Parrot 3.1.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Goals: Get more GSoC ideas on wiki; close tickets; stable 3.2 release; assess status of roadmap goals for 3/15 meeting.
Set by moderator on 9 March 2011.
00:12 mtk left 00:17 bubaflub joined 00:18 mtk joined
plobsing whiteknight: I've found my problem (I needed to pull), but I am seeing the failure in pbc_to_exe 00:26
whiteknight really? what is it?
I haven't looked too hard yet 00:27
plobsing I'm seeing an assertion failure in pbc_to_exe. trying to lookup string reg 9 when the ctx only has 5. 00:28
whiteknight hm, the error I was seeing was somethig about vtable elements not implemented in Integer 00:30
so somewhere along the line we are probably recycling a context
okay, thanks for that information 00:31
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#12030) fulltest) at 3_1_0-784-g34dda14 - Ubuntu 10.10 i386 (g++-4.5 with --optimize) 00:40
whiteknight plobsing: Nice project ideas
plobsing give the users what they want 00:41
whiteknight I would like to see either of those. I suspect the GCC one would be hard to maintain, but that's not really a criteria
plobsing hard to create, hard to maintain, hard to sell. but creating it would be a very interesting experience for a determined student and the insights gleaned along the way would be invaluable. 00:42
whiteknight GCC has something of a reputation of being..."difficult" internally
plobsing which is why people went to the trouble of translating MIPS instructions to JVM classes to avoid working with GCC internals. 00:43
whiteknight I've heard more than a few theories that the code is intentially obfuscated to prevent non-free additions to the code
plobsing and yes, I have looked inside GCC. sadly I do not know what a const pointer pointer const pointer const is good for. But I've seen one.
whiteknight ...lolwat?
at that point, you're just being difficult
I call those things a "void*" 00:44
plobsing yeah, they like their pointers around there. a lot apparently.
I exagerate a little, but not by much.
whiteknight I'm sure it's not much
plobsing there were definitely 4 levels of indirection but no consistency with constness
whiteknight: I doubt the obfuscation rumours. compilers are inherently complex. it is an old project. and nobody wants to work on plumbing when there is NEW SHINY OPTIMIZATION to be tried out. 00:46
whiteknight i sort of doubt it too. at the same time, richard stallman can be a little bit crazy 00:48
davidfetter "can be?" 00:49
"a little bit?"
plobsing he wrote emacs. we have proof of his insanity.
whiteknight his public arguments with linus torvalds, or the guy from Mono are nothing if not absurd and disappointing
davidfetter i think that being crazy is required for the thing he did, back in the 80s
...which is to flout all of common wisdom and start a movement in the middle of a market
whiteknight no doubt, getting open-source software started as a movement is certainly counter intuitive
davidfetter sadly, once the movement kicks off--and it has, far beyond his wildest dreams--the crazy is still there 00:50
plobsing I used to be an avid emacs user. Then I went to extend it (hey, what are those FSF freedoms for if you don't use them?). Now I use vim on principle.
davidfetter was thinking about switching to emacs, just to keep the brain cells active 00:51
whiteknight I tried using emacs once, but it was a long time ago 00:56
I may give it another spin eventually. I am perpetually unhappy with my text editors 00:57
mikehh whiteknight: I have been using Kate for a long time - it is very configurable, with lots of plugins 01:03
I also use gvim and have used emacs but you need to use it regularly if you want to remember the key combos 01:05
plobsing basic key combos for emacs are pretty easy to remember. (by default) they're the same for everything that uses readline. 01:06
Tene I've been very happy with vim.
davidfetter i'm good with it, and #vim over on fleanode is jammed with helpful, friendly people 01:08
mikehh plobsing: yeah, but if you use other editors you tend to get mixed up (or at least I do) and keep forgettin' how to exit :-}
davidfetter heh
plobsing mikehh: you mean it isn't (launch terminal)->killall emacs ? 01:09
01:10 Andy joined
davidfetter the cool kids are using pkill, mostly 01:11
cotto ~~ 01:15
Tene fwiw, "Tk for partcl" doesn't seem nearly as good to me as "HLL interop works, Tk library for parrot that can be used by partcl and all other HLLs." 01:19
plobsing is there something about partcl that would prevent a partcl Tk library from working in other HLLs? 01:21
Tene plobsing: Not prevent, it's just that Tk doesn't actually require Tcl 01:23
plobsing whiteknight: investigating further, it appears I had an invalid library/config.pir (not sure how that got built). realclean/config/make now gives no errors again (at b12bcf3e73137e6963b7652abc82988a91a08edd) 01:24
whiteknight plobsing: damnit 01:25
plobsing I've got to figure out why my computer doesn't fail like everyone elses. Then bottle it.
dukeleto whiteknight: i am not sure TapTinder is a valid Parrot GSoC project 01:27
sorear Tene: I think one of the goals of a gsoc project is that it be something people want 01:29
Tene: the desirability of HLL interop seems quite oversold
dukeleto Tene: i want HLL interop, but I think that gsoc project idea need more defined deliverables 01:30
Tene: perhaps, implement HLL interop enough for X to work 01:31
sorear: oversold in what way?
plobsing dukeleto: "HLL interop enough for X to work" might need a working HLL with an existing large body of software that we'd like to access from another language. I'm not sure any of our currently working HLLs have that. 01:32
dukeleto plobsing: why not "make Lua and Rakudo able to do interop" or somesuch
plobsing dukeleto: enough to do what with them? there is no large body of Perl 6 software a lua programmer would want to use. There is no large body of lua software (that I know of) that a Perl 6 programmer would want to use. 01:34
for example, if cardinal were working, we'd be able to say "make HLL interop work well enough to write a small Rails webapp in Rakudo"
Rakudo on Rails. that'd be freaky. 01:35
whiteknight dukeleto: I didn't add taptinder to this list. I just reformatted it
davidfetter
.oO(winxed on rails)
whiteknight dukeleto: remove it if you don't thnk it's workable
davidfetter winxed on wails? 01:36
dukeleto whiteknight: not that it isn't workable, just not parrot-related
whiteknight: wasn't blaming you, just notifying
whiteknight right, if it's not for us, we can remove it
no worries
dukeleto indeed
$self.nom($dinner)
davidfetter $apetit.bon() 01:37
plobsing ENOVERB
Tene plobsing: That's been my goal in cardinal all along. I have little personal interest in ruby, but I have a lot of interest in seeing what it might enable. 01:40
bubaflub Tene: any ideas for cardinal for GSoC? i'd definitely be interested... 01:41
Tene Once I get a metamodel impl working well enough on 6model, I expect to be able to port large amounts of ruby stdlib from projects like rubinius, start running the rubyspec test suite, etc.
whiteknight I keep thinking "migrate Cardinal to 6model"
but I think Tene and tewk and others will make that happen before GSoC
Tene whiteknight: I've been working on that intermittently for the past couple of weeks.
whiteknight Tene: right. 01:42
Tene whiteknight: tewk hasn't been involved in this iteration of cardinal. He started a previous iteration with the same name a while before this implementation started.
you mean treed, maybe.
whiteknight I only mention his name because he sent a mail to the list recently talking about it
Tene Ahh, I didn't see that. 01:43
whiteknight right around the time he submitted that Select PMC
Tene whiteknight: No he didn't; I did, though. 01:45
whiteknight oh, Maybe I misread an email
4 letters, both start with T
Tene Yeah, not the first time it's happened. No worries. 01:46
dalek tracwiki: v24 | dukeleto++ | GSoc2011 01:49
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v25 | dukeleto++ | GSoc2011
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
cotto busy day in here
Tene An implementation of R5RS or R6RS is probably feasible within 2-3 months, for an aspiring HLL developer for GSOC 01:50
(scheme)
01:50 davidfetter left
whiteknight R5RS? 01:52
plobsing there's also ERR5RS 01:53
whiteknight ok
Tene Implementation of *some* scheme standard. There's a fair amount of politics around which one is awesome and which one is horrible, I hear.
02:13 jsut_ joined 02:18 jsut left
mikehh oh forgot to report 02:34
rakudo (a38d453) - builds on parrot (3_1_0-784-g34dda14) - make test, make spectest_smolder[(#12055), roast (d38ece7)] PASS - Ubuntu 10.10 i386 (gcc-4.5 with --optimize)
27,629 ok, 0 failed, 604 todo, 1,799 skipped and 0 unexpectedly succeeded
bah that was g++ not gcc
02:36 Andy left
plobsing whiteknight: how is matrixy doing? how closer are you to say getting the controls toolbox on octaveforge to work? would that be a gsoc-sized thing? 02:38
bubaflub whiteknight: i'm looking at PLA, but it looks like there are known bugs with LAPLACK on Mac OS X
02:38 jsut joined
whiteknight bubaflub: are you on OSX? 02:38
bubaflub whiteknight: yeah 02:39
10.6 and change
whiteknight plobsing: matrixy has been on hold for many months now. I took a detour to work on PLA first. I think all the prereqs are in place now, but I probably want to start from scratch instead of salvaging what is there
plobsing: it will eventually be my excuse to learn ohm-eta, unless I want to try to do it all in NQP again
bubaflub whiteknight: from what i'm reading it looks like Apple includes some LAPACK stuff but not the full thing 02:40
whiteknight bubaflub. Well that's a bummer. You probably won't want to take that project if it's going to be a hassle on your system 02:41
bubaflub whiteknight: heh. yeah. i could work within a Virtual Box...
whiteknight: last GSoC with RTEMS i was in a ubuntu virtual box
02:43 jsut_ left
plobsing aw... I was hoping to be able to make misleading but pretty graphs with parrot :( 02:44
dalek sella: 9844036 | Whiteknight++ | s (6 files):
proxies have a target_object field, the object which is being proxied (if any). Fix AttributeIntercept to properly handle PMCProxy target_class
 02:49
sella: 7a4244c | Whiteknight++ | s (4 files):
add an Immutable proxy builder. cleanup and streamline the decorator builder
sella: 18d1cde | Whiteknight++ | src/proxy/builder/Passthrough.winxed:
fix so the Passthrough builder works correctly
whiteknight okay. apparently I can add a vtable override to a class, but I can't change it once it's been set and I can't remove it 02:52
there's an add_vtable_override method and VTABLE, but no remove_vtable_override 02:53
plobsing welcome to parrot's deficient object system 02:55
03:13 bubaflub left 03:14 ShaneC left 03:27 ShaneC joined 03:37 lopaway left 03:48 jrt4 left
whiteknight on the bright side, so long as I don't name any of my attributes "proxy", I can work around most of these warts 03:54
considering that the whole library is named "proxy", and it's a library about making proxies, that's a bigger restriction than you would expect 03:55
Why, if we're going to set aside an attribute name for internal-only use, we wouldn't pick a name like "!!!@proxy@!!!" instead of something almost any HLL could actually conflict with, I have no idea
I mean, attributes can be named any arbitrary string. So all we have to do is pick a string which most HLLs can't use to name identiiers 03:56
...what's that smell? I think I just found all our missing foresight 03:57
it's rotting in a pile, unused 03:58
cotto how internal is that name? 03:59
I think I might be smelling chainsaw oil.
whiteknight it's all over class.pmc, and the generated code in src/pmc/object.c 04:00
or, maybe PMCProxy pmc
cotto by "internal", I mean "exposed to the outside world and therefore subject to our support policy" 04:01
whiteknight oh 04:03
it's rare, and is always bug
well, maybe not "always", though I'm hard-pressed to find a reason why reserving the attribute name "proxy" in certain situations without warning is a good thing 04:07
I guess maybe some advanced hackers might want to get access to the underlying proxy object
although again, I can't think of a reason
anyway, I'm off to bed now 04:09
goodnight
cotto My best guess would be that perhaps Rakudo and kakapo do something with it.
'night
04:09 whiteknight left
dalek rrot: b024719 | petdance++ | config/gen/makefiles/root.in:
turn on initial hints
 04:38
04:38 cosimo joined, Andy joined
dalek rrot: a8cd489 | petdance++ | config/gen/makefiles/root.in:
Splint: Don't complain when mixing booleans and integers
 05:24
dukeleto ~~ 05:42
dalek rrot: b1dfb81 | petdance++ | config/gen/makefiles/root.in:
splinting src/gc is too annoying
 05:45
rrot: 236d9cc | petdance++ | / (3 files):
Only pointers get ARGIN() decorators
05:45 Andy left
dalek rrot: b96f999 | dukeleto++ | t/src/extend_vtable.t:
Merge branch 'leto/embed_grant'
 05:50
dukeleto it seems very hard to use the clone_pmc or cmp_pmc vtables. I am starting to wonder if it is possible at all. 05:55
# Exception is: type 44 severity 2 message 'Multiple Dispatch: No suitable candidate found for 'cmp_pmc', with signature 'PP->P'' 05:56
yet, PP->P seems to be it's signature
dalek rrot/cmp_pmc: 34a4162 | dukeleto++ | t/src/extend_vtable.t:
Add failing test for cmp_pmc
 05:58
cotto dukeleto, awesome 05:59
dukeleto yay, i snagged TT# 2^11 06:01
cotto: you know what is going on with that?
dalek TT #2048 created by dukeleto++: cmp_pmc vtable acting wonky 06:14
TT #2048: trac.parrot.org/parrot/ticket/2048
cotto dukeleto, looking now 06:15
dukeleto, it's probably because nothing in the inheritance chain (other than default) implements either the cmp_pmc VTABLE or a MULTI with that name. 06:18
07:00 rurban_ joined, theory left 07:02 rurban left, rurban_ is now known as rurban, Khisanth left 07:11 fperrad joined 07:14 Khisanth joined
cotto dukeleto, you still up? 07:43
07:59 mtk left 08:05 mtk joined
cotto seen nwellnhof 08:26
aloha nwellnhof was last seen in #parrot 2 days 11 hours ago saying "~".
09:03 cosimo left 09:13 lateau_ left 10:11 lucian joined
dalek rrot: b63c2c9 | fperrad++ | runtime/parrot/library/TAP/Harness.pir:
[TAP] refactor with StringBuilder
 10:33
rrot: b1fa14e | fperrad++ | runtime/parrot/library/Archive/Tar.pir:
[Tar] minor refactor
10:37 sigue left 11:08 whiteknight joined
mikehh switching to amd64 - bbiab 11:42
11:42 mikehh left 11:54 mikehh joined
mikehh looks like comment on TT #985 is SPAM 12:15
12:29 zby_home joined
dalek sella: 2b4af8d | Whiteknight++ | / (7 files):
fixes to immutable proxy builder. Method and Attribute intercept builder default behavior is to check the controller first then fall back to the target object. Add tests for Passthrough, MethodIntercept, and AttributeIntercept
 12:30
12:30 whiteknight left 12:43 dodathome joined 12:47 sigue joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#12103) fulltest) at 3_1_0-793-gb1fa14e - Kubuntu 10.10 amd64 (g++-4.5 with --optimize) 12:50
13:11 kid51 joined 13:20 kid51 left 13:22 kid51 joined 14:03 PacoLinux left 14:04 PacoLinux joined 14:25 plobsing left 14:29 kid51 left
dukeleto ~~ 14:47
14:57 ambs joined 15:01 rurban_ joined 15:02 rurban left, rurban_ is now known as rurban 15:09 M_o_C joined 15:14 M_o_C left, M_o_C joined 15:18 bubaflub joined 15:23 M_o_C left 15:24 M_o_C joined
lucian allison: hi 15:37
15:51 whiteknight joined
cotto ~~ 16:09
whiteknight hello cotto 16:11
cotto good morning, whiteknight 16:12
16:19 Patterner left, Psyche^ joined, Psyche^ is now known as Patterner, bubaflub left 16:22 bubaflub joined
dukeleto cotto: pong 16:29
cotto: so what should I do about writing cmp_pmc tests?
bubaflub good morning dukeleto 16:31
cotto dukeleto, I suspect that I pinged you to ask if you thought that having M0's ffi use the types in libffi would be sufficient for any ffi needs. 16:34
github.com/atgreen/libffi/blob/mas...ibffi.info ("primitive types") 16:36
dukeleto, I originally added cmp_pmc for Pipp's arrays. I need to dig up the exact reason. 16:38
dukeleto bubaflub: mornin' 16:42
cotto: i was ponging you in response to what I should do with cmp_pmc
bubaflub: started the new gig yet?
16:43 theory joined
dukeleto bubaflub: have any ideas for parrot in GSoC 2011 ? 16:44
cotto dukeleto, I'm wondering if it's necessary.
whiteknight dukeleto, cotto: have you two signed up for projects you'd be willing to mentor? 16:45
cotto whiteknight, 16:47
I will soon have
bubaflub dukeleto: not yet, this is my last week at Wolfram
allison lucian: hi
whiteknight There are a lot of good ideas on that page
bubaflub dukeleto: i was looking at PLA and the GSL bindings
dukeleto: possibly doing something with Plack for Perl
dukeleto whiteknight: i think i may just try to stay org admin again. I will of course meta-mentor any of our students, but I think being org admin is enough work 16:48
whiteknight dukeleto: Okay, that's fine too
dukeleto whiteknight: i am trying to think up more ideas for the ideas page now, though
cotto editing now
dukeleto bubaflub: we need more meat on our ideas page, it would be awesome if you could throw a few ideas on there
watch out for conflicts on the trac wiki 16:49
cotto done
dukeleto bubaflub: perhaps porting Math::Primality to parrot ?
bubaflub dukeleto: oooooh. do we have GMP bindings?
whiteknight not good ones 16:50
bubaflub so perhaps GMP bindings first?
dukeleto bubaflub: we have basic GMP bindings, but yes, we need access to more of GMP 16:51
bubaflub: that is a great idea, can you add it?
cotto you could do a pure-pir pmc to wrap those using nci
bubaflub dukeleto: yeah. anyone willing to mentor that? i'd love to work on that this summer
dukeleto bubaflub: are you still eligble to be a student in GSoC ? 16:52
bubaflub dukeleto: yep.
dukeleto bubaflub: or are you saying you want to be a mentor?
bubaflub: oh, cool!
bubaflub: i guess I my arm could be twisted to mentor you....
bubaflub dukeleto: hahaha
dukeleto: winning.
dukeleto bubaflub: only because I know you won't need much help ;)
dalek tracwiki: v26 | cotto++ | GSoc2011 16:53
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
bubaflub dukeleto: i think we can do what Sysiphus did with Math::GMPz and other related modules - separate integer, rational, and complex functions 16:54
dukeleto bubaflub: sounds like a plan, stan 16:59
lucian allison: you said at some point that there are some license/cp issues with pynie 17:00
allison lucian: not issues
lucian: just we'd like to donate it to the Python Software Foundation
lucian: or, if we start from scratch, make it a PSF project from the start 17:01
the PSF board is still open to the idea, the Chairman asked me about it yesterday
lucian allison: right. nice
dukeleto allison: who is the chairman these days?
allison (and full disclosure, I'm also on the PSF board)
lucian i didn't really expect that, i say PSF dead-set on cpython
allison Steve Holden
dukeleto allison: ok, that is what I thought, but making sure
allison no, not at all, the PSF actively supports all implementations 17:02
they just gave a $10k check to PyPy
lucian allison: sure, but they only seem to want to own cpython
dukeleto allison: he just moved to PDX. Might be interesting to meet with him
lucian likely had the wrong impression
allison and have actively stated this week that they're open to grant applications from all implementations
dukeleto allison: that is quite interesting
lucian indeed 17:03
allison especially if those grant applications are for implementations of Python 3
(which pynie currently is)
lucian allison: right. would you be willing (and have time) to mentor me on pynie-ng? 17:04
whiteknight getting a good python implementation up and running on Parrot should really be considered a high priority for us 17:05
17:06 M_o_C left
dukeleto lucian: if pynie-ng isn't on our ideas page yet, it should be 17:06
whiteknight lucian: Can you add it to the ideas page?
lucian here? trac.parrot.org/parrot/wiki/GSoc2011
allison lucian: sure 17:07
bubaflub dukeleto: i've updated the ideas page with GMP bindings
17:08 M_o_C joined
lucian needs to make a trac account apparently 17:08
dalek tracwiki: v27 | bubaflub++ | GSoc2011
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
Tene once I get cardinal running on 6model, I'd love to write a python object model in 6model too 17:09
whiteknight lucian: What are your ideas for pynie-ng? I'm very interested to hear what you are thinking
17:09 M_o_C left
dukeleto bubaflub++ 17:09
lucian whiteknight: as i said before, try to move as much as possible to pure-python. ideally reusing existing compilers/other code
Tene: i won't bother with that yet :)
whiteknight any specific existing compilers and other code you have in mind?
lucian is confused by 6model 17:10
whiteknight: PyPy has a good one
there are a few more
17:10 M_o_C joined
whiteknight Tene: how much work have you done with 6model already? 17:11
lucian do i need special permissions to edit trac pages?
whiteknight lucian: let me look
cotto lucian, no more than you have ;)
whiteknight lucian: are you a committer?
lucian whiteknight: no 17:12
lucian is likely not finding the obvious 'edit' button
Tene whiteknight: I haven't contributed anything to 6model yet. I've got a RubyClassHOW working for some things, and doing a few things that the parrot version couldn't, but it's not quite there yet.
bubaflub lucian: bottom of the page
cotto lucian, I just flipped the bit on your commit.
er, account
whiteknight Tene: I only ask because we want to get 6model into Parrot at some point, and we are going to need help with that from people experienced with it
Tene whiteknight: jnthn was talking about writing a simple repr class for me to use, that just does flat attribute storage without looking at ancestors; if he doesn't get to it before I'm done with these other tasks, I may end up doing it myself. 17:13
lucian cotto: uh, can't seem to edit anything. even searched for "edit" 17:17
Tene whiteknight: It seems likely that I'll help with that work.
whiteknight lucian: edit button should be at the very bottom of the page
bubaflub lucian: PM me the text you'd like, i'll put it on the page 17:18
whiteknight Tene: Awesome. When the time comes we will need all the help we can get
lucian whiteknight: not there
dukeleto lucian: stupid wiki spams makes us have to not allow editing by default
dukeleto kicks spammers in the face
Tene whiteknight: we currently need all the help we can get. ;)
whiteknight lucian: I fixed something in trac. Try again 17:19
Tene: of course, but with 6model in particular
17:19 bacek left
lucian whiteknight: ah, there we go 17:20
whiteknight lucian: you were in group "developers", not "developer"
17:22 bubaflub left
cotto sigh 17:22
17:22 bubaflub joined
cotto apparently I shouldn't be doing anything that involves a keyboard today 17:23
whiteknight I know the feeling. I had stupidfingers yesterday
cotto dukeleto, you can nuke cmp_pmc if there's no good way to test it, provided no active hll is using it 17:24
(which I'm pretty sure is the case)
though that does require a deprecation cycle 17:25
whiteknight what ever happened with pipp?
cotto whiteknight, it's dormant
17:25 kid51 joined
whiteknight which language was that, php? 17:26
cotto yes
lucian python3 on parrot added trac.parrot.org/parrot/wiki/GSoc2011 17:27
dukeleto lucian++
bubaflub mod_parrot with pipp would be cool 17:29
Tene rather
I had a lot of fun with mod_parrot a couple of years ago.
bubaflub Tene: what happened to mod_parrot? 17:30
Tene bubaflub: nothing.
dukeleto bubaflub: it existed before our embed/extend interface, so it relied directly on parrot internals
bubaflub ah, so a bit fragile 17:31
dukeleto bubaflub: which change about every other day. Development stopped on mod_parrot and it quickly got TEH BITROT
whiteknight a new mod_parrot using proper interfaces would be a great tool
bubaflub whiteknight: agreed
dukeleto a new mod_parrot using whiteknight++'s new embed api would be sweet
Tene bubaflub: Just... nobody was working on it, it was an external project that only (mostly) one guy was working on, and we didn't have a centralized replace doing smoke testing...
bubaflub there is a Parrot in Apache project, but no mentors listed
(on the GSoc2011 page) 17:32
whiteknight yeah, I added that
Coke mentioned it might be a good project
Tene I can probably be a resource for the templating engine project 17:33
bubaflub i've used Apache, but never done any module writing for it 17:34
if it doesn't go for GSoC, i'd be willing to work on it independently
i think it would help drive PHP, Python, Perl6, and Ruby HLLs 17:35
(among other things)
whiteknight bubaflub: my thoughts exactly
Tene argh I really need sleep. Very very bad that I stayed up this late two nights in a row.
wtf am I doing?
whiteknight Tene: The template engine was my idea too. We should talk about it when you're rested
Tene afk sleeping
whiteknight: Sure.
whiteknight goodnight 17:36
Tene we could also add a task for real threading, as an alternative to green threads
interpreter sandboxing, perhaps 17:37
whiteknight I am holding off on real threading because there are some things we need first (a data sharing system, for example)
sandboxing would make a great project too
bubaflub what happened to last year's GSoC project related to threads?
whiteknight I hate to think about anything like that when we are already talking about major interp refactors for lorito
Tene are there any lorito-related tasks that we could give out? 17:38
whiteknight bubaflub: Chandon was able to get a green threads implementation mostly working on Linux only
Tene: good question. That's for cotto to think about
Tene someone should definitely add a scheme implementation to the list of suggestions
that's a small enough language to do quite a bit of in 2-3 months
whiteknight Tene: okay, we'll keep that in mind
I don't know much about Scheme, so I probably won't add it as an idea 17:39
Tene Parrot as a browser plugin is an entertaining proposal.
dalek tracwiki: v28 | lucian++ | GSoc2011
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
tracwiki: v29 | jkeenan++ | GSoc2011
tracwiki: Add basic links to explain what 'LALR' means 17:40
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
Tene javascript impl would be nice, but would be a lot better after better 6model support.
okay fine sleep now really.
cotto Tene, do it now.
bubaflub goodnight Tene, goodnight moon
lucian scheme would be interesting indeed, i think there was already a partial implementation
Tene: g'night 17:41
cotto We'd need to be careful about M0/Lorito projects. M0 is deliberately too small to be implemented as a gsoc project on its own.
I suspect that the spec will be ready by the time gsoc ramps up.
lucian someone kick Tene :) 17:43
dukeleto I think M0 is too ill-defined and complicated to be a GSoC project, for the time being 17:44
let's not torture our students, at least not *that* much ;)
cotto dukeleto, right now yes, but I don't think it'll stay that way for long
once the spec is more or less ready, the actual implementation will be a 2-3 week project rather than a 2-3 month project 17:45
dukeleto cotto: yes, I think we are close, but would rather not rush anything. Slow and steady wins the race... 17:52
cotto dukeleto, of course 17:54
I think it'll be ready soon at a reasonable pace
cotto goes afk
lucian cotto: is there a wiki page?
dukeleto lucian: wiki page about what? M0 ? 17:55
lucian yep
dukeleto lucian: should be
lucian can't search for m0, too short
cotto lucian, there's a spec that contains the current state
m0-spec branch, pdd in drafts 17:56
afk for real
dukeleto lucian: there is also a 'lorito' branch which is a prototype in dynops, but it needs some love. I haven't had time in the last few weeks to update it with recent spec changes
lucian: M0 is the lowest level of Lorito 17:57
lucian not knowing what exactly lorito/m0 look like, i'm worried there will be too much reinvention of wheels
dukeleto lucian: which wheels are you worried that will be re-implemented? 17:59
bubaflub dukeleto: i'm thinking about doing the GMP bindings, where should i read about doing parrot level bindings and NCI stuff?
s/where/what/g
lucian dukeleto: jvm ones, mostly
lucian reads
dukeleto lucian: github.com/parrot/parrot/blob/m0-s...d32_m0.pod
lucian: we are basically refactoring parrot internals such that JIT is not impossible to implement 18:00
lucian: that is the gist
bubaflub is happy everytime he sees POD rendered in GitHub
dukeleto bubaflub++ for that
lucian dukeleto: the way i understand it, you're trying to implement everything in jittable stuff 18:01
18:01 whiteknight left
dukeleto lucian: we are basically creates the smallest set of opcodes that the rest of parrot can be implemented in 18:01
lucian: well, we want to minimize crossing the Parrot <-> C boundary, so basically, yes
bubaflub be back in an hour
lucian dukeleto: if you "forked" c, you could do that easily
18:02 bubaflub left
lucian is that what M1? wants to be? 18:02
dalek rrot/m0-spec: 6b78406 | dukeleto++ | docs/pdds/draft/pdd32_m0.pod:
PDDs no longer have versions
dukeleto lucian: M1 will basically be on the same level as PIR and will be implemented in terms of M0 ops
lucian: M0 is very low-level. Blobs of memory and indexing into them 18:03
lucian dukeleto: right. any reason to have m0 at all then?
dukeleto lucian: M1 will have a concept of objects and a MOP. M0 will not.
lucian wait, m1 ~= pir, misread that
and will m1 be assembly or structured?
dukeleto lucian: it currently looks like M0 will consist of somewhere between 30-40 opcodes 18:04
lucian: fixed width
lucian sounds reasonable
dukeleto lucian: which will make JITability possible. Very little Parrot <=> C boundary crossing, except when absolutely necessary (FFI/etc)
lucian finds GH's 404 funny 18:05
dukeleto lucian: indeed. those guys are hilarious 18:07
lucian dukeleto: so does M1 want to be more like jvm assembly, or more like C + parrot objects?
dukeleto lucian: i don't quite know what jvm assembly looks like, but M1 will mostly be what Parrot developers implement parrot core stuff with, if that makes sense 18:08
lucian dukeleto: it moves! that's the awesome bit
dukeleto: jasmin.sourceforge.net/
dukeleto lucian: M1 should be as pleasant to use as PIR *cough* ::ducks::
lucian dukeleto: that's exactly that i'd like to *not* see 18:09
i don't get why *anything* besides M0 should even resemble assembly
dukeleto lucian: sure. we can make M1 10x more pleasant than M0, if we want
lucian tower of increasingly complex assemblies seems silly 18:10
dukeleto lucian: it remains to be seen whether we will stop at M1 or if we need to have a candy-dipped M2 that comes with pink unicorns and bacon
lucian dukeleto: so why not retarget/fork winxed?
or finish Close, to target M0
dukeleto lucian: that could still happen. Nothing is impossible
lucian: Close is also a possibility, but it is currently bitrotten because the sole dev is not around anymore 18:11
lucian dukeleto: right
yes, i noticed
dalek tracwiki: v30 | lucian++ | GSoc2011
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
lucian dalek: i hate you
dukeleto lucian: the M0 spec is the most important thing now, how it will be implemented is still up in the air
lucian: once we have a solid spec, we will be in much better shape
lucian: does jasmin still work? 18:12
lucian dukeleto: i'm not sure a spec written in vitro will be very good
dukeleto: not sure. dalvik's asm is based on it, though, and that certainly works
dukeleto lucian: jasmin is still on CVS. yuck. 18:13
lucian dukeleto: oh, so much code is still in cvs/svn
dukeleto lucian: what do you mean by "spec written in vitro" ?
lucian or even git
lucian ducks
dukeleto lucian: yep, doesn't mean I can't complain about it :)
dukeleto loves to complain
lucian complaining is great!
dukeleto: i mean that the spec isn't targeted by a compiler 18:14
dukeleto I actually think we should just go back to posting tarballs and patches to Usenet. It worked great!
lucian: it will be targetted by a compiler, but how can a compiler target something that doesn't exist yet?
lucian dukeleto: by doing it incrementally, i guess. i'm just thinking of a compiler as a unit test 18:15
dukeleto lucian: we want a spec. we don't want to repeat perl 5's mistake of having no spec
lucian: there is atrodo++'s prototype that lives outside core 18:16
lucian dukeleto: how do you plan to validate it?
dukeleto lucian: github.com/atrodo/lorito/blob/master/ops.pod
lucian: validate our spec?
lucian yep
dukeleto lucian: i guess a working implementation is validation. Our spec will probably grow a test suite
lucian right 18:17
dukeleto lucian: atrodo++'s lorito pre-dates the current spec, so some stuff is different, but it really helped us in figuring out what we wanted for M0
lucian dukeleto: i should probably try to design a structured system language that targets the M0 spec, since i'm so fussed about it :) 18:36
dukeleto lucian: sure, works for me 18:38
lucian: and if you see any holes in the m0 spec, please let us know. Feel free to commit to that branch or send pull requests
lucian dukeleto: i'll most likely procrastinate away from it :) 18:39
anyway, i'll have a look anyway
tadzik . o O ( Not-Quite-Close ) 18:42
lucian tadzik: sounds good 18:43
tadzik that's actually what came to my mind a few minutes ago, whilst eating :) 18:45
dukeleto . o O (Pretty Close) 18:47
or even better, Pretty Close To Close (PCTC)
dalek tracwiki: v31 | dukeleto++ | GSoc2011 18:58
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
18:59 bubaflub joined 19:12 wknight-phone joined
dukeleto wknight-phone: new android phone? 19:28
dalek rrot/splint-quiet: f1f3fad | petdance++ | / (2 files):
pulled out the old flags
rrot/splint-quiet: a8862ff | petdance++ | splint01.log:
starting a new splint log tracking
rrot/splint-quiet: 0809f4e | petdance++ | / (2 files):
set two functions as pure. Use size_t instead of int in loop vars.
rrot/splint-quiet: 4ad0adc | petdance++ | splint02.log:
quiet warnings
rrot/splint-quiet: f8cda16 | petdance++ | / (3 files):
added new splint flags
rrot/splint-quiet: e5bfe34 | petdance++ | / (3 files):
added new flags
rrot/splint-quiet: 28cfe25 | petdance++ | / (3 files):
Merge branch 'master' of github.com:parrot/parrot into splint-quiet
wknight-phone dukeleto, in the last few months 19:29
dukeleto migrates back to his hotel from the coffee shop 19:30
19:33 zby_home left
dalek rrot/splint-quiet: 173b724 | petdance++ | / (3 files):
more flag reorganization
 19:38
19:44 wknight-phone left 19:46 dodathome left 19:47 dodathome joined 20:04 bacek joined 20:11 Andy joined 20:21 bubaflub left
dalek rrot/splint-quiet: cd0b569 | petdance++ | config/gen/makefiles/root.in:
save me some hassle
 20:31
rrot/splint-quiet: 25e7c49 | petdance++ | config/gen/makefiles/root.in:
quieting integer squawks
rrot/splint-quiet: 07d5250 | petdance++ | splint06.log:
quieted integer bummers
 20:34
20:42 M_o_C left
dalek rrot/splint-quiet: af6cde4 | petdance++ | s (2 files):
remove a potentially NULL path
 20:53
rrot/splint-quiet: 93ddd04 | petdance++ | / (3 files):
Parrot_ex_build_exception cannot actually return null
20:53 Andy left
cotto ~~ 20:56
21:23 lucian left 21:32 theory left 21:33 fperrad left 21:47 ambs left 21:58 dmalcolm joined 22:17 dmalcolm left, dmalcolm joined 22:21 dodathome left 22:42 kid51 left 22:49 dmalcolm left
dalek tracwiki: v32 | mj41++ | GSoc2011 22:52
tracwiki: TapTinder
tracwiki: trac.parrot.org/parrot/wiki/GSoc201...ction=diff
mikehh winxed (r856) builds on parrot 3_1_0-793-gb1fa14e - make test/test1/test2 PASS- ran some examples - - Kubuntu 10.10 amd64 (g++-4.5 with --optimize) 22:59
23:01 rurban_ joined 23:03 rurban left, rurban_ is now known as rurban
mikehh headed back to i386 - bbiab 23:13
dukeleto anybody know how to set a Key PMC to a value from C? 23:17
i have created the Key PMC, but I don't see how to actually set it to a value so I can use it to look something up
dukeleto reads the source of src/pmc/key.pmc 23:21
dukeleto cries 23:22
mikehh so sad 23:23
dukeleto i guess i need to set_string_native on it
mikehh: there are some sad comments in key.pmc
mikehh could be
hey dukeleto, I am happy to be a general mentor for GSoC - monitoring codetest etc and a specific one for anything that fits 23:24
cotto dukeleto, Parrot_key_new_X? 23:25
dukeleto mikehh: awesome!
cotto: that looks useful, thanks 23:27
cotto: now stuff is blowing up because I am creating a hash PMC, but it doesn't seem to be initialize
initialized, rather
mikehh dukeleto: I think I am registered through TPF, but would generally be more use for Pafo
dukeleto mikehh: once we orgs are accepted, we will figure it all out 23:28
cotto: what do I have to do, from C, to initialize a Hash PMC ?
cotto: Parrot_hash_new, i guess
cotto: i was using the Parrot_PMC_new(...) junk, which doesn't seem to initialize anything 23:29
cotto: it blows up less with Parrot_hash_new()
what does "# src/extend.c:127: failed assertion 's' 23:30
actually mean?
what is 's' ?
ah, i have seen the light 23:31
it is actually an error from Parrot_printf, duh.
dalek rrot/leto/embed_grant: b5c04c3 | dukeleto++ | t/src/extend_vtable.t:
[t] Parrot_PMC_defined_keyed_int
 23:34
dukeleto finally wrote a test that uses Key PMCs. The rest should be much easier. 23:35
dalek p: 82145f5 | jonathan++ | / (4 files):
Add a simple REPR that just stores values in a flat hash for Tene++.
 23:47
23:52 theory joined 23:57 benabik left

Powered by Cro and the Raku® Programming Language.

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

Thank you!