to argue this too strongly, because it's really not that important to me. 14:50
if I were a new participant, I'd feel really good about having my work immediately go into the trunk 14:51
I'd feel less good if it went into a branch but didn't really have to
I'd feel unappreciated if it went into a branch and then got forgotten (much like a warnocked patch)
14:51 Andy joined
Coke no one said we would be putting *everything* into a branch. I was responding specifically to your concern about things that would impact the release. 14:51
pmichaud I'm just asking why release dates are sacrosanct in this instance. 14:52
but rather than push it further, I'll just drop it. it's really not that important.
14:53 gryphon joined
Coke (drop it) ok. 14:53
particle pmichaud: is running Test.pm still on the priority list for rakudo? 14:56
pmichaud particle: *which* Test.pm ? 14:57
particle pugs
something that gets us todo/skip
pmichaud I thought that fudge handled that for us
particle it helps
i'm just thinking about old assumptions for milestones 14:58
pmichaud I'm not terribly impressed with pugs' Test.pm design -- it seemed like a lot of stuff was thrown in
I'd rather come up with a cleaner Test.pm
particle agreed
wonderful. i agree.
pmichaud I'll write up rakudo milestones shortly -- have one or two other morning tasks to take care of first 14:59
jonathan Will do my first full Rakudo day on Thu or Fri.
pmichaud jonathan++
Coke continues to wait for his parrotcode.org commit to work.
pmichaud I'll be out of town Fri and Sat
particle jonathan++ 15:00
Coke (it's taking a LONG time to time out.)
jonathan ok, better for you if I make it Thursday?
pmichaud if you can (and you think you'll want my input on things), then yes.
jonathan Well, your input is good to have. :-)
Will look at that inheritance bug someone posted on list.
particle t\\02-test-pm\\1-basic.........< postcircumfix colonpairs not yet implemented at line 41, near ", :desc('2"
jonathan Oh, I can do that too, if you like. ;-) 15:01
cognominal jonathan, what are your chance to come to fpw at the moment?
particle what's wrong with :desc('foo') ??
jonathan :desc('foo') should work...
Hmm
particle yeah. hmm.
jonathan :desc<foo> I know isn't done
particle ah.
pmichaud :desc<foo> should be really easy
particle then the error message is bad.
jonathan cognominal: Erm, I really need to work out how much it's going to cost me.
pmichaud: Agree, easy, just undone. 15:02
cognominal: The location is a tad unusual = more costly to get to.
cognominal You need to fly to Toulouse
pmichaud afk for a bit
cognominal and you can ride with BooK in a car, see the wiki
jonathan Is it's airport big? I'd hate tou-louse myself there... 15:03
Coke *groan*
jonathan cognominal: What dates should I be flying on?
The Nordic workshop want me too.
cognominal fpw Friday 30 May and Saturday 15:05
particle jonathan is the (unofficial) european embodiment of rakudo :) 15:06
jonathan cognominal: OK, so if I fly into Toulouse on the Thursday and leave from there on the Sunday, will that match up with the lifts?
cognominal conferences.mongueurs.net/fpw2008/w...oVoiturage # BooK is proposing a carpool from Toulouse de Albi if you arrive at the airport Thursday before 6pm
this year is on a tight budget so I can't ask the organizer to pay part of your trip :( 15:07
organizers 15:08
purl organizers are sincere enough.
cognominal indeed
Jonathan: and the organizer are quite short on talks, so you probably do as many as you can/will :) 15:10
jonathan cognominal: I'd very much like to be at both this and the Nordic one, I will try to make it happen. 15:15
15:15 sjansen joined
jonathan Sorry for not having an answer yet - my energies have been eaten up by finding a place to live from next month. 15:15
cognominal :)
you are welcome any time... 15:16
Coke would donate 5 bucks if you had a paypal link up somewhere.
I imagine there are more people who would do the same.
particle apply for a microgrant 15:17
$500 will buy you a hamburger in london, i hear.
Coke wonders if the microgrants are going through the GC.
jonathan?
Coke smacks purl.
purl Oh baby, you do it so *good*!
Coke jonathan, what's your email address? 15:18
cognominal well, they have better than burgers in Toulouse
jonathan y'know, it's funny hearing people saying about the pound being strong agains the dollar. :-)
I instead am having trouble with the weakness of the pound...aginst the euro.
Coke: jnthn@jnthn.net
particle seems search.cpan.org is down
jonathan particle: Microgrant could work.
particle no, microgrants are outside TPF GC
jonathan I know someone has got a travel grant before. 15:19
cognominal we have people like the Kerviel trader who works hard to lower euro currency :)
particle there are 5 open microgrants for the taking, and yes, they can sponsor travel
jonathan Given I've just had one grant given to me, I feel kinda odd about going chasing another one. 15:20
"All your grants are belong to me."
It would mean I could comfortably go to both events though. 15:21
Coke goes for 5 EU instead.
Coke urks at the exchange rate. 15:22
cognominal jonathan : what we can do is to auction you talks :) 15:23
but you will have to give one of your colorful t-shirts as a bonus :)
PerlJam jonathan: Do you believe that getting an additional grant will take that grant away from someone who is just as likely as you to do something productive? :) 15:24
jonathan PerlJam: There's plenty of productive people in the Perl community besides me. :) 15:26
PerlJam jonathan: sure, but how many of them are applying for a grant?
jonathan PerlJam: Given that there's five of them left now...
Seems not many.
cognominal jonathan : can you auction one of your t-shirts? 15:27
jonathan Coke++ # thanks! :-O 15:28
cognominal: Erm, maybe...
Most of my t-shirts are *from* Perl conferences. ;-)
As my mum frequently points out to me.
"ANOTHER Perl t-shirt?!"
PerlJam jonathan: Make your own via cafepress.com (put rakudo features on them or your picture or something) 15:30
cognominal no, we want one of his very colorful tshirts :) 15:31
jonathan hehe
cognominal: Will see what I can do.
Would people *really* want one of my t-shirts? ;-)
cognominal well, that's a pretext... 15:32
tewk How do people report Perl Foundation funding to the government? As a grant like an academic grant or as self employment income? 15:35
jonathan tewk: I've been doing it as the second of those. 15:37
particle i've been doing the latter, as well
however, it's a good question.
jonathan And highly dependent upon the country you're in too, I guess. 15:38
cognominal jonathan, I forgot your email address
jonathan cognominal: jnthn@jnthn.net 15:40
cognominal: Flying to Toulouse looks a lot more expensive, than taking a morning flight to Paris and getting the train down to Toulouse. 15:43
cognominal well but then we have to take a train to albi, that's seven hours of train. 15:44
jonathan From Toulous to Albi?
*Toulouse
cognominal no, from paris
jonathan Sure, it is. 15:45
cognominal I probably will take the train too from paris
jonathan Aha, maybe I can join you for the train journey? 15:46
Coke jonathan is jnthn@jnthn.net
cognominal sure
and one hour subway from Roissy to Gare Monparnasse to take the train.
jonathan What sort of time were you thinking of setting off on the Thursday? I can get a flight that arrives in Paris (ORY) at 08:50 15:47
cognominal maybe if we are two, and that the pooling car is not full we can manadge them to meet us at the Toulouse railway station 15:48
jonathan OK, that sounds like a plan too.
cognominal This is the Toulouse-Albi which is not fun.
jonathan And you would get train back on the Sunday after the conference?
cognominal From Paris to Toulouse it is TGV, a fast train
I must see with BooK on #perlfr if he agrees 15:49
your plane will land at Roissy or Orly? 15:50
15:51 iblechbot joined
jonathan cognominal: Orly 15:54
Infinoid YARLY
cognominal ok 15:55
I think we will be early at Toulouse so the best will probably to get a cab to Toulouse Blagnac airport (or some public transportation) to meet BooK. 15:56
jonathan OK. 15:57
particle has a three line patch for postcircumfix:<> colonpairs
jonathan particle++
15:57 davidfetter joined
particle wonders what it would take to get the :P5 regex modifier working 16:00
jonathan particle: I think just specifying the :language of the regex block differently.
erm, and parsing the adverbs to the regex. ;-)
particle right 16:01
jonathan OK, I'll put a microgrant application in for travel to the FPW and NPW.
davidfetter anybody have word on YAPC::NA? 16:03
particle that's a cryptic question 16:04
davidfetter well, like are all the talks accepted that will be?
particle yapcchicago.org
dalek r27217 | particle++ | trunk: 16:05
: [rakudo] implement postcircumfix:<> for colonpairs
diff: www.parrotvm.org/svn/parrot/revision?rev=27217
particle conferences.mongueurs.net/yn2008/schedule
davidfetter well, that answers my question 16:07
davidfetter a little disappointed not to have gotten any word on this from them :P
particle cognominal: you have a patch pending for rakudo hash composers, correct? 16:08
cognominal yes
I mean a function. 16:09
particle ok, i'd like to be able to pass isa_ok({ 'one' => 1 }, 'Hash');
cognominal I have not worked out, the grammar.pg that calls the hash() function
I will rewrite in nqp my parse trimmer first. 16:10
particle cool.
cognominal the perl awk like was plain ugly 16:11
jonathan particle: The => syntax is broken (in the parsing) at the moment, suggest sticking to colonpairs.
cognominal btw, coke, I have filed a bug about your perl script that generates template for languages 16:12
particle jonathan: i'd like to pass some existing tests
good to know => needs to be fixed, though
however, looks like we also need to implement rx//
particle checks STD.pm for that
i see a clear pattern in the failure results from spectest: rakudo needs list context 16:14
jonathan particle: If you have specific requests for stuff for me to look in to on Thursday, => jnthn@jnthn.net :-) 16:31
particle ok($foo.isa(Foo), '.isa(Foo)'); 16:35
is that valid perl 6?
looks like a bareword to me 16:36
guess it's a type
jonathan It's fine so long as Foo is declared. 16:39
It's a typename.
particle yeah. doesn't parse in rakudo
jonathan $foo ~~ Foo already works (or should)
particle Statement not terminated properly at line 23, near "($foo.isa("
t/spec/S12-class/basic.t
Infinoid particle: t/dynoplibs/myops.t passes for me on mingw, after my commit from saturday. do you have a chance to test it on msvc? 16:43
particle not yet. will do now
Infinoid ... well, I have a failure on the "one alarm" test, but not the "three alarms" test I modified. 16:44
oh, duh. its skipped under mingw, so my test was ineffective 16:47
after unskipping, it runs, but doesn't look like alarms ever fire under mingw 16:48
(that's why the "one alarm" test fails, too.)
... it also looks like "sleep 1" doesn't actually sleep. 16:51
particle on win32, sleep is guaranteed to be within 1sec of the requested time 16:52
dunno if that's same for mingw
but that's why sub-second times aren't greta 16:53
*great
Infinoid mingw doesn't sleep at all, which might explain why both of the alarm tests fail
I'm going to send that to parrotbug
particle: doing a ceil() on a float value is sufficient, if the API doesn't allow anything more precise 16:54
as long as sleep returns at or after the point you specify, that's fine... no guarantees beyond that
particle mingw sleeps for me 16:56
Infinoid pir sleep, or C sleep()?
particle perl sleep
perl -e
Infinoid ok. "sleep 1" in pir seems to be a noop 16:57
16:58 Theory joined 16:59 mire joined
Infinoid I'll see what I can do to fix it 16:59
17:01 particle[ventus] joined, grim_fandango joined
particle[ventus] is having trouble with ubuntu 8.04/amd64 17:02
Infinoid particle[ventus]: I saw your list of failed tests, any chance of seeing the stderr or prove -v or somesuch too?
17:02 grim_fandango joined
particle[ventus] sure, i'll post that 17:02
Coke robrt has fixed the websites; yapc::na link added to the front page. 17:03
17:03 grim_fandango joined
nopaste "particle" at 24.19.3.148 pasted "bt for perl6 -e "say 'Hello, world.'"" (39 lines) at nopaste.snit.ch/12842 17:03
particle[ventus] runs make test 17:04
17:04 cotto_work joined
Infinoid sleep() doesn't exist at the C level in mingw, I'm gonna find out how perl 5.10.0 does it 17:05
particle vaguely remembers something about 'delay'
rillian I'm still working through kj's tutorial, part 3 17:07
which is just assignments 17:08
my variables make it into the parse tree
but my values don't
the string literal ends up wrapped in a MySquaak::Grammar' => "Hash[0x587d4]"
particle rillian: have you seen languages/squaak in the parrot tree?
rillian and the integer literal ends up an undef 17:09
particle: I have, but I can't figure out what I'm doing wrong
pastebin.ca/1000571 and pastebin.ca/1000570
if anyone wants to have a look
cotto_work have you seen www.parrotblog.org/2008/04/solution...cises.html ?
rillian yes. I'm stuck in the part that says "I assume you don't need any help with this." :) 17:10
particle heh
i haven't read the tutorial yet. can you point me to the chapter you're on?
rillian in my integer_constant action, printing ~$/ (or +$/) give the correct value 17:11
but my make PAST::Val.new(...) doesn't seem to get passed up the tree
www.parrotblog.org/2008/03/episode-...first.html
although www.parrotblog.org/2008/03/solution...ode-3.html might be a better place to start 17:12
17:12 rdice joined
Patterner 27217 still cannot resize the FixedIntegerArray... 17:12
Infinoid isn't that why we call it Fixed? 17:13
particle[ventus] now sees All tests successful, 29 tests and 603 subtests skipped. 17:14
particle[ventus] tries rebuilding with -j
Patterner well, it complains... so I assume it's not A Good Thing(TM=). 17:15
Coke I vaguely remember this conversation. Is there a ticket, Patterner?
Infinoid We secretly replaced particle[ventus]'s bogons with cluons. Let's see if he notices.
Coke cluons are like gluons, only denser. 17:17
rillian what's the phase diagram of a frak-cluon plasma?
particle[ventus] wonders if make -j realclean caused the misbehavior 17:19
as make -j test returns all tests successful now, too
perl6 still segfaults :( 17:20
Infinoid I've been using make -j realclean quite a lot in the last week or so, with no ill effects
particle well, one bogon freed. but there are more. 17:21
Coke make -j realclean should be safe.
particle hrmm, apparently my mingw setup is now failing to find gcc :( 17:22
Coke uses it often, much quicker.
Zaba particle[ventus], tried running the .pbc instead of the perl6 executable? 17:24
Infinoid the perl6 fakecutable works fine for me on gentoo/amd64 17:25
Zaba it fails horribly on some things here, while .pbc does not
Patterner Coke: not yet... usually my bugs get resolved before I have a chance. 17:26
Infinoid I can't reproduce particle's nopaste.snit.ch/12842 crash :(
Patterner hmmm... "Configure.pl" works. "Configure.pl --optimize" does not work.
particle[ventus] particle@ventus:~/dev/parrot/trunk$ ./parrot languages/perl6/perl6.pbc -e"say 'Hello, world.'"
Hello, world.
purl rm -fr /
Coke removes some old release announcements from the front page.
particle[ventus] zaba++
japhb ce.
particle[ventus] Infinoid: you're running on the same hardware as me, right? 17:28
Zaba Infinoid, seems like I'm not the only with such a problem, then..
particle[ventus] gcc version 4.2.3 (Ubuntu 4.2.3-2ubuntu7) 17:29
This is perl, v5.8.8 built for x86_64-linux-gnu-thread-multi
Zaba gcc (GCC) 4.1.2 20070214 (Gentoo 4.1.2 p1.0.2)
Infinoid particle[ventus]: lenovo T61? yes. 17:30
particle[ventus] zaba: perl 5.8.8?
Zaba particle[ventus], yes.
Infinoid gcc version 4.2.3 (Gentoo 4.2.3 p1.0)
This is perl, v5.10.0 built for x86_64-linux-thread-multi
Zaba Infinoid, ricer! *hides*
particle[ventus] crap. now i gotta go build perl 5.10.0 17:31
is there an rpm somewhere?
Infinoid hmm... that's a long, difficult process. I spent about a day tracking down all the stuff that had to be rebuilt for 5.10 17:32
particle[ventus] yeah, hence the "crap".
Infinoid I'm wondering if there's a Better Way
Zaba Infinoid, perl-cleaner? 17:33
Infinoid Zaba: missed a lot of stuff.
I've got 5.8.8 on a linux/i386 box, one moment
particle[ventus] i wonder if it has to do with icu or another config option 17:34
Infinoid I can paste my config_lib.pasm, if you want to do a diff
or my Configure.pl output 17:35
particle[ventus] config_lib.pasm would be interesting
nopaste "Infinoid" at 96.238.213.50 pasted "gentoo/amd64 config_lib.pasm" (388 lines) at nopaste.snit.ch/12843
pmichaud 16:09 <cognominal> I have not worked out, the grammar.pg that calls the hash() function 17:38
grammar.pg doesn't call the hash() function
that will be done from the 'noun' method in actions.pm 17:40
or possibly in circumfix
cognominal well I need to work out a composer in grammar.pg/actions.pm that will call the hash() function.
in fact it must check if it is a block or a composer 17:41
pmichaud right -- that will be done in either circumfix or noun
cognominal anyways &hash is useful without the syntactic sugar 17:42
pmichaud correct.
purl no, it's not!
pmichaud particle: is that patch for <postcircumfix> correct? Looks like it might be missing an array subscript somewhere 17:44
cognominal bbl
particle[ventus] Infinoid: you have readline defined. that's all i can see that's different
pmichaud: maybe i botched it... i'll have to look in a few 17:45
Infinoid ok. install libreadline5-dev and try again? 17:46
(I doubt readline is the cause, but I haven't tested much without it) 17:47
cotto_work chromatic++ for working around the bug exposed by his mmd example 17:49
particle[ventus] yeah, i'm doing that now
17:52 Piper joined
Piper Hi there. I am Piper. I am now publicly logging this channel. If you don't want to be logged, please leave now. 17:52
cotto_work thanks, Piper 17:53
17:53 ambs joined 17:54 Piper joined
Piper Hi there. I am Piper. I am now publicly logging this channel. If you don't want to be logged, please leave now. 17:54
ambs Piper, go f...
ambs watches his tongue.
pmichaud ambs: yeah, I bet Piper logged that. :-P 17:55
ambs :D
pmichaud if you want to say something bad about Piper, be sure to do it when it's not looking (and make sure Tweety isn't watching either)
ambs I have it more when it opens a new tab just to warn me :) 17:56
Coke You can ignore those, neh? 17:57
ignore piper?
ambs I *hate*
ambs checks the documentation
17:59 grim_fandango joined 18:01 cotto_work joined 18:05 jalbo joined
jalbo Hello. 18:05
Infinoid hi! 18:06
dalek r27218 | pmichaud++ | trunk:
: [rakudo]:
: * Change ROADMAP to provide a list of intended milestones.
diff: www.parrotvm.org/svn/parrot/revision?rev=27218
jonathan ...ok, that was fast, I have the grant already. 18:07
FPW and NPW, here I come. :_)
ambs jonathan, PPW, no?
jonathan, or PtPW, if you preffer...
jonathan PPW? 18:08
purl PPW is the Pittsburgh Perl Workshop pghpw.org or perl paddy wagon
Coke SMOP?
purl it has been said that SMOP is a Simple Meta Object Programming or Simple Matter of Programming at www.perlfoundation.org/perl6/index.cgi?smop
pmichaud I'm planning to attend PPW
jonathan Oh, argh, I'm meant to be meeting someone five minutes away, two minutes ago...gotta run. 18:09
ambs PTPW 18:10
PTPW?
purl PTPW is probably the Portuguese Perl Workshop (6-7 June, 2008)
ambs that one :)
dalek r27219 | pmichaud++ | trunk:
: [rakudo]:
: * Update to ROADMAP milestones.
diff: www.parrotvm.org/svn/parrot/revision?rev=27219
18:15 Ivatar joined
particle Zaba: i'm sorry, i've been way to distracted to give your code a look. i'll try to get to it when i can 18:21
pmichaud zaba: what's the code/issue? I might be able to take a look. 18:22
jalbo Someone can try this patch buliding with c++? rt.perl.org/rt3/Public/Bug/Display.html?id=53352
particle ack, i even messed that up
it's rillian with the code issues
pmichaud: you should see the issues in backscroll 18:23
pmichaud I scanned but they didn't pop out
18:23 AndyA joined
rillian pmichaud: still stuck debugging tutorial 3. pastebin.ca/1000571 and pastebin.ca/1000570 18:24
pmichaud rillian: okay, just a sec (finishing use.perl post)
rillian if I compile 'x = 1' I get bind(var('x'), undef) in the parse tree
nopaste "particle" at 24.19.3.148 pasted "earlier posts from rillian" (18 lines) at nopaste.snit.ch/12844 18:25
rillian and 'x = "one"' turns into bind(var('x'), val('Hash[0xaddr]'))
I've not been able to figure out what I'm doing wrong
18:29 grim_fandango joined
Coke ponders moving all the Tcl tickets out of parrot's bug queue. 18:29
Coke thinks this would be for the best.
pmichaud Coke: perhaps launchpad?
ambs rt is so damn slow :( 18:30
Coke ambs: use the command line tool. Much faster.
pmichaud: i was thinking googlecode. 18:31
ambs Coke, is there any command line tool for RT?
Coke (bug tracker, eventually svn repo for when we get booted out of perl.org...)
rt command line?
rtcli?
purl rtcli is, like, RT CLI
Coke rt cli?
purl i guess rt cli is www.perlfoundation.org/parrot/index...mmand_line
pmichaud rillian: can you post the output of --target=past ?
Coke rt command line is RT CLI
18:31 mj41 joined
particle coke: i have a shiny empty repo created for parrot hlls 18:31
ambs googles
particle currently populated with none
ambs coke++ 18:32
particle if you'd like to start using it for tickets, i think we can make that happen
Coke depends on where it is.
particle googlecode 18:33
purl rumour has it googlecode is google's sourceforge replacement
Coke which project?
purl which project is this?
particle squawk.googlecode.org
Coke wget: unable to resolve host address `squawk.googlecode.org'
particle s/org/com/
ambs purl, forget which project
purl ambs: I forgot which project
pmichaud rillian: on line 63 of actions.pm, you have
#
make $/{$key}; 18:34
it probably needs to be
make $($/{$key});
rillian pmichaud: pastebin.ca/10000697
pmichaud otherwise it's putting the parse node into the tree instead of the PAST node
rillian er pastebin.ca/1000697
particle coke: hopefully, we can implement our own columns, like HLL or something 18:35
Coke particle: that's exactly the kind of wonkery I'm trying to avoid. =-)
pmichaud files a ticket to generate a more useful error for rillian's case
rillian pmichaud: I did try that in my punctuation-fiddling phrase. "Method 'get_scalar' not found of invocant of class 'Undef'" 18:36
particle coke: wcoleda@gm...?
Coke yes
particle you're now a project owner
Coke I'll see if mdiep has a strong preference on whether or not to share.
pmichaud rillian: okay, then that means that something is returning 'Undef' for expression
rillian pmichaud: that does make the string literal work though, so you're on the right track
pmichaud then it's a problem with integer_constant
rillian how to I print the class of a variable? 18:40
pmichaud could try $var.WHAT
but I'm not sure that works yet.
dalek r27220 | pmichaud++ | trunk: 18:41
: [rakudo]:
: * Another ROADMAP update (modules and versions)
diff: www.parrotvm.org/svn/parrot/revision?rev=27220
particle c:\\usr\\local\\parrot\\clean\\languages\\perl6>..\\..\\parrot perl6.pbc "say 1.WHAT" 18:42
Error: file cannot be read: say 1.WHAT
pmichaud particle: need a -e
particle that's a *very* misleading error
er, right.
Int
pmichaud particle: yes, but this isn't rakudo rillian is working on
particle yes, i know. 18:43
pmichaud I don't know if .WHAT is supported by Protoobject or NQP yet
particle it's not.
pmichaud I will undoubtedly add it. Sooner if someone files a ticket. :-)
rillian: you're going to hate me for this, but....
particle it's another feature to backport from rakudo
pmichaud in grammar.pg, line 50 18:44
| <integer_constant> {*}#= integer_constant
has an extra space at the end of the line
so the key is coming in as "integer_constant " and not "integer_constant"
particle ooh! Protoobject.pir *does* support .WHAT 18:45
pmichaud oh. Then .WHAT should work on PAST::* nodes
particle i'm so far behind in parrot these days :(
yes, indeed.
rillian pmichaud: no, I am *very* grateful you noticed that. because I never would have.
time to make my editor ugly on such things, I guess 18:46
pmichaud I guess PGE should strip trailing blanks when figuring out #= keys
particle rillian: do you use vim?
rillian nano
particle has yet to find someone that uses kate
pmichaud
.oO(...someone uses an iPod for parrot development? wow!)
rillian "smart whitespace matching, except in magic comment strings!" 18:47
pmichaud: :)
particle has a command in vim to trim trailing whitespace 18:48
" remove trailing whitespace
map ,ts :%s/\\s\\+$//<Enter>
imap ,ts <Esc>:%s/\\s\\+$//<Enter>a
rillian yay, now it works! thanks much, all of you. 18:49
pmichaud thanks for being patient with us :-)
particle[ventus] ok, how do i set up ccache? 18:53
particle[ventus] reads the manpage 18:54
Coke do you want it to just run for parrot, or for everything?
cotto_work ./configure --cc='ccache gcc' 18:55
perl Configure.pl
Juerd (note: ccache is default on feather)
particle[ventus] coke: iunno, what do other folks do?
nopaste "pmichaud" at 76.183.97.54 pasted "my rebuild script (w/ccache)" (9 lines) at nopaste.snit.ch/12846
rillian would find distcc more useful if it were possible to sanely cross-install cross-compilers. 18:56
particle[ventus] pmichaud: not make -j?
18:57 mj41 joined
pmichaud I haven't seen that make -j makes things significantly faster for me 18:57
and I don't want to have to worry about whether the current parrot is -j capable or not 18:58
I have other bugs to chase. :-)
rillian er, just so I understand, having extra whitespace at the end of the key didn't stop my integer_constant method from being called, because *that* name matched the rule name, but when my expression method called make $($/{$key}), it did so with a $key that didn't exist, so we got an undef instead of the Past::Val object the integer_constant method had set. Is that right?
pmichaud rillian: exactly right. 18:59
rillian ta
pmichaud and Undef doesn't understand get_scalar yet
Coke in general, don't fear the -j anymore. =-) 19:00
(but yes, avoiding borkage in parrot is good when you're working on a language.)
rillian if only I'd debug-printed $key with quotes around it like I did the actual constant
Coke particle[ventus]: (ccache) on feather, I just use gcc which is symlinked to ccache. 19:02
on my mac, for historical reasons, I invoke ccache directly and pass the compiler as an arg.
feather's setup is much easier.
Infinoid just adds /usr/lib/ccache/bin/ to his PATH in his /etc/profile 19:04
Juerd I wanted a shared ccache initially
But that's problematic with file access permissions.
Never could get that to work well.
Infinoid hmm. could probably merge everything into the same user using distcc, and then ccache that... but is it worth the overhead? 19:05
particle[ventus] well, i'm running on my laptop, and don't expect to be building much other than parrot
so, i'll likely use the link approach
Infinoid particle[ventus]: do you have a /usr/lib/ccache/bin/ full of symlinks?
particle[ventus] /usr/lib/ccache is full of links 19:06
Infinoid you should be able to just add that to the front side of your PATH 19:07
ambs mine configuration
[ambs@rachmaninoff SEPLN2008]$ ls -l /opt/ccache/
total 32
lrwxr-xr-x 1 root admin 19 6 Abr 20:32 c++ -> ../local/bin/ccache
lrwxr-xr-x 1 root admin 19 6 Abr 20:32 cc -> ../local/bin/ccache
lrwxr-xr-x 1 root admin 19 6 Abr 20:32 g++ -> ../local/bin/ccache
lrwxr-xr-x 1 root admin 19 6 Abr 20:32 gcc -> ../local/bin/ccache
:D
ambs is a spammer
jalbo Did you sell replica ccaches? 19:08
Infinoid r3p l1c a
ambs hehehe
Juerd, but yes, for feather a common ccache would be great 19:09
dalek r27221 | coke++ | trunk: 19:10
: [tcl]
: Begin movement of tcl tickets out of RT queue to simplify things for parrot.
diff: www.parrotvm.org/svn/parrot/revision?rev=27221
Juerd ambs: Maybe feather needs a sysadmin wiki, with its todo list 19:11
Infinoid TODO: add todo here
Juerd Right 19:12
I'm adding it to my personal todo list
It's probably worth the tuit investment.
ambs doesnt use feather, for now
19:12 particle[ventus] joined
jalbo feather? 19:12
purl feather is feather.perl6.nl/ or a perl6 community development server
particle[ventus] how do i tell gnome-terminal to run my .profile every time?
Juerd particle[ventus]: You don't. You tell your *shell* to do that.
Infinoid run it from .bashrc or .zshrc or whatever 19:13
Juerd particle[ventus]: e.g. echo 'source ~/.profile' >> .bashrc
particle[ventus] that doesn't seem to do what i want 19:15
Coke and then you have to re-exec bash to get it for this session.
or, "what do you want that that doesn't do?" =-) 19:16
particle[ventus] if i open a new terminal after doing that, i get a blank screen
if i hit ctrl-c, it gives me a prompt 19:17
Coke can you nopaste your .bashrc and your .profile?
is this on feather?
particle[ventus] no, it's on ventus, my laptop 19:18
Juerd Coke: I would be surprised if people used gnome-terminal on feather. 19:20
Coke oooh: event.perlbelarus.org/minsk2008/
jalbo Maybe your .profile exec your .bashrc, some distros do that by default. 19:21
particle[ventus] indeed, .profile contains #include .bashrc if it exists 19:22
Coke freaky.
particle[ventus] still, setting PATH="/usr/lib/ccache:$PATH" in .profile doesn't work 19:23
Infinoid put it in .bashrc then
Coke you'll need to export it, too.
particle[ventus] no need to export. works from .bashrc. 19:24
Infinoid I'm not so sure... you will need to export if you want GNU make to run the same thing your shell does
jalbo PATH is always already exported, unless you make some unusual.
Infinoid ok :)
Juerd jalbo: If you change it, you get a local non-exported copy afaik 19:25
Coke wonders how much of his unix experience is cargo culted from 12 year old aix and xenix, and 10 year old solaris. 19:26
jalbo Juerd: if you change it in a subshell, but in a .profile is not the case.
Juerd Aha
jalbo Coke: I believe that Unix is a cargo cult, entirely. 19:27
rillian my favorite definition of a scripting language is that you don't ever actually know what you're doing, you just cut-and-paste from something similar and twiddle the punctuation until it works 19:28
Infinoid well, that covers Java pretty well 19:29
jalbo rillian: I have seen a lot of C code written that way, also.
Infinoid hrm. win32 has a Sleep() that takes a millisecond argument. but its not in time.h, its in windows.h. well, whatever, maybe I can use that to fix mingw's sleep
particle[ventus] ok, so how do i know that ccache is working? 19:30
Infinoid particle[ventus]: rm -rf ~/.ccache, compile something, see if it was recreated
particle[ventus] it's enough for me to know that ~/.ccache now exists. Infinoid++
Infinoid you probably want to set a maximum size on the cache directory, at some point. ccache --help 19:31
it filled up my 5GB /home partition before I noticed it :P
Juerd particle[ventus]: Compile something twice with a make clean in between, and if it's much faster the second time around, then ccache works ;)
particle[ventus] i have 1 20GB partition here
Infinoid a big SVN::BinarySearch will use up a surprising amount 19:32
or App::SVNBinarySearch or whatever it is :)
Juerd It sucks when a make script regenerates a non-changed file. Then ccache thinks it has to be recompiled.
jalbo If it not works, build a longer airport until cargo comes. 19:33
Infinoid uh, no, ccache decides that based on the md5sum of the preprocessor output
Juerd Infinoid: Really? Has it always done that?
Infinoid it has for the last couple of years at least... that's how long I've been using it
Coke (App::SVNBinarySearch) holy crap, people use that? 19:34
19:34 jrockway_ joined
Coke feels guilty. 19:34
Infinoid Coke: yeah! it's a great way to fill up your ccache. :)
Coke Infinoid: HA! glad to help.
particle[ventus] is surprised that coke is still shocked about that
Juerd Infinoid: Hm...
particle[ventus] however, particle[ventus] has never used it
Coke particle[ventus]: it was basically a proof of concept someone convinced me to put on cpan. 19:35
Juerd someone++
particle[ventus] it's too stupid to use svn log, right?
i mean, to determine which revs are in the branch you're testing 19:36
jalbo Talking about cpan, is amusing when you can't update Test::Harness because his test need an updated Test::Harness 19:40
19:40 ruoso joined
Infinoid jalbo: perhaps they should add a 'use blib' to the test? 19:41
jalbo Don't know, I just forced the install and tested later.
But maybe that's a barrier for possible testers. 19:42
Infinoid yeah, probably
purl yeah, probably is something that needs doin/working on
Coke particle[ventus]: (too stupid) yes.
particle[ventus] coke: the overhead might be worth it on large projects like parrot, but probably not for small stuff 19:43
anyway, maybe someday i'll code it up. can't be too hard
Coke nope. pretty easy. 19:46
Infinoid contemplates a svn-bisect command (similar to git-bisect)
Coke Infinoid: wots that?
Infinoid same idea, bit more manual intervention 19:47
you give it some initial bounds, and then say "git-bisect good" or "git-bisect bad"
and it goes and updates to the next rev for you to try
for some reason, that fits into my brain a bit better than writing a "realclean; configure; make; test && echo 1" script
Coke Infinoid: next version of app:svnbs will not do an update, but a co or export avoiding the need for most of the infrastructure. 19:48
particle[ventus] export++ 19:49
(unnecessary metadata)--
Infinoid Coke: sounds like that would be a lot cleaner
there's always the danger of some rev having a bug that modifies a file tracked by svn, which means I've had to add a "svn revert -R ." to the cleanup section of the script 19:50
so if the script doesn't need a cleanup section, I'm all for it :)
19:51 jrockway joined
Coke obra++ 19:59
Infinoid adds a Parrot_usleep() to the various platforms
20:04 barney joined
dalek r27222 | chromatic++ | trunk: 20:05
: [src] Added STRUCT_COPY_FROM_STRUCT macro to perform struct copy when the
: source is not a pointer (NotFound, RT #53352).
diff: www.parrotvm.org/svn/parrot/revision?rev=27222
Coke obra++ 20:07
20:07 IllvilJa joined
jalbo is NotFound 20:13
20:15 Zaba_ joined 20:22 rdice joined
Infinoid ah. jalbo++ 20:49
21:05 sjansen joined 21:34 Theory joined 21:54 jrockway joined 21:56 jrockway joined 21:58 grim_fandango joined 22:03 rdice joined 22:12 AndyA joined 22:13 jalbo left
dalek r27223 | chromatic++ | trunk: 22:15
: [src] Ported more STRUCT_COPY uses to STRUCT_COPY_FROM_STRUCT (missed in
: r27222, more of RT #53352 from NotFound).
diff: www.parrotvm.org/svn/parrot/revision?rev=27223
22:18 wknight8111 joined 22:20 davidfetter joined 22:31 Limbic_Region joined
dalek r27224 | chromatic++ | trunk: 22:38
: [spec] Updated spec file; tested with Fedoras 7, 8, and 9-beta (Gerd Pokorra,
: RT #53450).
diff: www.parrotvm.org/svn/parrot/revision?rev=27224
Limbic_Region just noticed that rakudo's make spectest seems to be passing more tests on Win32/Cygwin 22:58
wknight8111 I'm getting an error for "make perl6" on WinXP. Problem linking to "MSVCR90.dll". is this a real issue, or just a problem on my system? 22:59
I don't ignore the possibility that my system is just borked from years of haphazard configuration changes
23:09 japhb joined
Tene Hm. Thinking about trying to add state vars to rakudo. 23:11
japhb Can Rakudo use modules from the 'parrot' namespace yet? 23:12
And if so, what's the current syntax?
japhb remembers back in Pugs land using 'use perl5:Foo::Bar;' 23:13
Limbic_Region wknight8111 - I do Win32/MinGW and Win32/Cygwin without issues
Tene japhb: jonathan said he was going to consider working on that, but I haven't seen any commit messages mentioning anything about it. 23:15
japhb jonathan: please please please please work on getting Rakudo to access other HLL namespaces (parrot at the least), sometime soon -- that's I think the next Big Block to me doing my OpenGL stuff in Perl 6 (since things are progressing on the Parrot side, and the triangle spins in PIR on at multiple platforms) 23:17
s/ on at / on / 23:18
wknight8111 Limbic_Region, I'm using cl.exe. I must just have a bad configuration or something 23:20
23:23 tetragon joined
jonathan japhb: You're not the first person to ask for this kinda thing...I will try and do something for you. 23:31
japhb jonathan: I very much appreciate it, with ample realization that I'm asking someone *else* to do work for something *I* want. :-) 23:32
jonathan japhb: Sure, but I'm funded to improve Rakudo, and if you can make it do pretty graphics stuff, well, that's just cool. ;-) 23:33
japhb :-)
23:39 LimbicRegion joined 23:42 kid51 joined
jonathan sleeps 23:46
dalek r27225 | chromatic++ | pdd25cx: 23:48
: [exceptions] Removed Parrot_init_exceptions() and its residue.
diff: www.parrotvm.org/svn/parrot/revision?rev=27225
chromatic@wgz.org | Concurrency Tasks: 23:54
link: www.perlfoundation.org/parrot/index...ency_tasks
shorten dalek's url is at xrl.us/bjqpy

Powered by Cro and the Raku® Programming Language.

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

Thank you!