#parrot Parrot 0.6.4 "St. Vincent Amazon" Released | parrotcode.org/ | 15 new, 648 open, 80 stalled tix | logged in irclog.perlgeek.de/parrot/today
Set by moderator on 28 July 2008.
00:09 AndyA joined 00:50 petdance joined 01:04 Khisanth joined 01:40 bacek joined 02:17 kid51 joined 02:18 cognominal joined 02:31 jennyf joined 02:35 cybergirl joined
tetragon Argh, I tried building and testing languages/lua, and all I got for my trouble was a bunch of segfaults referring to the address 0x00000000deadbef7 02:37
Hrm... Now I got a single segfault referring to a different addres 02:38
(lua is one of the places outside of the OpenGL examples that uses call_toolkit_init from NCI) 02:39
nopaste "tetragon" at 76.10.153.9 pasted "A typical 0x00000000deadbef7 crash report" (138 lines) at nopaste.snit.ch/13773 02:42
02:48 britneypire joined
japhb tetragon: I assume he's using call_toolkit_init for the same reason the OpenGL examples do ... for GLUT 03:03
or rather, glutInit
tetragon Yep
The point at which my attempt to run the OpenGL examples fail on my 64-bit build looks to be in call_toolkit_init 03:04
japhb reaches back into the depth of memory 03:05
... yep, I remember now.
tetragon Hrm... Actually, one of the comments near the failing line in call_toolkit_init is quite the hint 03:06
japhb The magic trick that call_toolkit_init uses to handle argv is an ugly kludge, because it's a workaround
tetragon # XXX: This is unportably wrong; it assumes sizeof(INT) = sizeof(PTR)
japhb YUP
I put that comment in there for a reason. ;-)
There's just no way in NCI to say "put a null pointer here"
Or at least, there wasn't back when I wrote that.
tetragon Well, guess what. 64-bit OS X is generally LP64, not ILP64
japhb tetragon: yep, I knew that. 03:07
tetragon: I forgot about it, because it works on Linux/x86_64
So I didn't remember to keep pestering chromatic about it. ;-)
tetragon (Although, in the depths of time, it or a predecessor was ILP64. There's a whole stack of #ifdefs in the platform headers that relate to that history) 03:08
japhb tetragon: ... in the 10.3 days? I know there was much incorrect typedef'ing in the GL headers from old OS X ...
But maybe that was just a coincidence 03:09
tetragon GL wasn't built as 64-bit in the 10.3 or 10.4 days
japhb ah, gotcha.
tetragon (and it was the 10.4 headers with the strange typedefs that I sent you)
japhb Still, it boggles my mind that they had the wrong typedefs. Since every other platform uses the same typedefs, known to work virtually everywhere ...
tetragon: nodnod 03:10
japhb is running at reduced memory capacity.
tetragon And it was Mach that was ILP64
japhb tetragon: how did Mach specify a 32-bit integer type? 03:12
tetragon It was easy on 32-bit platforms
It was an int
On 64-bit, well, int was 64-bit (as far as I can tell by some of the comments) 03:13
japhb Right ... so ...
japhb imagines some ugly hack like 'long short'
03:18 StephenPollei left
japhb Any git-svn users about? 03:22
03:24 olgagirl joined 03:39 tetragon joined 03:52 tetragon joined 04:10 tetragon joined
bacek japhb: I use git-svn 04:16
japhb bacek: thank you -- I think the freenode #git folks are answering my question. I'll let you know if not. 04:18
:-)
bacek ah, ok. ;)
japhb The short of their answer: 'git-svn rebase' does do a fetch, but it is not the fetch you are looking for. In fact, its semantics are basically insane. 04:19
It fetches from whatever branch is the parent of the current SVN HEAD.
so, actually need to do 'git-svn fetch' separately first.
They were right 04:26
joy
04:52 Ademan joined
Infinoid ah. I use git-svn a lot at work, but I don't set it up to track svn branches, because they are basically insane 05:05
05:18 tetragon joined 05:34 tetragon joined 05:36 Psyche^ joined
japhb Infinoid: would that I could avoid doing so myself -- but until Parrot switches completely over, we're still stuck with the SVN branch mambo. 06:35
06:36 cotto_work joined
dalek r30137 | fperrad++ | opengl4lua: 06:37
: [Lua] merge from trunk (30102 -> 30136)
diff: www.parrotvm.org/svn/parrot/revision?rev=30137
06:41 tetragon joined
dalek r30138 | fperrad++ | opengl4lua: 06:51
: [Lua] imrprove layout
diff: www.parrotvm.org/svn/parrot/revision?rev=30138
06:56 sandra_f joined
cotto_home 'bout time we had some commits 06:58
fperrad++ #not that he'll ever know 06:59
karma fperrad
purl fperrad has karma of 847
07:06 jennyf joined 08:10 Ademan joined 08:41 barney joined
cotto_home barney, ping 08:50
08:54 sandra_f joined
barney hi cotto 08:54
cotto_home yapc would be a good time to brainstorm for what to do with pipp.org, now that we have it 08:55
several bots from that addr have been connecting and disconnecting 08:57
that was more general than I'd intended 08:58
barney cotto: YAPC might be a good occasion to discuss HLL interop. Writing all those PHP functions in PIR is a pain. fperrard++ 08:59
cotto_home very much so. 09:01
barney For pipp.org, it might be good to have a 'more content is coming' page there,
cotto_home I'm hoping for a lot of headway on that tricky and interesting topic.
barney and a status page of what is done, (not much so far) 09:02
cotto_home I don't see much reason to start putting content there (except for collaborative purposes, which seem to be covered ok by #parrot and email) until pipp gets to the point where people will want to play with it. 09:05
That's just me, though.
barney Well, when talking to people it is easier to say go to pipp.org then to say there is a Wiki-page somewhere. 09:08
So, a referal to the Wiki would serve that purpose too
cotto_home a redirect to the wiki would be a good placeholder until we decide to put something more substantial up 09:10
it shouldn't be hard to find the tuits to get that up before yapc starts on Wed 09:11
barney cotto++ 09:18
I need to do some shopping and chores, bbl 09:19
cotto_home I'll be asleep by then
'night
btw, progress on releasing the phparray code is slow and there's not really a way I can apply more pressure (there's already quite a bit) 09:24
otoh, all vtable methods are implemented and fairly well tested 09:25
barney Cool. 09:26
BTW on Aug 23-24 I will attend FrOSCon, www.froscon.de. I suppose that some core PHP devolopers will be there. 09:27
cotto_home there might be some guys from the OSS lab there too 09:28
I know they're trying to hit the less-advertized conferences where the core PHP hackers get together 09:29
I also am fairly sure I won't be going. :( 09:30
barney St. Augustin is not a must-see place. I've been there for a German Perl Workshop. 09:31
09:37 masak joined
cotto_home good night 09:47
purl sleep well too
cotto_home sleep &
10:24 tetragon joined 10:43 Ademan joined 10:46 rurban joined
dalek r30139 | rblasch++ | trunk: 11:57
: [src] Apply bitmask before casting.
: Otherwise the MSVC "convert to smaller type check" may trigger.
diff: www.parrotvm.org/svn/parrot/revision?rev=30139
r30140 | rblasch++ | trunk: 12:02
: [win32] Adjust Parrot_floatval_time to standard epoch time.
diff: www.parrotvm.org/svn/parrot/revision?rev=30140
12:15 lafille joined 12:26 Infinoid joined 12:48 Ademan joined 12:51 tetragon joined 13:29 kid51 joined 13:31 Ron joined 13:40 Ron_ joined 13:55 gmansi joined
dalek r30141 | jkeenan++ | stepdescription: 14:13
: Creating stepdescription in svn.perl.org/parrot/branches
diff: www.parrotvm.org/svn/parrot/revision?rev=30141
14:14 jhorwitz joined 14:37 petdance joined
dalek r30142 | jkeenan++ | stepdescription: 15:02
: Trim all configuration step descriptions so that we can fit step name and
: description on one line as Configure.pl runs.
diff: www.parrotvm.org/svn/parrot/revision?rev=30142
kid51 svn.perl.org is misbehaving again. Failing to send successful report back to sender. Instead spending this spurious message: Transmitting file data ...................................................................svn: Commit failed (details follow): 15:04
svn: MERGE request failed on '/parrot/branches/stepdescription'
svn: MERGE of '/parrot/branches/stepdescription': 200 OK (svn.perl.org)
svn: Your commit message was left in a temporary file:
svn: '/home/jimk/work/stepdescription/svn-commit.tmp'
While 'svn up' makes things right by merging (G) files, this is still not correct behavior. Haven't seen it in several weeks, though it was endemic early in this year. 15:05
jonathan hi from Sveeeeden. 15:31
masak jonathan: ditto :)
rurban Malm�? 15:32
masak rurban: you're sending latin-1 in a utf-8 environment.
rurban hmm, chatzilla 15:33
purl chatzilla is fucked or very beautiful :)
jonathan Linkoping.
Nice town.
However, I managed to be out walking during the 15 minutes of the day where it rained heavily. 15:34
15:34 jq joined
japhb One of my few regrets about having English as my native language is unintentional i18n rudeness. (This from someone who wrote a Western-European languages website translation engine almost a decade ago.) Of course, I get mine -- I'm colorblind, and its amazing how much of the computer world assumes that red and green are obviously different .... 15:35
(That was re: masak's latin-1/utf-8 comment)
jonathan I just moved to a country where they have charcters that ain't in latin-1. 15:36
masak japhb: out of curiosity, how would you personally solve the red-green thing?
japhb masak: tone, texture, or shape are easiest. If you're very careful, *different* color contrasts.
For example, Using red triangle - yellow square - green circle works. 15:37
masak japhb: that sounds beneficial to everyone, actually
the more clues, the better
japhb masak, yep, it is, actually.
masak seems to me that encoding problems will exist for a long time. part of the problem is that there is no fundamental difference between two strings with different encodings. 15:39
japhb It turns out that a lot of the things that make computer interaction better for visually handicapped users improve speed and comprehension for everyone
masak jonathan: re $str.subst, I now have only two problems left :) 15:40
jonathan: whenever you have the time.
jonathan masak: OK, you got a patch of it so far? 15:41
masak jonathan: hold on 15:42
nopaste "masak" at 130.238.45.242 pasted "The commented out code works. The code after it doesn't." (63 lines) at nopaste.snit.ch/13775 15:44
masak so, I wrote .substr(Str,Str) first.
that was lucky, because I never really got .subst(Regex,Str) working. 15:45
nopaste "rurban" at 212.183.85.173 pasted "#56996-fhs-runtime.patch" (94 lines) at nopaste.snit.ch/13776 15:46
masak my first blocker is getting the second version to work. the other blocker is getting the two versions to run together.
jonathan match = regexp.'ACCEPTS'(self)
masak yes?
jonathan I'm pretty sure that will return a boolean and put the match object in $/
masak I sort of had that feeling
jonathan Rather than returning the match object iself. 15:47
*itself
masak is it clear what I _want_ to do? how do I do it?
jonathan See the code in Regex.pir that sets $/
In the ACCEPTS method in that file
It should be almost copy-paste to get the match object.
masak Regex.pir?
jonathan src/classes 15:48
I *think*, wait..
masak there's no such class in rakudo.
you mean Code.pir?
jonathan oh, yes
Codepir
aye.
$P0 = getinterp 15:49
$P1 = $P0['lexpad';1]
$P1['$/'] = match
masak since that's the method that throws the error, I've been there a few times already :)
so I want to inline that code, with some changes?
jonathan Is the error something like, outer has no lexpad?
masak the cause is that, yes
jonathan Or something lexical-ish?
OK
masak the error is actually "Null PMC access"
jonathan OK 15:50
I think :lex on the sub you're writing will work.
But also:
match = regexp.'ACCEPTS'(self)
If you just write
match = regexp(self)
It should work too
masak did that, got a bus error :(
jonathan Ah. 15:51
masak do I still need :lex for that?
jonathan That's certainly wrong.
Shouldn't
masak that's what I thought too
jonathan Oh, actually, if the sub you're calling looks at $/, then you may.
oh, no, this is the replace with a string version 15:52
So that should not be an issue.
masak yes.
ok.
jonathan The way that should work giving you a bus error is rather sucky, though. :-(
masak I'll try the :lex fix, and if that doesn't work I'll try not to get a bus error using regexp(self)
if that doesn't work I'll submit a tear-filled bug report to rakudobug 15:53
jonathan Yes, it seems you are close.
Try it with -G just to be sure, it's not a regex bug. 15:54
masak ./perl6 -G?
15:55 ruoso joined
masak jonathan: did you happen to catch the moment when I realized that all my 'double free' errors disappear if I use ../../parrot perl6.pbc instead of ./perl6, by the way? 15:55
don't really know what that means, but thought it might be important.
jonathan Didn't see that, no. 15:57
But it probably helps.
I never actually use the fakexecutable. 15:59
Maybe we should have a way of running the tests with it, though.
(If there isn't one already)
masak ok, I (1) added :lex, (2) got "lexical $/ not found", (3) added declaration of $/ to method, (4) got 'bus error' when running :(
jonathan :( 16:00
masak :(
16:00 cybergirl joined
masak seems what I want to do contains a bus error. 16:00
16:01 rhr joined
jonathan Yes. 16:02
Backtrace would be good to have.
masak gdb? 16:03
purl somebody said gdb was beyond abeltje's current skillset, do I need a perl with DEBUGGING?
masak jonathan: gdb?
purl i heard gdb was beyond abeltje's current skillset, do I need a perl with DEBUGGING?
jonathan Yes.
masak ok.
nopaste "masak" at 130.238.45.242 pasted "Here's what gdb had to say." (20 lines) at nopaste.snit.ch/13777 16:04
masak seems to me a rogue pointer is involved. 16:06
jonathan I think if you type "bt" and hit enter it shows us how we got there too. 16:07
masak ah.
yikes. we got here through a long and circuitous path. 16:08
nopaste "masak" at 130.238.45.242 pasted "gdb output, now with backtrace" (136 lines) at nopaste.snit.ch/13778 16:09
jonathan Wow! 16:10
masak maybe the match actually succeeded, and the error occurs in one of the substr calls 16:11
jonathan That may or may not be it 16:12
masak either that, or the regexp match makes its own substr calls
jonathan --trace 1 16:13
No, --trace=1 16:14
Second one as a parameter to perl6.pbc
masak oki
I don't know how many lines you need, but the line just before "Bus error" says ' 1607 substr S1, S0, 0, I1 S1="(null)" S0="(null)" I1=3' 16:15
there's our rogue pointers.
jonathan: hm, I'll paste the last 20 lines so you can see. 16:16
jonathan Ooh...the fun of null strings.
nopaste "masak" at 130.238.45.242 pasted "last 20 lines of ./perl6 --trace=1 -e 'my $s = "abcd"; $s.subst( /d/, "D" )'" (21 lines) at nopaste.snit.ch/13779 16:17
16:17 iblechbot joined
masak jonathan: it is at times like this that one actually starts to appreciate the Maybe monad of Haskell :P 16:17
jonathan: hm, maybe you actually need more than the 20 lines. S0 and S1 aren't mentioned until a bit further up 16:19
I'll nopaste the whole trace. it wasn't too much. 16:20
nopaste "masak" at 130.238.45.242 pasted "the whole trace" (179 lines) at nopaste.snit.ch/13780
jonathan ok, lookng 16:22
NotFound I think there was an unresolved ticket about string_substr not accepting NULL as first argument. 16:23
jonathan masak: It seems that you get the match back, and call .from and .to on it successfully 16:24
nopaste me your latest code,
masak jonathan: coming up. 16:25
nopaste "masak" at 130.238.45.242 pasted "my latest code." (32 lines) at nopaste.snit.ch/13781 16:26
masak hm... I think I see the problem. 16:28
I never declare $S0, do I?
16:29 kid51 joined
jonathan Yes, that's what I was wandering 16:29
masak seems that line of code got lost somehow.
jonathan I think you are missing a $S0 = self
masak recompiling...
it works! \\o/ 16:30
jonathan Yay!
masak jonathan++
jonathan masak++
masak ok, so how do I make it do either string or regexp replacement?
jonathan masak: Slightly off-topic, what is Swedish for "chinese", as in the type of restaurant? 16:31
masak jonathan: it's the same word, but spelt with a 'k'
'kinesisk mat' or 'kinamat'
'kinarestaurang' 16:32
jonathan gah! Google Maps only suggests places in Noorkoping! 16:33
masak (and for someone who has tried real Chinese food, I usually make a clear distincion between 'kinesisk mat' (home-made food) and 'kinamat' (which you get in the restaurants))
jonathan wants Chinese nom
In answer to your question - I'd do it with multi subs.
masak jonathan: I'm sure there's at least one Chinese restaurant in Linköping 16:34
jonathan Me too.
masak jonathan: so just declare two :multi subs one after the other?
jonathan Yup, I think they do fairly different things.
masak what's up with using quotes around subnames and class names sometimes in PIR, and sometimes not? 16:35
jonathan You generally should use them. 16:36
masak oki
jonathan I think that's the direction we've been heading.
masak oki, my patch works now 16:37
should I nopaste it, or send it via rakudobug? 16:38
jonathan Send it to the list.
Can get some more eyes on it that way.
But it looks good.
masak I'll take a look at the corresponding .t file too
I fudged it a few days ago. maybe it passes now 16:39
jonathan The regex one no doubt may want some tweaks in the future, since the :g once implemented will let us match many things, I guess requiring many replacements. but we don't have that yet.
masak jonathan: ACK 16:40
from a Perl6 user perspective, I can tell you that even a partial .subst implementation is an improvent over manually stitching strings, though
:g would be even better, of course... :) 16:41
jonathan Oh, of course. 16:42
purl Indubitably.
jonathan I'm very much up for partial implementations of stuff that take us in the right direction.
masak to be honest, I really see no alternative, apart from "think long and hard, and then suddenly emerge with a perfect solution", which seems what Knuth likes to do. he dösn't use tests, either. 16:44
jonathan Perl 6 is just too hard to have a perfect solution first time. :-) 16:45
masak we should get Knuth interested in the project and let him try :) 16:46
it'd be an "immovable object, unstoppable force" situation.
16:46 Limbic_Region joined
masak is there a way to run just one testfile, fudged? 16:55
16:57 Ron joined 16:59 tetragon joined 17:05 rurban_ joined
jonathan masak: Fudge it manually first. 17:07
perl t/spec/fudge rakudo path/to/test.t
masak jonathan: thanks.
jonathan So annoying! Lots of nice blue sky now.
masak jonathan: is there any hope of making .subst work for all scalars, not just strings?
jonathan: told ya. 17:08
jonathan masak: If it's meant to do that, then move it to str-any.pir
masak but don't worry, tomorrow and Monday will be awful. 17:09
jonathan: I don't know, isn't it?
jonathan I'm not sure. :-)
masak most string replacement functions work on all scalars.
jonathan If it's one of them it should be in any-str
masak 65536.subst(5, 7)
seems reasonable to me 17:10
dalek r30143 | julianalbo++ | trunk:
: throws on invalid null argument to string_substr
diff: www.parrotvm.org/svn/parrot/revision?rev=30143
jonathan Best to leave it just in string unless the spec is clear where it should be.
masak oki
jonathan And check on that.
masak just wondering, becuase one of the tests tries to send in an integer
or rather, a block returning an integer
oh well, it's a start
jonathan Yeah, but just because spectests do stuff sadly doesn't always mean it's right.
I've been burnt by making that assumption before. 17:11
masak heh.
17:15 particle joined
jonathan OK, time for a hopefully dryer walk, and if I'm lucky some nice food. 17:18
masak jonathan: good luck. kudos. see you. 17:20
jonathan cu, on Tuesday if not on IRC before :-)
masak yep. :) 17:24
dalek r30144 | jkeenan++ | trunk: 17:31
: MANIFEST and SKIP were not updated before a recent commit.
diff: www.parrotvm.org/svn/parrot/revision?rev=30144
r30145 | jkeenan++ | stepdescription: 17:41
: Eliminate --fatal-step=n option; allow only --fatal-step=step::class.
diff: www.parrotvm.org/svn/parrot/revision?rev=30145
17:44 s1n joined 18:12 slavorgn joined
dalek r30146 | fperrad++ | trunk: 18:23
: [Lua] now os.time() uses the float version of time
diff: www.parrotvm.org/svn/parrot/revision?rev=30146
r30147 | fperrad++ | opengl4lua: 18:30
: [Lua] merge from trunk (30137 -> 30146)
diff: www.parrotvm.org/svn/parrot/revision?rev=30147
18:45 davidfetter joined 19:37 cybergirl joined 19:39 lauanana joined 20:04 iblechbot joined 20:08 olgagirl joined 20:23 iblechbot_ joined
dalek r30148 | chromatic++ | trunk: 20:26
: [Rakudo] Fixed SVN metadata on new file.
diff: www.parrotvm.org/svn/parrot/revision?rev=30148
r30149 | chromatic++ | trunk: 20:36
: [PMC] Fixed context reference counting in Exception PMC, so that exception
: contexts don't get destroyed early but do get destroyed on time. This fixes
: some double-free errors during Parrot full destruction and probably prevents
: other weird memory errors in long running programs.
diff: www.parrotvm.org/svn/parrot/revision?rev=30149
r30150 | chromatic++ | trunk: 20:39
: [config] Made Darwin-specific memalign respect 64-bit builds (reported by
: Seneca Cunningham, RT #57536 and RT #57700).
: Tidied code.
diff: www.parrotvm.org/svn/parrot/revision?rev=30150
20:52 davidfetter joined
dalek r30151 | chromatic++ | trunk: 20:54
: [PMC] Fixed compiler warnings (use of uninitialized variables) and tided code.
diff: www.parrotvm.org/svn/parrot/revision?rev=30151
21:18 lafille joined 21:28 baest joined 21:29 cybergirl joined
dalek r30152 | pmichaud++ | trunk: 22:08
: [rakudo]: spectest-progress.csv update: 121 files, 2200 passing, 3 failing
diff: www.parrotvm.org/svn/parrot/revision?rev=30152
22:23 olgagirl joined 22:41 kid51 joined 22:58 particle joined
jonathan pmichaud: ping 23:09
jonathan is sleeping now 23:22
23:24 britneypire joined 23:31 gmansi joined 23:41 lauanana joined 23:50 particle 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!