Parrot 1.1.0 Released | parrot.org/ | 332 RTs left
Set by moderator on 21 April 2009.
wayland76 indeed 00:03
cotto darbelo, using something like perl tools/dev/mk_language_shell.pl --with-pmc foo /usr/src/decnum-dynpmc will take care of all those kinds of details. 00:04
darbelo Hmm. The Configure.pl generated by mk_language_shell can't find my parrot_config. 00:05
cotto Is it in your $PATH? 00:07
Infinoid that Configure.pl should support --parrot_config=<path_to_parrot>
00:09 AndyA joined
darbelo I looks like it needs an installed parrot: Can't open perl script "/usr/local/lib/parrot/1.1.0-devel/tools/dev/gen_makefile.pl" 00:10
NotFound Please look at the logs, we were talking about that some hours ago 00:11
Infinoid NotFound++ # I was just thinking this sounded familiar
davidfetter HAI 00:12
Infinoid ohai davidfetter
davidfetter i'm confused about version numbers. is 1.1.0 a new major version, as in the packagers should take notice? 00:13
Infinoid the .0 is just there in case we screw up something so badly that we have to immediately release a fixed version
otherwise, our versioning is intended to be as linear as possible.
davidfetter um, in the sense of being a major upgrade from 1.0.0 00:15
Infinoid from a distro perspective, 1.1.0 is definitely worth paying attention to because the latest rakudo needs it 00:16
otherwise, 1.1.0 is just the latest development release; 1.0.0 is still the latest "supported" one
davidfetter hrm 00:17
darbelo The problem is that the generated Configure.pl tries to use "$config{libdir} . $config{versiondir} . '/tools/dev/gen_makefile.pl';" .
davidfetter is there a fixed number for the next "supported" release?
Infinoid yes, 1.4.0
davidfetter k
Infinoid the version number just indicates another month has gone by, that's all. if you're lazy, just package 1.0 and 1.4 (the longterm-supported releases), otherwise I think they're all worth paying attention to 00:18
wayland76 And a fixed date too
it's in the roadmap
davidfetter k
wayland76 trac.parrot.org/parrot/wiki/ParrotRoadmap 00:19
I've found that Rakudo doesn't package properly yet 00:20
darbelo Why is $config{libdir} expanding to something inside /usr/local for a parrot that has never left my $HOME dir?
Whiteknight hello all
purl It's a crazy world, but hello to you too!
Whiteknight hello purl
Infinoid hai Whiteknight 00:21
wayland76 So my suggestion for distros would be to leave the Parrot 1.0.0 package until Rakudo packages properly, and then release an updated Parrot at the same time
davidfetter thginketihW pus'
Infinoid yeah. most people are just interested in rakudo or other HLLs, and can safely pretend parrot doesn't even exist 00:22
(ideally)
wayland76 (Well, when I say "until Rakudo packages properly", I probably should have put in "or some other HLL that needs a higher Rakudo") 00:23
That's the plan, but I think until Christmas, Rakudo is going to be tracking Parrot pretty closely
Infinoid yep. fperrad++ has done an incredible job of this for win32 so far, on his parrotwin32.wiki.sourceforge.net/ site
he's already updated all the HLL plugins except "fun" 00:24
davidfetter better him than me 00:25
(re: windows)
wayland76 Question -- is the plan to remove all Perl5 from Parrot sometime soon?
NotFound For some values of 'soon'
cotto wayland76, eventually, but it's a big project
Infinoid sometime, yes. I think the plan is 2.0 or after
davidfetter are we talking cosmologic soon? geologic?
wayland76 Does that include the build processes too? 00:26
NotFound Before the conjuction of the million of spheres
Infinoid yes. and I was wrong, it's the 3.0 plan
see also trac.parrot.org/parrot/wiki/Parrot...ermRoadmap
wayland76 Ok.
cotto ideally it'll be before the sun goes out 00:27
Infinoid That would be nice.
NotFound Yeah, programming in obscurity is bad for the eyes
darbelo I think the sun will go out brightly enough that you won't have to worry about obscurity anymore. 00:29
wayland76 So probably not something for me to worry about until 2010 anyway
Infinoid yeah, we'll ping you about it then 00:30
NotFound I remember an olf joke about windows nt and the end of world...
cotto wayland76, if you want to help now, it'll be welcome.
wayland76 No, I'm just thinking of making the Rakudo build system depend on some of the Parrot P5 modules
cotto Converting Perl PIR tests to pure PIR is a good incremental change you can work on now.
oic 00:31
wayland76 So let me ask a different question -- what should the build frameworks for HLLs be written in?
NotFound I'm toying with the idea of making at least one HLL build process based only in parrot 00:32
wayland76 Or should they just be written in the HLL, but then precompiled into PIR before distribution?
NotFound wayland76: that can be difficult, because the generated pir may depend of other parts of the hll 00:33
Infinoid It's an interesting idea, though. If you could compile rakudo down to a simple perl6.pir, then you can treat the parrot-dev package the way we currently treat bison/flex; you only need it if you're hacking on it. 00:34
wayland76 Good point. So unless the HLL has some --compile-as-static type option, the build system would have to be in PIR or something
Infinoid NotFound: Are you thinking of the dynpmcs?
We'd probably need parrot-dev for dynpmcs/dynops anyway. 00:35
NotFound Infinoid: for example
Ah, yes I mean parrot-dev, not pure parrot
wayland76: nqp can be used as higher level approach than pir 00:37
wayland76 Ok. I think my plan then will be to use P5 depending on Parrot for now, and we can convert to something appropriate later on. But thanks for the nqp tip. 00:38
darbelo Until I can think of something better I've hacked Configure.pl to be overridable from the command line. If you don't have an installed parrot, you should know where you've put it. 00:44
NotFound Even if I put it in a pen drive that is not plugged, your script must be able to find it. 00:45
cotto darbelo, good to know. Are you going to have something worth committing soon? 00:46
Coke Infinoid: (all the hll plugins) - like, partcl? 00:48
darbelo Not for a few hours, I'll go get some food and then hack some more.
Im trying to get a prototype PMC using just decQuad.c and decContext.c 00:49
just to make sure we can correctly build the library. 00:50
cotto right 00:53
darbelo, I'm making a Google group for discussion of your soc project. Is announce ok with you (i.e. anyone can read, only approved people can post)? 00:54
darbelo Sure, go ahead.
Be back in ~1.2 hours. 00:55
00:58 kid51 joined 01:50 amoc joined 01:57 eternaleye joined
rg uhm is smolder down? 02:00
Infinoid seems so 02:15
02:37 kid51 joined 02:41 janus joined 03:23 braceta left 03:24 Xrblackwe joined
dalek rrot: r38335 | jkeenan++ | branches/install_tools (6 files):
Convert doc_dir to docdir throughout.
 03:32
04:14 Theory joined 04:24 gaurav joined
bacek hi there 04:30
purl salut, bacek.
cotto hi, bacek 04:34
bacek cotto: good morning :)
cotto: I'mm going to merge PMC fixes from pmc_pct branch into trunk. Objections? 04:35
cotto I'm happy to hear it.
well, read
bacek running make test 04:41
dalek rrot: r38336 | bacek++ | trunk/src (20 files):
Merge PMC fixes from pmc_pct branch into trunk. cotto++
 04:55
cotto bacek, what's next? 04:58
bacek cotto: I'm reading PCT docs to understand how properly implement emitting of .c and .h files 04:59
cotto good idea 05:00
purl cotto: Good Idea: Visiting the circus. Bad Idea: Having the circus visit you.
dalek rrot: r38337 | bacek++ | branches/pmc_pct/docs/pct/gettingstarted.pod:
Fix link to supplimentary doc.
 05:08
05:27 darbelo joined
Tene What's the path that parrot uses when I say ".include 'foo.pbc'" ? 05:30
Ah, it uses Parrot_locate_runtime_file_str 05:36
05:38 flh joined
wayland76 That function, IIRC, returns one of three paths depending on which you ask for 05:49
Paths are the only part of Parrot that I know anything about (well, and a little of the build system) 05:50
Tene the PARROT_RUNTIME environment variable 05:51
bacek looks like all power of PCT is useless for PMC "compiling"... 05:53
Tene orly?
purl YA RLY.
bacek Or, more likely, that I'm dumb and dont understand how to use it... 05:54
Tene: yeah... For generating .c and .h files from PMC I don't need POST and PIR stages. 05:55
And looks like using PAST isn't useful too
Tene bacek: you can customize the stages it uses, right?
bacek Tene: yes.
But I "generate_h" and "generate_c" stages are parallel. 05:56
It's more like "PCT::Dumper"
So, I'll have only one stage - "parse"... 05:57
bacek definitely missed something crucial
cotto bacek, you should ask pmichaud about it. 06:00
bacek purl: seen pmichaud
purl pmichaud was last seen on #parrot 17 hours, 14 minutes and 43 seconds ago, saying: cotto: I'm not aware of an easy way to do it.
06:28 eternaleye joined
bacek purl: mgs pmichaud And ideas how to use PCT to generate .h and .c files using PCT? I'm dumb enough to figure out how to do it... 06:28
purl bacek: i'm not following you...
bacek purl: msg pmichaud And ideas how to use PCT to generate .h and .c files using PCT? I'm dumb enough to figure out how to do it...
purl Message for pmichaud stored.
bacek good girl
purl thanks bacek :)
wayland76 bacek: I think that last sentence says the opposite of what you want it to 06:38
It should probably be "I'm dumb enough to TRY TO figure out how to do it" (which means you're stupid for trying), or "I'm too dumb to figure out how to do it" 06:39
cotto He'll get the idea. 06:42
wayland76 cotto: Hopefully, yes. But I wasn't sure which of the two bacek meant, so I thought I'd try to help clarify. But it's probably not important :) 06:44
bacek whatever :) 06:45
English is my forth language.
After Russian, Rude Russian and Very Very Rude Russian :) 06:46
06:47 eternaleye joined
wayland76 bacek: Well, it's been my experience that multilingual speakers appreciate language clarification more than native speakers, so I stereotyped you that way :) but maybe I'm wrong 06:48
Theoretically I speak two languages, but I haven't spoken any French since I was maybe 16 or so, although I've been able to help translate Franglais into English :) 06:49
bacek wayland76: I can't understand difference between "dumb enough" and "too dumb". I'm too dumb for it
wayland76 So you're 2 ahead of me
To give an example, dumb enough = "if you score 10 or more on the scale, then dumb enough is true" 06:50
whereas too dumb = "if you score 10 or more on the scale, then too dumb is false" 06:51
(where 10 is a random number I picked)
bacek oh...
It's even more complicated than PGE sources. 06:52
wayland76 So, dumb enough = ! too dumb
Yes. That's why PGE can be compiled, and English can't :)
bacek :)
wayland76 (or other natural languages)
Although after Christmas, I might be dumb enough to try :)
dalek tracwiki: v11 | cotto++ | ParrotQuotes 06:53
tracwiki: bacek speaks four languages
tracwiki: trac.parrot.org/parrot/wiki/Parrot...ction=diff
shorten dalek's url is at xrl.us/beqb7u
wayland76 "so dumb" is a bit like "dumb enough"
My Dad says the Greek men (in Cyprus, where he was) used to boast "I have too much money" 06:54
bacek cotto: :)))
wayland76 Which means "I have more than I need", and implies "I'd be happier if I had less" :)
bacek wayland76: "World Financial Crisis" can help those men :) 06:57
wayland76 No doubt 06:58
bacek ok. I give up. I'll create my own AST nodes for PMC... We can scratch them later.
darbelo Finally! I got *something* out of the decQuad lib. 07:04
cotto Heh. It always takes a while to get started. 07:05
darbelo Notthing really to get exited about: 07:07
.sub example :main $P0 = loadlib 'decnum' .local pmc decimal decimal = new 'DecNum' $S0 = decimal.'version'() print $S0 07:08
.end
prints decNumber 3.61%
and that's all you are going to get out of this dynpmc right now. 07:10
'version'() is the only implemented METHOD right now.
bacek It compiles? SHIP IT! 07:11
darbelo It compiles if I drop the sources into src/dynpmc/ and hand-hack the Makefile. 07:12
cotto go ahead and commit it 07:14
commit early and often
darbelo I'm still fighting with the Configure.pl that mk_language_shell.pl generates. And the makefiles that Configure.pl generates. 07:17
Everything is full of references to /usr/local. I'm starting to think that parrot_config is assuming that it got installed even when it didn't. 07:21
cotto sounds wacky 07:22
darbelo I might cave in and just install parrot, but I want to avoid that. 07:24
cotto You'll probably have to. 07:25
darbelo That AND install-dev. 07:29
Configure.pl should be finding stuff for me, not forcing me to put stuff where it can find it. 07:31
07:42 gaurav joined
darbelo Well, first commit is in. Not very usable, but it's there 07:49
cotto sweet 07:50
room for improvement
07:51 cognominal joined, deepak joined
darbelo An empty room, ready to be filled with improvement ;) 07:51
07:52 deepak left
darbelo I'm going to get some sleep. I'll try building against an installed parrot later. 07:56
cotto night
darbelo It should be just a matter of adjusting Makefile.in 07:57
tangent: Is is reasonable to ask for an installed parrot and parrot-dev to build dynpmcs? 08:00
08:01 uniejo joined
darbelo I'll go ask my pillow about that :) 08:02
08:02 darbelo left 08:06 iblechbot joined
NotFound purl: msg darbelo You can try to use create_language.pl instead of mk_language_shell.pl 08:33
purl Message for darbelo stored.
dalek rrot: r38338 | bacek++ | trunk/compilers/pct/src/PCT/HLLCompiler.pir:
[pct][cage] Comment out useless pop_eh because push_eh was already commented out.
 08:40
09:06 jq joined 09:07 szabgab joined
cotto If the underlying library handles the code the same way, is there any advantage to having a separate bigint and bignum PMC? 09:50
(thinking about decnumber, not gmp) 09:51
bacek cotto: it depends on library. But parrot's pmc shouldn't
cotto shouldn't have separate PMCs? 09:52
bacek shouldn't depends on particular library 09:56
cotto I see. 09:57
10:02 Ademan joined
bacek cotto: btw, I broke PMC parser yesterday... 10:06
dalek rrot: r38339 | bacek++ | branches/pmc_pct/compilers/pmc/README.pod:
Add little bit into README.
 10:07
bacek *incoming*
bacek hides
rrot: r38340 | bacek++ | branches/pmc_pct/src/pmc/scheduler.pmc:
[cage] Add VTABLE to scheduler.pmc. bacek-- for braking parser.
rrot: r38341 | bacek++ | branches/pmc_pct/compilers/pmc/src/parser/grammar.pg:
Fix PMC grammar.
rrot: r38342 | bacek++ | branches/pmc_pct/src/pmc/schedulermessage.pmc:
[cage] Add VTABLE to schedulermessage.pmc
rrot: r38343 | bacek++ | branches/pmc_pct/src/pmc/key.pmc:
[cage] Add VTABLE to key.pmc
rrot: r38344 | bacek++ | branches/pmc_pct/src/pmc/namespace.pmc:
[cage] Add VTABLE to namespace.pmc
 10:10
rrot: r38345 | bacek++ | branches/pmc_pct/src/pmc/role.pmc:
[cage] Add VTABLE to role.pmc
rrot: r38346 | bacek++ | branches/pmc_pct/src/pmc/object.pmc:
[cage] Add VTABLE to object.pmc
cotto bacek, breaking it is half the fun
bacek cotto: yeah... If you have tests for it. 10:12
cotto It's a lot faster now. 10:13
bacek I've put few '::' in it.
So it will not try to backtrack
cotto I was just going to ask that. 10:14
bacek "done" :)
cotto bacek, are you trying to fix it atm? 10:17
bacek fix what?
purl i guess fix is to just override new
bacek purl: forget fix what
purl bacek, I didn't have anything matching fix what
bacek purl: forget fix
purl bacek: I forgot fix
cotto bacek, nm. I ran the compiler without --target=parse 10:18
bacek ah, it's totally broken atm
cotto yup. That's what I'm seeing too. 10:19
bacek I'm trying to construct proper AST
EPIC FAIL... 10:20
purl hmmm... epic fail is chopping chili and go pee without washing hands or whatever led to you being born
bacek for last few hours
dalek rrot: r38347 | NotFound++ | trunk/examples/streams (4 files):
[cage] drop 'library' from load_bytecode path in some examples/streams, see "load_bytecode 'library/...'" thread in parrot-dev list
 10:26
10:30 flh joined
bacek hooray! It works! 10:32
I need ideas how to unittest PAST creating... 10:34
cotto will there be blood, sweat and tears in the next commit?
You can examine them programatically in pir 10:35
dalek rrot: r38348 | bacek++ | branches/pmc_pct/compilers/pct/src/PCT/HLLCompiler.pir:
Comment out pop_eh
rrot: r38349 | bacek++ | branches/pmc_pct/compilers/pmc/t (2 files):
Check only parse in parse tests
bacek dropping blood tears swearing 10:37
10:39 bsdz joined, uniejo joined
dalek rrot: r38350 | bacek++ | branches/pmc_pct (5 files):
Initial implementation of PAST creating.
 10:42
bsdz bacek: re unittesting past creation/pct. have you seen t/compilers/pct/complete_workflow.t?
bacek bsdz: Yes, I did. But it something different. 10:43
bsdz you could use _dumper to dump your past nodes 10:44
bacek looks like it only one reasonable way to do it...
bsdz how are you thinking? 10:45
slightly different question. was wondering how will we bootstrap parrot on a new architecure with then new PMC compiler you are creating? 10:46
bacek Something like "shallow" comparison of two objects.
bsdz: "lex/yacc"
bsdz ah okay. cool :) 10:47
bacek Yes, certainly desirable, but we're not shooting for that level of
bootstrapping in the 1.0 release. When we eliminate all dependencies on
Perl, we'll move over to a PGE-based compiler, and keep the generated C
code checked into the repository like we keep the generated flex/bison
files checked in.
(c) allison
cotto L1?
purl L1 is a hypothetical language that would be used to implement PMCs and PIR-visible ops so that they could all be easily jitted. or irclog.perlgeek.de/parrot/2009-04-21#i_1083550
bacek purl: L1 is also rt.perl.org/rt3/Ticket/Display.html...txn-471982 10:50
purl okay, bacek.
shorten bacek's url is at xrl.us/beqcfo
bacek it's not L1 but closely related.
bsdz yes recall that conversation. definitely the way to go. will purl pick up that L1 is the PMC compiler and PMC compiler is L1
bacek thinks about HURD 10:52
cotto hopes we don't end up with silly names for intermediate languages like gimple and generic. 10:54
bacek PIG - PIG Is Generic :) 10:55
cotto PING - PING Is Not Generic 10:56
cotto tries to figure out something for "pingu"
bacek PMSE - PMSE Mean Something Else 10:57
"u" for "useful"!
cotto Ping Is Not Generally Useful
bacek indeed! :)
OMG - Our Main Generic :) 10:58
N!
NO!
LOL - Level One Language!
Lets implement PMC in LOL! :) 10:59
cotto lol++ 11:00
karma lol
purl lol has karma of 1
bacek lol++ # vote for lol!
cotto These are much better silly names.
bsdz i'm pretty bad at these but perhaps Parrot is Assembler Runtime Register-based Open-source Type-2-VM :/ 11:06
cotto tl;dr
11:10 iblechbot joined
bsdz does the trac roadmap still think milestone 1.1 hasn't been reached? or am i misreading it? 11:11
11:15 uniejo joined 11:39 he left 11:52 Whiteknight joined 11:59 darbelo joined 12:29 Santtu joined 12:52 Casan joined 12:57 cheflog__ joined 13:22 kid51 joined
mikehh I can't seem to connect to smolder.plusthree.com 13:27
13:29 rdice joined
mikehh I did a sake smoke at r38350 which PASSes but get a connection refused error when it tries to send it 13:29
Whiteknight confirmed, I can't connect to it either
kid51 I believe there was talk on list last night about Smolder being down. 13:30
I wish I had remembered that before I just started to run Smolders on 2 OSes :( 13:31
Whiteknight yeah, I know what you mean 13:34
I still need to get smolder working properly on my win64 box
mikehh oh well - I am building and testing rakudo now and will some other tests on parrot after that 13:35
BTW I installed Ubuntu Jaunty i386 yesterday and had great fun with CPANPLUS getting the perl modules up to date 13:43
kid51 Fortunately, we have some POD on dealing with Smolder outages: 13:44
On network problem or for offline use you may send tar reports later
with that command:
perl -Ilib -MParrot::Harness::Smoke \\
-e'Parrot::Harness::Smoke::send_archive_to_smolder(Parrot::Harness::S
moke::collect_test_environment_data())'
dalek rrot: r38351 | whiteknight++ | trunk/DEPRECATED.pod:
Add two function names to DEPRECATED.pod for TT #155 and #156
 13:48
kid51 From 70.42.41.97 icmp_seq=21 Packet filtered 13:49
--- smolder.plusthree.com ping statistics ---
21 packets transmitted, 0 received, +14 errors, 100% packet loss, time 20100ms
mikehh you can also connect to the site and upload the parrot_test_run.tar.gz file 13:50
13:53 fperrad_ joined 14:03 HG` joined 14:39 amoc joined
Infinoid good morning 14:43
Whiteknight mornin' Infinoid! 14:50
14:53 Woody4286 joined 15:19 tetragon joined 16:08 HG` joined
dalek rrot: r38352 | whiteknight++ | trunk (4 files):
Remove the unused no-op function stack_destroy() per TT # 523
 16:08
Infinoid ooh, Whiteknight++ # working on attr parsing 16:14
Whiteknight Thanks. That patch is buggy though, so if you can take a look at it, that would be awesome
Infinoid I will, thanks 16:15
I think I recall some cases elsewhere in pmc2c which attempted to strip off the type to get the name, which expected the type to be a simple \\S+\\s*\\* prefix 16:19
getting the name out of a function pointer declaration is a little more exciting.
Whiteknight I didn't see any code that looked obviously malformed, everything in src/pmc/pmc_managedstruct.h looked reasonable 16:20
and it did compile and run, so that's something
Infinoid you're right, that does all look good 16:21
Infinoid running make test now
hmm. # Attribute 'oid' already exists in 'ManagedStruct'. 16:22
is that some kind of attribute inheritance issue?
nopaste "Infinoid" at 75.5.243.250 pasted "Backtrace from adding the redundant 'oid' attr (caused by this PIR: test = new ['ManagedStruct'])" (33 lines) at nopaste.snit.ch/16410 16:28
Infinoid it seems to be creating a PMCProxy rather than a ManagedStruct. I don't know if that's normal or not. 16:30
16:32 kid51 joined 16:33 gaurav joined
Infinoid after the patch, pmc2c emits an invalid vtable->attribute_defs string 16:43
before: Isize Finit :custom_free_func :custom_free_priv :custom_clone_func :custom_clone_priv 16:44
after: Isize Finit :(*custom_free_func)(PARROT_INTERP, void *, void *) :custom_free_priv F(*custom_clone_func)(PARROT_INTERP, PMC *ptr, void *priv) :custom_clone_priv
(I'm guessing the type of "size" was "I", this is pulled from printing the half-parsed string)
so it's another place where pmc2c needs help parsing the name from the declaration 16:46
Whiteknight okay, so that makes good sense then 16:56
now we just need to find the place where that attribute string is being defined
or, being populated
okay, I think I found that point in the code. 17:04
Whiteknight starts hacking a fix
Infinoid yeah 17:15
I was halfway done finding the place for a fix, but then I got a phone call, sorry
Whiteknight Infinoid++ I think I've got it 17:19
i'm going to commit since all tests pass 17:20
Infinoid Whiteknight++
17:22 masak joined
dalek rrot: r38353 | whiteknight++ | trunk (4 files):
A fix for TT #544, plus proof-of-concept implementation in ManagedStruct PMC. Allows parsing arbitrary function pointers as ATTRs, to reduce the use of void* pointers floating around.
 17:26
17:26 wayland76 joined 17:48 iblechbot joined
Infinoid Parrot has a reasonably sane deprecation policy, but one thing I feel is missing is a migration guide. That is, when we remove a feature or API, we should drop a note somewhere saying "If you used this, now you should use this instead" 18:02
I've had to work with some projects which do this, and some projects which don't. I really like it, because as a third party developer, it reduces the time for me to find a fix by at least an order of magnitude 18:03
If we were to do this for parrot, where would we put it?
(Or are we already doing this and I just haven't found it yet?) 18:05
18:07 wayland76 joined
Whiteknight maybe sounds like a Trac thing to me 18:23
pmichaud we already have this. 18:26
trac.parrot.org/parrot/wiki/Deprecation
18:30 Casan joined
Infinoid oh good, thanks. 18:31
18:54 eternaleye joined 18:58 s1n joined, barney joined 19:02 kid51 joined 19:14 eternaleye joined
Tene gdb takes a while to walk 280k stack frames. 19:22
... oh, gdb segfaulted.
Whiteknight 280k stack frames? sounds to me like an infinite loop 19:35
Infinoid hmm, it would be nice if Devel::NYTProf saved its sessions to separate files for each invocation, by including the pid in the filename or somesuch 19:36
turns out perl is invoked almost 500 times during a normal "make"
looks like c2str has the biggest traces 19:38
I'll find out in a moment whether that's relevant. 19:39
Tene Whiteknight: it is. 19:40
Whiteknight: was I also talking with you about the pmc hll stuff? 19:44
Whiteknight yeah 19:46
nopaste "tene" at 166.70.38.237 pasted "pmc hll small example for whiteknight" (7 lines) at nopaste.snit.ch/16411
Whiteknight is that the infinite loop? 19:47
Tene No.
19:47 pfig joined
Whiteknight okay 19:47
Tene The infinie loop was from trying to migrate rakudo to .HLL 'Perl6'
Which I would also love help on... I'm out of ideas for now.
I'm not sure if it's relevant to this issue, though. 19:48
Whiteknight I'd love to help, you have any information about the work you've been doing so far?
pfig evening
in utc, at least :)
Tene Whiteknight: the 'hll' branch in rakudo
Whiteknight Tene: okay. I don't know GIT, how do I check it out? 19:49
Tene git fetch; git checkout origin/hll
will work
if you want to make some local commits, you can then 'git checkout -b hll' to make a local branch 19:50
pfig there's a problem with dynpmc which causes ld to fail. after googling i added -c manually to the c++ invocation, but then all sorts of tests fail 19:51
is this a known issue? mac os x 10.5.6/intel 19:52
both parrot 1.0.0 and 1.1.0
Tene Whiteknight: I must leave now... let me know what you find? 19:54
dalek rrot: r38354 | jkeenan++ | branches/install_tools/lib/Parrot/Install.pm:
Correct bug I introduced in r37913: should have been 'lib/Parrot/Install.pm'.
 19:55
Whiteknight Tene: Ok, if I can fnd anything
rakudo is very new to me
Tene Whiteknight: I think that PMCs need a way to say what HLL namespace they're installed in.
Whiteknight okay 19:56
20:03 gryphon joined
kid51 msg wayland76 I believe I fixed bug in install_files ticket. Please svn up in branch and try out. thanks 20:07
purl Message for wayland76 stored.
20:19 s1n joined 20:31 Whiteknight joined
pfig ah, g++ is used as linker... gcc runs with -c, though so i'm stumped. everything works fine in linux 20:41
dalek rrot: r38355 | whiteknight++ | trunk/src/dynext.c:
Remove some unhelpful comments from src/dynext.c. One is about the now-defunct struct val, and the other is just the string 'XXX'
 20:42
pfig right, that's parrot done for another month then
Whiteknight pfig: I've never heard that issue before 20:45
open a ticket
20:54 pfig joined 20:55 eternaleye joined 21:30 eternaleye joined 21:34 eternaleye_ joined
Infinoid Are we tied to the current hash function c2str.pl uses, for any reason? (This is the second column of output in all_cstring.str.) 22:09
nopaste "infinoid" at 75.5.243.250 pasted "[PATCH] [tools] Make c2str.pl stop using Math::BigInt and simplify the hash algorithm a bit; this speeds builds up parrot builds by 10 seconds (8%) for me." (36 lines) at nopaste.snit.ch/16412 22:11
Infinoid before: make -j8 112.34s user 16.40s system 173% cpu 1:14.01 total 22:13
after : make -j8 101.88s user 15.56s system 174% cpu 1:07.24 total
uh. It doesn't look like the hash value is ever used, anyway. I can probably kill that and save some more time 22:28
22:40 s1n joined
bacek Good morning 22:45
Infinoid: patches welcome to pmc_pct branch :)
Infinoid hi bacek 22:46
what's pmc_pct?
bacek Experiment with compiling PMC using PCT 22:47
trac.parrot.org/parrot/browser/bra...grammar.pg
shorten bacek's url is at xrl.us/beqdye
bacek I can already parse all PMCs :) 22:48
Infinoid bacek: Be aware, Whiteknight++ improved pmc2c slightly today, allowing for function pointer types for ATTRs, see trac.parrot.org/parrot/changeset/38353
So you might need to update it for the changes to trunk/src/pmc/managedstruct.pmc within that diff
bacek Infinoid: no, it already allowed in my grammar. 22:49
Infinoid nice stuff tho!
dalek rrot: r38356 | Infinoid++ | trunk/tools/build/c2str.pl:
[tools] Make c2str.pl stop using Math::BigInt, simplify the hash algorithm.
rrot: r38357 | Infinoid++ | trunk/tools/build/c2str.pl:
[tools] The hash_val output of c2str isn't used anyway, just get rid of it.
bacek Hm... I was wrong. "Failed to parse source". 22:50
Ah. Function pointers...
Infinoid Yeah. Don't you love C's syntax for those? 22:51
bacek have to steal more stuff from c99.pg
gotta go. Will be back in ~1 hour 22:53
22:53 barney joined
Whiteknight bacek: This is really nice work 23:18
bacek++
bacek Whiteknight: little step to rule the World :) 23:34
23:34 Theory joined
Whiteknight bacek: Where does the C code get generated? 23:35
or does that not happen yet?
bacek Whiteknight: not yet
I just started 2 days ago. 23:36
Whiteknight ah. Well, looking forward to it
bacek "patches welcome" :)
23:45 kid51 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!