Parrot 1.7.0 "African Grey" is coming soon, no non-release-related commits to trunk, please | Test CallSignature PMC | pcc_reapply is almost there! trac.parrot.org/parrot/wiki/Callin...nsOverview
Set by moderator on 19 October 2009.
darbelo but I prefer manual addition, it let's you edit several files and only commit what's done. 00:00
dukeleto Whiteknight: you have to git add before committing. comit -a is convenient, but could be dangerous if you don't check what it did before committing
Whiteknight darbelo: I have to add it every time I modify it?
00:01 abqar joined
darbelo yep, and if you modify after your last add, that last change doesn't get commited. 00:01
dukeleto Whiteknight: no. just 'add' it once. the only exception is if you get merge conflicts in the file. if you resolve conflicts in an added file, you 'git add' it again to tell git that you resolved them
darbelo: hmm. i think you are righter. i use commit -a often, so I don't notice 00:02
Whiteknight: yes, 'git add' tells git 'add the current state of the file to the index'
Whiteknight: if you use 'commit -a' then it will basically re-add already-added files and use the current state of them in the commit 00:03
Whiteknight: git status -a is your friend, as well as a healthy .gitignore
darbelo Also, commit -a is closer to the svn workflow you might be used to.
except for the whole local/remote thing. 00:04
Whiteknight okay
so I make changes, review them, git add, git commit, git push
dukeleto Whiteknight: yes.
Whiteknight okay. that all makes good sense
dukeleto Whiteknight: also, pull before you push, unless you use the stash 00:05
cotto_work dukeleto, are you planning on cutting the release tonight?
dukeleto Whiteknight: i use this alias up = !git stash && git pull origin master && git stash pop
Whiteknight: so I can pull with dirty working copy (i.e. it has uncommitted changes) 00:06
cotto_work: i was planning on tomorrow morning-ish, but I am prepping now. anything I should know?
cotto_work nope. just curious
Whiteknight dukeleto:without stash, git pull wil clobber local changes? 00:07
darbelo Ohhh. we are passing tests in matrixy! 00:08
darbelo is so excited!
dukeleto Whiteknight: no, it will make fun of you and point 00:10
Whiteknight: git pull won't do anything if your tree is dirty
Whiteknight ok 00:11
00:11 KatrinaTheLamia joined
dukeleto Whiteknight: check out my .gitconfig, you might find some of it useful: github.com/leto/Util/blob/master/co....gitconfig 00:13
darbelo Whiteknight: Matrixy has a new test harness, have fun fixing the test failures.
dukeleto darbelo: exciting! i will start breaking it after I am done releasing 1.7.0! 00:14
KatrinaTheLamia hmm... okay... how does the African Parrot's air speed velocity compare to the African Swallow's air speed velocity. Of course, this is in Miloticean(sp?) space that this is occurred. XD XD
KatrinaTheLamia is shot
Whiteknight nice 00:15
cotto_work The African Grey is substantially larger and smarter than the African Swallow. I'
m sure it'd figure something out.
Whiteknight I like that like "praise = blame"
the philosphical implications are quite profound
darbelo git who-da-man 00:16
KatrinaTheLamia you know, there is going to be another group to make an implimentation of Parrot for their own purposes... and it will be named Sparrow. I mean, it is not rocket science to see this occurring. Sparrow will also likely be the test bed of some Python 3 (Python 4? o.o) projects. 00:17
00:17 pjcj joined
jonathan git: 'who-da-man' is not a git-command. # yeah, I tried, hoping for an easter egg... 00:18
cotto_work goes home
dukeleto Whiteknight: yeah, i try to give praise, but occasionly I blame
KatrinaTheLamia: really? 00:19
KatrinaTheLamia dukeleto, as a queen I must know these sorts of things XD
darbelo jonathan: nah, it was a joke on dukeleto's praise = blame alias
Infinoid Whiteknight: pong 00:20
KatrinaTheLamia naw, I just read the topic, and thought of something that at least I thought was funny--which is very rarely a requirement for others to think the same.
dukeleto is watching the Monty Python 'dead parrot' sketch for the first time right now
jonathan darbelo: I was fairly sure so, but that never stops me trying. ;-) 00:21
KatrinaTheLamia you know, thinking about it, the Dead Parrot sketch has a large amount of implications as far as Perl 6 is concerned XD
as well all know Perl 6 is just pining for the fjords.
KatrinaTheLamia is shot
oh right... I need to get back to work on learning Perl 6.. so at least my ignorant statements can have a slightly better basis to them. 00:22
jonathan shoots KatrinaTheLamia again, since it seems to be the done thing after jokes :-)
Whiteknight Infinoid: I've moved matrixy to github, but it doesn't appear in dalek
updated the Languages page on the wiki too 00:23
Infinoid Whiteknight: Yeah, I think dalek is just ignoring the link because it already had the old one
I'm just verifying that before I stab it.
Whiteknight okay, it's not a big deal, mostly just vanity for the new project
thanks for looking at it
Infinoid karma is good
purl is good has neutral karma
Infinoid Whiteknight++ # just coz I can.
KatrinaTheLamia jonathan, naw, I mostly just add "/me is shot" after things I say in a joking manner, because if I do not, there are other places on the web where I will be flamed until I is roasted snake. 00:24
darbelo We want the karma
Infinoid stabs dalek.
00:24 dalek joined
darbelo That works? 00:25
Infinoid++
KatrinaTheLamia also... kind of wondering. I have yet to look into it, as I said I would over a month ago. Is Perl 6's threading capabilities part of Perl 6--or is it part of Parrot? I forgot after the conversation where I suggested a few fixes that could be made to how Perl 6 does its threading.
Infinoid darbelo: now commit something clever. :) 00:26
Whiteknight KatrinaTheLamia: Parrot provides all the functionality and primitives 00:27
jonathan KatrinaTheLamia: I'd expect Parrot will provide primitives that the Perl 6 threading model can be built on top of.
However as Parrot last I checked currently seems to struggle to spawn a thread inside Rakudo without segfaulting...I fear we have a little way to go.
KatrinaTheLamia you know, speaking of karma, the only sites where I have found karma system to actually work are PerlMonks and porn sites. Is there a corrolation to be made here?
Whiteknight++
jonathan++
thanks btw for helping me in the right path. 00:28
jonathan KatrinaTheLamia: Well, only in that you visit both. ;-)
Whiteknight karma works on porn sites? 00:29
purl works on porn sites has neutral karma
Whiteknight thanks purl
jonathan (works on porn sites)++
KatrinaTheLamia I was mostly going to make the corrolation that Perl is inherently smexy.
Whiteknight yes, Parrot's threading system needs a lot of work
but, in theory, it works great and provides everything Perl6 needs
KatrinaTheLamia Whiteknight, well, with porn sites it is simple... you post nekkid pics, your karma goes up.
jonathan The same theory where Parrot is lightning fast and never crashes ever? ;-)
KatrinaTheLamia alright--I will look into it. 00:30
Whiteknight KatrinaTheLamia: With me, I suspect posting nudes would cause my karma to go down. But I see what you are saying
KatrinaTheLamia may be simpler to look into Parrot's threading system, rather than Perl 6's retarded idea of copying the entire program's instance and interpreter to do it.
Whiteknight that's Perl 5's system
jonathan KatrinaTheLamia: Huh? That was, erm, Perl 5's idea. 00:31
KatrinaTheLamia Whiteknight, well, it is rarely of the person themselves that was posted.
jonathan KatrinaTheLamia: We're *so* not doing that one in Perl 6.
I hpoe...
:-)
darbelo waits for the karma
KatrinaTheLamia jonathan, Whiteknight, ah, okay--I was hoping you were not doing that. Despite having it suggested that was how it works in another perl channel. If anything, I would have started working on a method to do it better, in such a case. 00:32
Whiteknight adds "ake", "mak", and "mae" as aliases to "make" out of sheer frustration
perl5 and this implementation of perl6 are very diferent
darbelo ksh: mkae: not found
Whiteknight jonathan: just added :call_sig! 00:33
(well, in a patch anyway)
Infinoid watches dalek suspiciously
KatrinaTheLamia alright--going to work on actually getting Parrot to compile (I had issues doing that when I tried midSeptembre) 00:34
KatrinaTheLamia I am safe from the dalek... as I am not human.
Infinoid hmm, is dalek broken?
dukeleto Whiteknight: i alias 'm' for make :) 00:35
KatrinaTheLamia impossible! Daleks are perfection! They cannot be stopped!
dukeleto Infinoid: i just pushed some commits to matrixy and didn't see anything
KatrinaTheLamia sorry... that was just the first thing that popped into my mind... I apologise for my geeking out. 00:36
japhb KatrinaTheLamia, sure they can. You just have to be Bad Wolf.
nopaste "Whiteknight" at 69.249.176.251 pasted "patch to add :call_sig for jonathan++" (173 lines) at nopaste.snit.ch/18378
Whiteknight put that in your pipe and smoke it
Infinoid it says it added it as a github feed... just not sure why it hasn't emit any output
jonathan Whiteknight++ # nice!
Whiteknight that'll get committed after pcc_reapply lands in trunk and gets settled 00:37
trying to keep the branch diff as small as possible (which isn't very small)
jonathan Whiteknight: This is for calling with one, or receiving it on the other end, or both? 00:38
KatrinaTheLamia japhb, I dunno... I vaguely remember the Daleks coming back after Bad Wolf wiped them from existence. Pissed off at Bad Wolf, they set a course of events that made Bad Wolf forever seperated from the Doctor. Pretty bad ass there.
Whiteknight jonathan: calling with one 00:39
hold on, demo coming
japhb KatrinaTheLamia, IIRC they had the help of the creator of the Daleks to do that ... they couldn't do it themselves.
jonathan Whiteknight: Hmm.
japhb But it's been a while since I watched that season.
jonathan *accessor->pmc(interp, arg_info, 0) = call_object;
Where does that get the call object from?
nopaste "Whiteknight" at 69.249.176.251 pasted ":call_sig example for jonathan++" (11 lines) at nopaste.snit.ch/18379 00:40
jonathan That is, we aren't constructing a CallSignature with it in, and pulling it out of that, are we?
KatrinaTheLamia japhb, well, part of being bad ass is known who to ask for help from.
Whiteknight jonathan: Parrot builds the CallSignature in the set_args opcode
and calls this function, fill_params, inside the get_params opcode
jonathan Ah, sorry, I had it on the wrong end. :-)
japhb There was the "crossing over from another universe" arc, but that's hardly fair, since they Daleks of *that* universe hadn't had to fight Bad Wolf.
Infinoid Hmm, it's also possible that 33 feeds is a little too much for a poor simple bot. Have the rest of them seemed to be holding up recently?
Whiteknight set_params eventually will be able to take a CallSignature or some other type to act as the sig
Infinoid: yes, everything else is working fine 00:41
jonathan So that line I pasted is what's setting the register receiving the CallSignature.
OK, got it, and great.
In fact, that's the most helpful side of it for me for the moment too. :-)
Whiteknight++
Whiteknight jonathan: can obviously add more functionality pretty quickly. Just let me know what you need
(the new system is TEH M4J0R R0XX0R 00:42
jonathan Whiteknight: This looks great for now. 00:43
Whiteknight I've got most of the parser nonsense in place, could probably add it on the returns side too if that was needed for some reason
jonathan Whiteknight: Right now, I build a slurpy array and slurpy hash and give them to the binder. 00:44
Being able to not do that, and just give it the call signature object, will be sweet.
And hopefully a performance win at some level.
dukeleto needs stuff to put in NEWS 00:45
jonathan (e.g. it's one less PMC to create)
dukeleto starts to pour of parrot-commits for the last month
s/of/over/
dalek rrot-plumage: 47689e5 | darbelo++ | :
Add metadata file for matrixy.
 00:46
shorten dalek's url is at xrl.us/bfs6y9
Infinoid oh, so now suddenly parrot-plumage works, but matrixy doesn't? 00:48
silly bot.
purl *giggle*
dukeleto hmm
darbelo Infinoid: plumage has been working for a while now.
japhb Infinoid++ # And thank you for that ... 00:50
Infinoid great. last time I looked at the bot, it seemed suspiciously silent
which, come to think of it, seems strangely familiar
maybe matrixy will spontaneously start working too :)
Whiteknight Do I need to add any special hooks or anything to the matrixy project?
Infinoid no. the github link on Languages is all dalek should need 00:51
japhb darbelo, OK, I believe I've pushed a parrot-plumage fix for you. Give it a try.
Infinoid and it says in the log that it's tracking the atom feed... maybe it's just slow.
dalek rrot-plumage: 64d797d | japhb++ | :
[plumage] Fail entire command if install_required_projects() fails
shorten dalek's url is at xrl.us/bfs6zy
dalek trixy: 9732075 | (Duke Leto)++ | CREDITS:
Add myself to CREDITS
 00:52
shorten dalek's url is at xrl.us/bfs6z2
darbelo Infinoid: Oh, and before I forget. Can you teach dalek about trac.parrot.org/parrot/wiki/Modules too?
dukeleto do we have any new functionality in 1.7.0 ?
Infinoid darbelo: should it be handled the same way as Languages?
Whiteknight YAY! 00:53
darbelo Yeah, the link points to the repo, same as in languages.
Infinoid cool. hopefully it won't generate conflicts with the existing decnum-dynpmcs parser 00:54
darbelo You can stop special-casing decnum-dynpmcs too.
darbelo is going to hunt down some food now. 00:55
japhb: I'll try your fix later.
japhb darbelo, roger that
00:55 darbelo left
japhb IWFM 00:55
Whiteknight Modules would be good, cause I want to see changes to Parrot-Linear-Algebra too 00:56
dukeleto dammit, no I have karma going to dukeleto, leto and "Duke Leto". madness 00:57
s/no/now/
dukeleto should L to T
should a function being added to the Parrot API (like this: trac.parrot.org/parrot/changeset/41897) be documented in NEWS ? 00:58
dukeleto is going to be asking a lot of questions today and tomorrow
Whiteknight dukeleto 00:59
purl i think dukeleto is still employeed. i "made the cut" in our CEO's words
Whiteknight yes
dukeleto lulz re: purl
Whiteknight dukeleto: what you do is go through the SVN log and add whatever you think is noteworthy, based on no particular standard
if people want things added or clarified, they can do it their damn selves 01:00
01:00 whoppix joined
Infinoid dukeleto: do you have karma aliases in CREDITS? 01:01
dalek rrot-plumage: 61f101b | japhb++ | :
[META] TODO updates
 01:02
shorten dalek's url is at xrl.us/bfs634
dalek rrot-plumage: 36e5520 | japhb++ | :
[META] TASKS: Include status for partially implemented TASKS
shorten dalek's url is at xrl.us/bfs636
dukeleto Infinoid: which CREDITS do I need to alias my karma? 01:08
Whiteknight: that seems to be the current situation
Infinoid parrot/CREDITS
I'd do it, but it may be a few minutes... looks like my git-svn has a couple months of catchup to do.
Whiteknight dukeleto: that's how it is every month 01:09
peopel should be updating CREDITS all month, butnobody does
Infinoid I haven't updated CREDITS at all, because I haven't done anything. :)
Whiteknight purl msg chromatic if we extend fastcore functions to take a Parrot_Context** pointer to a context structure, we can avoid all those repeated calls to Parrot_pcc_get_context_struct. Ops that modify context could update the pointer. 01:11
purl Message for chromatic stored.
Infinoid that sounds like something that should be buried deep within a typedef 01:12
Whiteknight Contexts are PMCs now, but we need to access the guts of that PMC inside almost every single op
01:13 Zak joined
Whiteknight if we can cache a pointer to the pointer, we can avoid having to fetch the pointer each time from the PMC 01:13
it's a theory, anyway
and I suspect those repeated calls to Parrot_pcc_get_context_struct create a substantial slowdown for Parrot
dukeleto Hunk #1 succeeded at 1 with fuzz 2 <-- always makes me laugh 01:15
japhb "Hunk with fuzz" sounds so 70's
dukeleto wonders if larry came up with the terminology 'hunk' or if it pre-dates 'patch'
Whiteknight Larry is a total hunk, so it was a self-reference 01:16
Infinoid but how fuzzy is he?
dukeleto to add a karma alias, I just add more U: lines to CREDITS?
Whiteknight I've never had the pleasure to find out :)
Infinoid dukeleto: A: leto, A: Duke Leto
s/, /\\n/
dalek rrot: r41944 | dukeleto++ | trunk/docs/project/release_manager_guide.pod:
[cage] Update release manager guide about committing to trunk near a release, based on cotto++'s advice
 01:17
01:17 athomason joined
Infinoid it'll map your A: lines to the name from your U: line 01:18
cotto I thought that advice was already there. Maybe it was just for a "supported" release. 01:19
either way, free karma
dalek rrot: r41945 | dukeleto++ | trunk/CREDITS:
[cage] Add some karma aliases for myself
 01:20
01:21 dalek joined
dukeleto should I put stuff like 'partcl moved to github' in the NEWS for the release? 01:21
Whiteknight purl msg allison I think I have :call_sig implemented. Check out nopaste.snit.ch/18378 for the patch and nopaste.snit.ch/18379 for a quick test case. I'll apply after pcc_reapply lands, unless you spot any obvious problems 01:22
purl Message for allison stored.
Infinoid msg darbelo I've killed the decnum-dynpmcs special case. Please ping me if it doesn't still work. 01:23
purl Message for darbelo stored.
cotto Whiteknight, ideally without the debug printf? 01:25
;)
Whiteknight ah yes. Good catch
I have some tweaking to do in any case, so i need to scan it all very closely
cotto also, wouldn't it be easier to compare arg_flags[i] with ~VT_CALL_SIG in case other flags get added or removed? 01:27
(in the start of the pcc.c chunk
)
Whiteknight++ for giving me an excuse to dip my toes into the pcc code 01:29
Whiteknight it might be, yes 01:31
if it has VT_CALL_SIG, it can't have anything else, and vice-versa 01:32
so I thought that was a decent short-circuiting check
cotto minor nit, at any rate
Whiteknight at this point, any change I make is going to have "PERFORMANCE" at the front of my thoughts 01:33
cotto patching from nopaste got much easier once I remembered that I could copy/paste to stdin
Whiteknight yeah
I don't like the extra call to VTABLE_get_integer_keyed_int in fill_params, but I can't think of any good way around it 01:36
cotto how expensive is it? 01:38
dalek kudo: 9d76f3b | (Solomon Foster)++ | src/setting/Any-num.pm:
Add proto declarations for all the trig functions, allowing you to call them using named arguments.
 01:39
shorten dalek's url is at xrl.us/bfs7e3
cotto also, what's the target to regenerate imcc's generated files? 01:40
Whiteknight it's not too expensive, just annoying that we have to fetch the same value twice on a false positive
and ideally I would like to do this in a way that doesn't require an additional if() check on every loop iteration 01:43
I could change it so we call VTABLE_get_integer_keyed_int above the loop and then at the bottom of it. That would reduce the extra calls, at the expense of a little more code complexity on this already-bloated function 01:47
although I've got some ideas to refactor it that would still be O(1)
cotto probably not worthwhile unless a profiler tells you otherwise 01:48
Whiteknight every little bit is worthwhile, when Parrot is 4x slower then it was a few months ago 01:51
but you're right, there are probably better fat deposits to jazzercise for now
...and on that note, I'm off to bed. Goodnight 01:53
02:07 JimmyZ joined 02:42 janus joined 03:04 bacek joined
bacek o hai 03:06
JimmyZ good localtime, bacek. ;) 03:07
bacek G'Day JimmyZ
KatrinaTheLamia uh... issue with github... where do I go to ask about it? 03:10
dalek rrot: r41946 | bacek++ | branches/pcc_reapply/config/auto/frames.pm:
Temporary disable buildframes detection.
 03:36
03:44 bacek_ joined
dalek rrot: r41947 | bacek++ | branches/pcc_reapply (2 files):
[cage] Get rid of old call_state structs and enums
 04:25
05:11 particle joined
cotto dukeleto, ping 05:45
dukeleto cotto: pong 05:53
cotto dukeleto, I have a patch to fix some compiler warnings in the profiling runcore (though there's still one) 05:55
ok to apply?
dukeleto cotto: go for it 05:56
05:59 uniejo joined
cotto dukeleto, done 05:59
It's not ideal, but it's more a stylistic nit than functional. 06:00
dukeleto cotto: i will probably try to release sometime mid-morning-ish to mid-afternoonish, Pacific time
dalek rrot: r41948 | cotto++ | trunk/src/runcore/profiling.c:
[build] address some compiler warnings
cotto ok 06:01
dukeleto i am assuming that bumping PARROT_MAX_ARGS should be in NEWS? 06:08
cotto your call. I don't think it's a huge deal. 06:09
06:11 bacek joined
dalek rrot-plumage: 9cc0f6c | japhb++ | :
[TOOLS] WIP: import_proto.p6 (import proto project information into Plum...
 06:39
shorten dalek's url is at xrl.us/bfs8dp
japhb Time to catch a few Z's, I think 06:40
'night, all.
07:03 masak joined
dalek p-rx: 49572a7 | pmichaud++ | src/ (4 files):
Allow multiple alias subrule bindings (e.g., <abc=def=subrule>).
 07:05
shorten dalek's url is at xrl.us/bfs8es
07:14 Zak joined
dalek p-rx: b0d3c56 | pmichaud++ | src/ (3 files):
Make use of <OPER=...> bindings to promote <prec> and <sym> to
 07:17
shorten dalek's url is at xrl.us/bfs8fp
07:26 fperrad joined 07:30 eiro_ joined
mikehh dukeleto: ping 07:32
dukeleto mikehh: pong 07:34
mikehh dukeleto: I am getting bignum and bigint tests skipped in t/arithmetics_pmc.t even though I have gmp 07:36
dukeleto mikehh: this is since that file has been translated to PIR?
mikehh: so everything passes, but it should be running the tests, but is not?
mikehh obviously not reporting failure but I don't see why they are skpped 07:37
if I remove the 'gmp' test the test passes - for me anyway and gmp should be 'defined'
dukeleto mikehh: perhaps the check for gmp is broke? 07:38
mikehh yeah but dunno why - it should work 07:39
dukeleto mikehh: which platform?
purl I'm running on OS/2 on an Atari, can you help?
dukeleto purl, die in a fire
purl HALP
mikehh Ubuntu 9.10 (beta updated) amd64
dukeleto mikehh: does the same thing happen on other platforms?
mikehh dukeleto: dunno 07:40
dukeleto mikehh: this definitely needs to be fixed, but I am not sure i want to tweak how that test works so close to the release. I haven't played with that stuff much before.
mikehh dukeleto: what are you on 07:41
dukeleto mikehh: can you make a TT?
mikehh: i am on OS X 10.5, but have access to linux/freebsd
mikehh: i am also pretty close to going to sleep
purl okay, dukeleto.
mikehh dukeleto: what does prove -v t/arithmetics_pmc.t give you? 07:42
on either platforn
platform
dukeleto mikehh: it skips for me too, when it shouldn't
mikehh: this is somewhat my fault. I reviewed and applied the patch that converted those tests. No good deed goes unpunished... 07:43
mikehh: perhaps config['gmp'] is not being set correctly? 07:44
mikehh dukeleto: it should be - but obviously not 07:45
07:45 TiMBuS joined
dukeleto mikehh: this is definitely an issue we need to take care of, nice job noticing it. i don't think it is going to be part of 1.7.0 though. 07:46
mikehh dukeleto: I don't know how to check the (iterate over) the keys in config = interp[.IGLOBALS_CONFIG_HASH] without specifying the key - i.e. list all the keys (and values) 07:47
dukeleto mikehh: there should be some kind of iterator for hash keys
mikehh: but I don't know the syntax off the top of my head 07:48
mikehh: t/pmc/orderedhashiterator.t may be of use 07:49
mikehh: we should be able to create an iterator on config and assert that things in there are correct. this would be a really good test....
mikehh dukeleto: all the examples I have seen try with known keys - not to list the keys in the hash 07:50
dukeleto mikehh: huh. yeah, we definitely need to test that
mikehh: i am going to sleep. will catch up in the morning. keep up the good work! 07:51
bacek mikehh: "iter SomeHash" 07:54
just check t/pmc/hashiterator.t for exact syntax
mikehh bacek: was looking - the tests for it all use specific keys - I just want to list all the key/value pairs 07:56
bacek it = iter Hash; unless it goto done; thingy = shift it; key = thingy.'key'(); value = thingy.'value'() 07:57
mikehh: src/pmc/hashiterator.pmc, top of the file
mikehh ok will try that
bacek: helps if you look in the right place :-} 08:01
bacek mikehh: I just know it because I reimplemented whole Hashes in parrot some time ago :)
mikehh let me set up a TT and see how far I can get (for the test that is) fixing it 08:05
allison msg whiteknight The patch looks good. A few things: Make the flag Pk instead of K (it is a PMC, so should be a modifier on the PMC flag). Then move the semantic checking into src/call/args.c instead of in IMCC (so, all IMCC should do is set the flag like it does for the others). 08:11
purl Message for whiteknight stored.
allison msg whiteknight Also add a reverse test case for "foo($P0 :call_sig)" 08:13
purl Message for whiteknight stored.
moritz oh hai 08:35
I just heard about a nice idea from the DebConf people 08:36
for their IRC meetings, they have a bot which you can tell meeting results
something along the lines of protocoldroid: record improve test coverage of CallSignature PMC 08:37
which makes it trivial to obtain a summary from the last meeting
would you want me to investigate how to get such a bot, or maybe write one, for #parrotsketch?
08:40 bacek_ joined
mikehh bah - trac is not talking to me 08:42
allison moritz: sounds useful 08:43
mikehh if fact parrot.org is taklking to me apart from the irc 08:54
is NOT talking
moritz and irc.parrot.org is just an alias for irc.perl.org 08:55
mikehh well I can't seen to access trac or svn
moritz the rest of the website seems also down 08:56
mikehh that's all I've tried
I managed to fix the test t/op/arithmetics.t and now I can't commit it 08:57
svn: OPTIONS of 'svn.parrot.org/parrot/trunk': SSL negotiation failed: Secure connection truncated (svn.parrot.org) 08:58
anyway I will try in again soon - got to go out for a bit - bbl 09:00
09:57 donaldh joined
dalek kudo: f5b38a0 | masak++ | src/builtins/guts.pir:
[src/builtins/guts.pir] arrays/hashes & handles

traits. These now behave as any other attribute, as per spec change r28846.		 10:04
shorten dalek's url is at xrl.us/bfs8pq
10:06 jsut_ joined
dalek p-rx: 4e5addc | pmichaud++ | src/ (4 files):
Handle postcircumfix: parsing.
 10:19
shorten dalek's url is at xrl.us/bfs8qv
dalek p-rx: 9bf6977 | pmichaud++ | src/ (3 files):
Handle postcircumfix: ast requirements.
shorten dalek's url is at xrl.us/bfs8qx
dalek p-rx: bcd5543 | pmichaud++ | src/cheats/hll-grammar.pir:
Correct stacked postfixish operators.
shorten dalek's url is at xrl.us/bfs8qz
mikehh parrot.org still seems to be down - can't access trac or svn 10:27
and it's definately not my internet connection - I have just done a distribution update on Ubuntu 9.10 beta (101.9MB) 10:32
10:34 plobsing joined
mikehh I think I need to re-boot - bbiab 10:38
dalek p-rx: 7591d1c | pmichaud++ | src/ (4 files):
Add list associativity.
 10:44
shorten dalek's url is at xrl.us/bfs8sr
10:44 mikehh joined
mikehh msg dukeleto - I got a fix for t/op/arithmetics.t - I just can't connect to svn.parrot.org at the moment 10:53
purl Message for dukeleto stored.
10:56 cconstantine joined
mikehh msg duckleto that is t/op/arithmetics_pmc.t - the .local int gmp is not an int - it's a string - 'defined' -> .local string gmp 10:57
purl Sorry, I've never seen duckleto before.
mikehh msg dukeleto that is t/op/arithmetics_pmc.t - the .local int gmp is not an int - it's a string - 'defined' -> .local string gmp
purl Message for dukeleto stored.
10:58 nbrown joined
mikehh bah 10:58
nothing works for parrot.org - it's down 10:59
11:16 cconstantine joined, payload joined 11:40 payload1 joined
dalek kudo: 1ab0699 | (Solomon Foster)++ | src/setting/ (2 files):
Remove troublesome atan2 is export declarations to make atan2 work properly.
 11:42
shorten dalek's url is at xrl.us/bfs8wp
Coke sd 11:45
ww 11:48
mikehh parror.org still down - going to lunch - bbl 11:56
12:00 payload joined, whiteknight joined
whiteknight good morning #parrot 12:03
plobsing hi, whiteknight! 12:11
whiteknight hello plobsing!
plobsing when trac comes back up, can you test the latest patch I attached to TT #1105?
it adds support for non-386, non-patched libjit
so it should work on your machine 12:12
whiteknight trac is down?
I'm on x86 here at work, have x64 at home 12:13
plobsing for me at least. I can't get to anything on parrot.org right now
seems like mikehh has the same problem 12:14
whiteknight yeah, me too 12:16
DNS finds it correctly, so server must be down 12:17
other projects hosted by OSU appear to be working correctly 12:19
Coke allison sent a ticket to OSUOSL qbaout it. 12:23
(2.5 hours ago) 12:24
looks like an HTTP issue.
12:28 bluescreen joined
whiteknight ok 12:28
Coke i fan a fulltest on osx yesterday and got an error. 12:31
(missing test files, IIRC)
... I would svn up and double check, but svn is over http. :| 12:35
whiteknight awesome
Coke has dukeleto said when he's doign the release? 12:36
whiteknight no idea
Coke parrot release?
whiteknight the /topic says it's coming soon, but no idea about the actual release time 12:43
dukeleto: ping
12:44 ruoso joined
Coke http back. 12:56
whiteknight w00t 12:59
plobsing: so how do I apply these patches? Specific order? 13:11
or is that latest patch against branch?
13:12 bluescreen joined
dalek rrot: r41949 | mikehh++ | trunk/t/op/arithmetics_pmc.t:
fix test - change int to string
 13:35
mikehh seems to working again
dukeleto said earlier that he was ready to do the release but was going to get some sleep first 13:37
and was going to do it this afternoon PST 13:39
Coke Test Summary Report 13:46
-------------------
t/op/exit.t (Wstat: 0 Tests: 7 Failed: 0) TODO passed: 6
no clue which core that was.
(I hate fulltest. =-)
dalek TT #1124 created by mikehh++: t/op/arithmetics_pms.t is skipping bigint/bignum test even when gmp is ... 13:51
Coke t/codingstd/test_file_coverage.t .. Dubious, test returned 1 (wstat 256, 0x100) 13:52
(is no one else seeing this fulltest error?)
mikehh Coke: running now - and logging :-} 13:58
I do something like make -k fulltest TEST_JOBS=5 2>&1 | tee fulltest.41949.amd64.log 13:59
Coke ok. you know that with -k you'll have to go back through and manually find the errors, yes? 14:02
mikehh sure - I load it into the editor and look for Result: then check for any problems 14:04
it has got to codetest now :-} 14:06
dukeleto 'ello 14:10
whiteknight: pong
Coke dukeleto: hio. seeing a failure in t/codingstd/test_file_coverage.t; probably skippable for the release. 14:11
whiteknight dukeleto: ETA on the release?
dukeleto whiteknight: just woke up, ENOCOFFEE
14:11 ruoso joined
whiteknight dukeleto: Okay, I'll open a coffee ticket, you go work on a coffee patchset 14:12
dukeleto whiteknight: do you want to apply stuff or just curious? All I can say is 'today' right now
mikehh All tests PASS (pre/post-config, smoke (#29222), fulltest) at r41949 - Ubuntu 9.10 (beta updated) amd64
whiteknight dukeleto: just keeping track. I can do very very limited helping if you need
mostly win32 testing
Coke dukeleto: ping me when you do the release; I can do the docs.parrot.org step (I seem to be the SPOF there.) 14:13
purl I can't find me in the DNS.
Coke (if you have access to ssh to the parrotvm box, I can step you through it.)
dukeleto Coke: interesting. I didn't know about the docs.parrot.org 'step' 14:14
Coke: and I don't think I have ssh access to parrotvm, but who knows?
mikehh Coke: that test passes for me
Coke dukeleto: not i. =-)
moritz is it in the release guide? 14:15
mikehh dukeleto: that problem with t/op/arithmetics_pmc.t is fixed for me now - can you check it on your box (r41949) 14:16
dukeleto mikehh: yep, checking now
14:18 hercynium joined
dukeleto mikehh: yep, the gmp tests now run and pass. nice work! mikehh++ 14:19
whiteknight: i would appreciate win32 testing, since I don't have any of those 14:20
whiteknight purl msg desertm4xI worry that CBLAS is hard to get. I don't think there are any binaries or prebuilt packages for it, and I've had some problems building from source. Maybe we should look at an implementation that's easier to get. GSL comes to mind.
purl Sorry, I've never seen desertm4xi before.
whiteknight purl msg desertm4x I worry that CBLAS is hard to get. I don't think there are any binaries or prebuilt packages for it, and I've had some problems building from source. Maybe we should look at an implementation that's easier to get. GSL comes to mind.
purl Message for desertm4x stored.
dukeleto whiteknight: i would say, once coffee is involved, i assume that Parrot 1.7.0 will get released around early afternoon, Pacific time 14:21
Coke dukeleto, moritz : it's there now. =-)
dukeleto Coke: in trunk or branch?
Coke dukeleto: the docs? in trunk.
dukeleto Coke: ok, i see it 14:22
i am currently on step c) of the release guide. must. find. coffee. 14:23
Coke I avoided listing explicit hostnames and directories there, perhaps in an overabundance of paranoia.
dalek rrot: r41950 | coke++ | trunk/docs/project/release_manager_guide.pod:
Add brief outline of docs.parrot.org update step.
14:25 davidfetter joined 14:27 mikehh joined
dukeleto so, can I add myself as a core dev in RESPONSIBLE_PARTIES, or is there some kind of process for that? 14:32
sjn + 14:33
whiteknight dukeleto: no process. Just add it 14:34
dukeleto whiteknight: just do it. check.
whiteknight I would avoid stuff like "supreme leader of the Parrot multiverse", but anything you reasonably want to add is fine
dukeleto does anybody know the status of PBC_COMPAT? are we still PBC-compatible? 14:35
whiteknight: Lead Parrot Troll
whiteknight I thought that was Purl?
dukeleto whiteknight: you may be right. 14:36
dukeleto notes that 'make distro_tests' pass. yay
so did we rename or remove any ops this release-month? i don't THINK so ... 14:37
particle diff it!
purl i guess diff it is :)
dukeleto particle: easiest way to find the svn rev of the previous release in svn ? 14:40
wow, the steps in part 10) about docs.parrot.org really need to be automated
whiteknight for serious 14:41
Coke dukeleto: er... just get the /tag/ of the previous release.
(why need the svn #?)
dukeleto Coke: touche. I blame lack of coffee. 14:42
14:46 ash_ joined
whiteknight dukeleto: stop complaining about lack of coffee! Go make some coffee a reality 14:46
whiteknight is certain that somewhere on the west coast there is a starbucks 14:47
on Sunday I went to a shopping center that had two of them, only a few buildings apart
14:48 Andy joined
dukeleto whiteknight: starbucks is sacrilege, but let's not go there :) I am crashing at a friends and they are not up yet. I will soon shake them awake while screaming NEED COFFEE 14:48
jonathan whiteknight: Go to San Francisco. It felt impossible to walk more than a couple of blocks without seeing a Starbucks there.. :-)
whiteknight oh right, I forgot that you're in town for the GSoC thing
ash_ isn't that like a sign of the apocalypse? having a starbucks across the street from a starbucks
whiteknight according to Lewis Black that is THE sign of the apocalypse 14:49
ash_ ah, thats where i heard that
dukeleto jonathan: i am actually in San Francisco, but there is better coffee here : 14:50
:) even
Coke was in leeds a few weeks ago, and standing in the middle of the road, could see a starbucks on one side, and then turning 180deg, another starbucks. 14:53
jonathan dukeleto: Oh, I'm sure there is. :-)
Coke has only ever had that happen in Houston, TX, before.
whiteknight purl msg desertm4x ATLAS would be a good option too. I am seeing a few BLAS packages around but they are all based on the non-optimized netlib reference version. ATLAS should provide better performance then that (and I see a few packages for it around too) 14:55
purl Message for desertm4x stored.
14:56 Psyche^ joined
Coke someone should consider re-formatting PLATFORMS now that we don't have really have a JIT. 14:59
I also see lots of ?? for many options.
dukeleto Coke: what do you recommend? and as part of this release or after?
dalek rrot: r41951 | coke++ | trunk/PLATFORMS:
parrot doesn't install on this platform.
dukeleto wow, svn diff from the last release is still running. svn diff is sloooooooow 15:00
Coke dukeleto: if I had a recommendation, I'd've made it. =-)
your judo won't work on me. =-)
15:05 hudnix joined
Coke going to remove any PLATFORM entries older than a year. any objections? 15:08
ash_ how are the jit changes coming along?
Coke ash_: I don't think anyone has touched JIT recently.
(most devs have been concentrated on refactoring the calling conventions) 15:09
ash_ ah, thats cool, i just remember on whiteknights blog posting something about libJIT NCI frame builder stuff 15:10
dukeleto Coke: do it 15:15
renesd.blogspot.com/2009/09/py3kpyt...n-096.html <-- interesting read. Was looking for Parrot's pynie to be mentioned, but nope 15:16
shorten dukeleto's url is at xrl.us/bfs9om
dukeleto how should I describe the JIT deprecation? 15:22
particle sadly. 15:23
jonathan "With a tear in our collective eye, we removed the that caused us to stab the other eyes we used to have out."
dukeleto lulz 15:24
jonathan *the code...
Coke dukeleto: (do it) rip out year old platform entries?
moritz "with a happy heart, and a tear in either eye, I'm proud to announce that we killed JIT with fire"
dukeleto Coke: yes
Coke done, along with another minor cleanup. 15:28
dalek rrot: r41952 | coke++ | trunk/PLATFORMS:
remove results older than a year.
 15:29
rrot: r41953 | coke++ | trunk/PLATFORMS:
remove 2 unused comments, renumber remaining comments.
dukeleto particle: diff from 1.6.0 to trunk is 84144 lines 15:31
moritz yikes
dalek rrot: r41954 | whiteknight++ | trunk/PLATFORMS:
update win32-x86-mingw_gcc3.4.5 to current date. All tests pass locally
 15:32
15:33 mikehh joined
dalek rrot: r41955 | whiteknight++ | trunk/PLATFORMS:
[PLATFORMS] remove JIT column from platforms listings. Don't have JIT, don't really need to report it's status anymore. Can update this later once we know what the JIT implementation will look like.
 15:36
15:37 athomason joined
particle dukeleto: for src/ops/ops.num? 15:37
sorry, i should have been more specific. 15:38
svn diff -rREL_1_6_0:HEAD src/ops should give you the info you want
Coke particle: you can't, IIRC, use tags as revision numbers. 15:40
dukeleto particle: i think that helps, let me see if I can get a diff of only src/ops 15:41
svn diff hates me 15:43
when I do "svn diff svn.parrot.org/parrot/tags/RELEASE...0/src/ops/ svn.parrot.org/parrot/trunk/src/ops" it shouls pic.ops as being removed, but that was described in the last release. what gives? 15:44
s/shouls/shows/ 15:45
15:54 nopaste joined
dalek kudo: f76c57e | moritz++ | docs/ROADMAP:
[docs] update ROADMAP

  * biding positionals by name now works		 15:54
shorten dalek's url is at xrl.us/bfs9wd
mikehh what is the Inst column in PLATFORMS
Install? 15:55
purl i heard Install was extremely finicky...i ended up using the RPM, but of course that is just pure trash...I am through with postgres!!!
whiteknight mikehh: yes, install I think
there should be an explanatory note lower in the file
yes, it's "make install"
mikehh well ok I am updatein amd gcc 4.4.1 as installing
whiteknight ok 15:56
mikehh ok then - got to go out for a bit - bbiab 15:58
15:58 theory joined
dalek rrot: r41956 | mikehh++ | trunk/PLATFORMS:
add linux-amd64-gcc.4.4.1
 15:58
p-rx: 7451fb8 | pmichaud++ | src/ (3 files):
Allow nextterm specifier, nulltermish tokens.
 16:00
shorten dalek's url is at xrl.us/bfs9xc
16:04 nopaste joined 16:06 allison joined
Coke whiteknight: if all tests pass, why do you have it marked as one failure? 16:07
whiteknight where, in PLATFORMS?
Coke yes.
trac.parrot.org/parrot/changeset/41954 shows Y/1 - "one test failure". but your commit message disagrees. 16:08
whiteknight There had been an intermittent failure in t/op/io.t on this sytem, which i didn't see when I just ran the test
Coke so the /1 is correct? ok. 16:09
whiteknight Let me run it again.
I did see that failure recently, but didn't see it when I tested this morning 16:10
Coke k. the /1 is confusing there. you might want to replace it with a *NOTE instead of a /FAILING_NUM 16:14
(pointing to the TT) 16:15
dalek kudo: 0a9860f | masak++ | docs/ROADMAP:
[docs/ROADMAP] updated wrt nqp-rx
 16:17
shorten dalek's url is at xrl.us/bfs9z3
16:23 integral joined
dalek rrot: r41957 | coke++ | trunk/PLATFORMS:
update platform
 16:27
16:29 kj joined 16:34 darbelo joined
darbelo Piper needs to die in a fire. 16:35
cotto_work It's like nobody here has heard of the ignore list. ;) 16:36
dalek rrot: r41958 | rblasch++ | trunk/PLATFORMS:
Updated VC9 platforms.
 16:40
Coke piper? 16:41
purl well, piper is the logbot for #parrot
16:42 Coke left, Coke joined
Coke I haven't seen piper in ages. =-) 16:42
nopaste "he" at 158.38.152.119 pasted "Enable shared libs for NetBSD, as for linux, freebsd, dragonfly, openbsd etc." (32 lines) at nopaste.snit.ch/18381 16:45
darbelo That was disabled on NetBSD? 16:47
he As near as I can tell from "svn log" it was never really enabled.
cotto_work I wonder how long it'll be before someone makes money from something running on a current version of Parrot. 16:48
mikehh I think 'he' was re-enabled yesterday 16:49
darbelo cotto_work: I did that alreday ;)
Coke dan did that back in... '03 ?
cotto_work *current version*
Coke it was current then! =-) 16:50
dalek kudo: 7eed769 | jonathan++ | (6 files):
Make Rakudo_binding_bind_signature actually take a low level signature object rather than an already unpacked one - this will make it far neater to recurse into the binder later. Also, cache the nameds to positionals hash. Doesn't actually win much performance, though we're building probably several less PMCs per dispatch, so less memory overhead for sure. Also remove a leak.
 16:51
shorten dalek's url is at xrl.us/bfs966
16:52 desertm4x joined 16:55 mikehh joined, payload joined 16:58 bacek joined
Coke is anyone building their languages dynamic ops from their top level src dir? 17:01
whiteknight no 17:02
dalek kudo: 55fd86e | jonathan++ | build/Makefile.in:
jnthn-- puts tabs in the makefile again.
shorten dalek's url is at xrl.us/bfs98s
jonathan Coke: If you mean, from one Makefile in their top level dir, yes 17:03
Coke: e.g. we do cd $(OPS_DIR) && $(OPS2C) C --dynamic $(OPS_SOURCE)
Parrot's default seems to be to scatter lots of little makefiles all over the place. 17:04
Coke jonathan: then you're not doing it from the top level. you're cd'ing into a subdir... =-)
jonathan The makefile is in the top level. :-P
Coke jonathan: (little) yah, trying to avoid that with partcl now.
jonathan But yes. :-)
mikehh gotta do another re-boot - bbiab
Coke merging the ops makefile into the root makefile first.
jonathan Note that we're compiling the ops there and then, not cd'ing and calling another makefile. :-) 17:05
darbelo Coke: I recently convinced PMC2C to work from the top-level in decnum-dynpmcs, not the same thing, but the approach could help here. 17:06
Coke jonathan: ah, I did misread that.
yay. was missing a -I . 17:07
17:09 mikehh joined
dukeleto is back and caffeinated 17:10
mikehh piper is bad enough but tweety 17:11
Coke wonders about RT #69887 17:12
I see no porn.
jonathan Well, you can always try googling. 17:13
;-)
dukeleto i have no bugfixes so far as part of 1.7.0, NEWS, but I don't actually believe that. we must have fixed a bug in trunk in the last month.... 17:14
Coke also rant: 20 line sig files with ascii art.
darbelo dukeleto: we fixed lots of bugs, almost as many as we introduced ;) 17:15
dukeleto mikehh: i think trac.parrot.org/parrot/ticket/1124 is closeable
kthakore where is teh p0rn?
hi jonathan dukeleto
dukeleto kthakore: hello 17:16
darbelo: name a bug we fixed this month that wasn't introduced since 1.6.0 (seriously) 17:17
darbelo line numbering in IMCC.
Biggest one I can think of.
dukeleto darbelo: got that already, in the compilers section
darbelo: perhaps it should be in the bugfix category
Coke That was only a partial fix. meh. 17:18
dukeleto is there a way to see closed tickets since 1.6.0 in trac?
or most recently closed tix.... 17:19
Coke probably. moment.
darbelo trac.parrot.org/parrot/query?statu...;report=11
shorten darbelo's url is at xrl.us/bftabi
dukeleto anybody remember which PMCs we improved coverage for this month?
i have CallSig
moritz various arrays, perhaps? 17:20
dukeleto darbelo++
moritz: oh yeah
Coke dukeleto: I see no way to restrict ticket query by "date of last activity on ticket"
moritz irclog.perlgeek.de/search.pl?channe...q=coverage
shorten moritz's url is at xrl.us/bftabn
Coke you could probably write a custom query for this. 17:21
whiteknight darbelo++ 17:22
Coke sorry, custom report.
darbelo whiteknight: huh?
whiteknight you're a super hero
I didn't realize you added Matrixy metadata to plumage 17:23
darbelo Okay.
17:23 gaz joined
darbelo whiteknight: doesn't work yet. It depends on parrot-linear-algebra. 17:24
whiteknight right, but that's still awesome
darbelo I never disputed that :)
japhb Now that dependencies are working in Plumage, I'm working on importing proto's projects, and then we'll have several dozen. 17:25
dukeleto japhb: exciting! 17:26
purl i heard exciting was good enough.
whiteknight I really need to learn more about plumage. Is there any good documentation about it?
darbelo the metadata/ dir is going to get crowded.
whiteknight: the code ;)
nopaste "he" at 158.38.152.119 pasted "atan part 2 now succeeds on netbsd" (23 lines) at nopaste.snit.ch/18382 17:27
he However, I'm getting a couple of other strange test failures, one of them related to ... recently enabled shared libs.
dukeleto darbelo: you added tests for Exception. that is going in NEWS...
cotto_work Hasn't that been failing forever? It's good to see it working.
dukeleto moritz: that search is very helpful, thanks!
whiteknight yeah! I didn't think I would see a fix in my lifetime 17:28
he++
he Yep, I think the change I did a while back to improve the math library initialization licked it.
Fix is config/gen/platform/netbsd/math.c r40837 17:29
cotto_work #ps in 60 17:30
darbelo dukeleto: Are you sure *I* added that. I can't recall that commit.
he I'm going to do a smoke test so you others can look at the test failures.
whiteknight he: awesome. Let us know when it's posted 17:31
he One says "Invalid shared object handle 0xbb8b7000" in the test output (uh-oh), the other appears to core-dump in a couple of unexpected places.
whiteknight urg
darbelo Ouch.
whiteknight core dumps are evil
darbelo But debuggable!
whiteknight wants to invent a computer with no core, so there is nothing to dump 17:32
darbelo wants backtrace.
dukeleto darbelo: yeah. that was me. /facepalm
whiteknight wants some icecream
darbelo whiteknight: Computers haven't used cores in qite a while.
whiteknight so then what does parrot keep dumping? 17:33
CORE
that's what
purl That's *WRONG!*
whiteknight you're wrong
Coke hurm. pmc2c seems to really like to be run in the pmc dir.
darbelo Coke: Yes, bu you can trick it if you are clever about it.
whiteknight or we could just fix it, if we were super-clever 17:34
darbelo I don't want it fixed, I want it replaced.
japhb whiteknight, re: Plumage docs: Right now there are pretty good docs for a couple of the internal modules, but the main program is rather underdocumented. I'm mostly planning to fix this by continuing to refactor code out of the main program into well-documented libraries. :-)
whiteknight ah, nice 17:35
japhb The metadata project in general is documented on the Parrot wiki.
whiteknight I would like to lend my moral support, even if I can't contribute directly
japhb "ModuleEcosystem", IIRC.
darbelo goes afk
japhb yup: trac.parrot.org/parrot/wiki/ModuleEcosystem
whiteknight, moral support is appreciated, thank you. :-) 17:36
Coke jonathan: does "cd FOO && " work on windows?
17:36 cconstantine_ joined
jonathan Coke: Aye. 17:38
dukeleto is running a fulltest in prep of committing what will be 1.7.0
jonathan Coke: What I copy/pasted you is straight from the Rakudo makefile, which builds fine on Windows. :-)
dukeleto no commits to trunk in the next hour, please, unless something is seriously on fire 17:39
whiteknight dukeleto++ 17:40
17:40 theory joined
dukeleto is also hacking on some release automation scripts that I will commit after the release 17:41
cotto_work dukeleto++
17:41 theory joined
he smolder.plusthree.com/app/public_pr...ails/29228 17:44
shorten he's url is at xrl.us/bftaeq
cotto_work I thought those extend tests got fixed 17:45
Tene japhb: how would you like LOLCODE installable from plumage? 17:49
darbelo Tene: +1! 17:50
japhb Tene, I would, yes, thank you!
Tene I just need to track down where it went, or pull it out of svn history.
japhb I'll consider it a nice milestone when all living Parrot HLLs can be installed via Plumage
Tene, totally worth the effort. ;-)
darbelo Hmm. I'll have to get back on the 'install partcl' thing. 17:51
Tene japhb: it was the first language supporting HLL interop.
japhb Tene, good choice. :-)
whiteknight darbelo: Matrixy doesn't currently depend on Parrot-Linear-Algebra. It can work without it
darbelo whiteknight: I'm planning to change that ;) 17:52
whiteknight darbelo: yeah, I'm planing to change it too :)
but for now, it works alone
jonathan I just discovered that Parrot_call_sub doesn't handle multis.
Tene Oh, it's in svn.parrot.org/languages/
jonathan Not sure if that's by design or an unhappy accident.
I can work around this easily of course (write an only that re-dispatches to the multi - the multi I want to dispatch to also tail calls, so I get to avoid that bug too). 17:53
japhb Tene, commit us some metadata!
jonathan But since it's the documented way to call a sub from C...I just wondered if it should be able to handle multis too. 17:54
I cna file a Trac ticket if so.
Tene japhb: I need to make it work against an installe dparrot, give it an install make target, etc.
I bet.
japhb nodnod
dukeleto please test the shit out of r41959
dalek rrot: r41959 | dukeleto++ | trunk (11 files):
[release] Parrot 1.7.0
 17:55
darbelo Tene: I can probably help there. I just did that for Matrixy.
Tene darbelo: any chance you can give me a link to a matrixy commit?
dukeleto is doing 'make release' now and checking that
cotto_work dukeleto fulltest pases on x64/ubuntu 17:56
darbelo Tene: github.com/Whiteknight/matrixy/comm...21792c7671 17:57
shorten darbelo's url is at xrl.us/bftaij
darbelo There were a few more commits after that, but that's the big one. 17:58
Tene darbelo: thanks.
dukeleto cotto_work++ 17:59
he Hmm, seems that that parrot doesn't properly deal with the NetBSD ELF shared libraries.
mikehh #ps in 30 18:00
he On NetBSD, the real-time linker really likes to find the shared library where it was told it is to be found via -Wl,-R
s/real-time linker/ld.elf_so/
dukeleto is testing the release tarball now 18:01
whiteknight "dukeleto: Rock Star"
cotto_work dukeleto if you've tagged the release the rest of us can resume breaking stuff
whiteknight PREPARE TO SET ENGINES TO MAXIMUM DESTRUCTION!!! 18:02
Tene cotto_work: only if he's comfortable that the release is good. He doesn't want to have to deal with reverting things if he needs to make additional commits to fix up the release.
cotto_work I'll leave it to him. 18:03
whiteknight once he is done testing the tarball, that should be it
dukeleto cotto_work: i need to do the tag. very soon now. i will wait a few minutes for a few more "tests pass" from some peeps 18:04
18:04 chromatic joined
dukeleto also, I will test parrot on some other machines I have access to 18:05
japhb We've had at least one time in the pass when people broke things too soon -- IIRC Rakudo couldn't find a master-branch commit that was kosher. Better to give everyone some breathing room.
er, in the past
darbelo clock? 18:06
purl darbelo: LAX: Tue 11:06am PDT / CHI: Tue 1:06pm CDT / NYC: Tue 2:06pm EDT / LON: Tue 7:06pm BST / BER: Tue 8:06pm CEST / IND: Tue 11:36pm IST / TOK: Wed 3:06am JST / SYD: Wed 5:06am EST /
whiteknight #ps in 23 18:07
dukeleto can someone test the release rev on windoze? 18:08
whiteknight on it
what rev? 18:09
purl i think rev is nice too
he Test on NetBSD/macppc 5.0: smolder.plusthree.com/app/public_pr...ails/29231
shorten he's url is at xrl.us/bftaku
japhb purl, forget rev
purl japhb: I forgot rev
he Demonstrates the "could not find shared library" problem (I've not done "make install" before the test)
dukeleto whiteknight: 41959 18:10
whiteknight gotcha. On it
dukeleto dammit. parrot -V still says 1.6.0-devel . why? 18:12
cotto_work dukeleto ./parrot?
dukeleto the release guide says to make sure that it has the right number, but not what to change if it doesn't 18:13
whiteknight chromatic: you get my message yesterday?
chromatic Did you make a release tarball?
dukeleto cotto_work: (kadath)(~/svn/parrot-1.7.0-release/parrot-1.7.0)$ ./parrot -V
This is Parrot version 1.6.0-devel built for nojit.
chromatic: yes, I did make release=1.7.0
chromatic I did, whiteknight. I'm not sure how that change would look, but it's an idea.
whiteknight you updated all the metadata files
? 18:14
he oh, t/src/atomic.t specifies C code to compile and run. I guess it doesn't get passed the proper -Wl,-R flag, but where should that happen?
chromatic There's no single separate fastcore; we'd have to modify all invocations of functions in the cores that share those functions.
That's doable.
whiteknight chromatic: yeah, that's what I'm thinking.
so all fastcore op functions take three arguments (opcode_t*, interp, ctx**) instead of two 18:15
dukeleto needs halp with parrot -V
whiteknight I've never been exactly sure where that value comes from. Now is a good time to learn
dukeleto: make sure you don't have a previously-installed libparrot anywhere? 18:16
18:16 nbrown joined
dukeleto whiteknight: shall i delete my installed parrot? 18:16
whiteknight yes 18:17
dukeleto whiteknight: it scares me that an installed parrot should effect that version number
whiteknight: scary but I will
whiteknight goddamnit. Failed t/op/annotate.t
(but looks like a harmless, though obnoxious, error) 18:18
have: 0, want: -0
dukeleto whiteknight: yep, it was the installed parrot
whiteknight okay, awesome
dukeleto whiteknight: !!! like i said a few days ago. testing numbers exactly as strings is ALWAYS A BAD IDEA
dukeleto looks at annotate test 18:19
whiteknight hold on, need more testing
cotto_work dukeleto it could be a bytecode issue.
dukeleto whiteknight: which test was that?
whiteknight don't know, I already flushed my console buffer
cotto_work nm. There's no PBC_COMPAT bump for non-supported releases.
mikehh smoke #29233 PASSes at r41959 - 11,103 ok, 0 failed, 262 todo, 573 skipped and 0 unexpectedly succeeded - Ubuntu 9.10 (beta updated) aqmd64 18:20
whiteknight allison++
dukeleto mikehh++
whiteknight ++ for everybody!
dukeleto whiteknight: so, do we still have a failing test on windoze? 18:21
chromatic Someone should delete the pcc_optimize_sig test.
whiteknight dukeleto: still running.
chromatic s/test/branch/
whiteknight we didn't have any test failures at r41949, so I'm hoping failures didn't magically appear in 10 revisions after that 18:22
dukeleto whiteknight: i can change those tests over to use 'fp_equality.pasm' if needed. let me know as soon as you can
whiteknight ok
dukeleto whiteknight: if windows passes, i will tag it and bag it 18:24
dalek rrot: r41960 | darbelo++ | branches/pcc_optimize_sig:
Remove pcc_optimize_sig, this branch has alredy been merged and is no longer needed.
whiteknight failed
t/op/arithmetics.t 21 and 24
Tene argh darbelo beat me
whiteknight t/op/io.t, failed test 4
dukeleto whiteknight: fuck
whiteknight t/pmc/env.t didn't run 14 tests
dukeleto whiteknight: can you nopaste the whole log? 18:25
also, can we triage these failures? I can work on some, but I can't test them
Coke whiteknight: what platform ar eyou on?
whiteknight win32
Coke the -0 sounds like a familiar (and old) bug.
whiteknight r41959, right? 18:26
dukeleto whiteknight: yes
chromatic Thanks, darbelo.
darbelo No problem.
dukeleto whiteknight: so annotate isn't failing now, but 3 other tests?
whiteknight apparently. Let me try again, from the top
dukeleto whiteknight: realclean and perhaps rm your installed parrot, just for funzies 18:27
can anyone else verify failing tests on win32 in trunk?
like, have these been failing for a while?
'cause that would suck
darbelo Do we have any recent smolders to confirm? 18:28
he Build emits "trap 1253 experimental, not in ops.num" during ops2pm.pl run. 18:29
whiteknight that's expected
he ok 18:30
chromatic #ps time 18:31
dukeleto smolder.plusthree.com/app/public_pr...ails/29207
shorten dukeleto's url is at xrl.us/bftarh
he I'm testing a fix for the "need to point run-path to shared libs in blib/lib/ on NetBSD"
Essentially, if "rpath" is defined, use it in libparrot_linkflags. 18:32
dukeleto so, these tests have been failing for a while on win32. this sux.
the questions now is, what does a Duke do? 18:33
mikehh All tests PASS (pre/post-config, smoke (#29233), fulltest) at r41959 - Ubuntu 9.10 (beta updated) amd64
18:33 kj joined
dukeleto conditionally skip them on win32, i guess. can someone help me with this? i can't test them 'cause i got no access to win32 18:34
dukeleto begs for halp
whiteknight working on it
japhb dukeleto, I would help you if I could. I don't have win32 either. :-(
whiteknight chillax
dukeleto whiteknight++ 18:35
whiteknight urg, that time it didn't even build 18:36
cannot open output file libparrot.dll: Permission denied
Coke etc?
purl etc are why php sucks
Coke etc is also partcl!
moritz etc? 18:38
purl somebody said etc was partcl!
dukeleto good thing I automated some parts of the release process, since I will have to do it again once win32 is happy. yay for forward-thinking
japhb dukeleto, plumage thought: when I finally get the metadata up on a server, we could move the existing metadata/ dir directly to t/metadata/, and have it be the test data collection. Just a random thught. 18:40
darbelo make fulltest PASS on OpenBSD i386 18:43
18:43 bacek joined 18:44 nopaste joined
dukeleto darbelo++ 18:45
darbelo This is the first release I've been able to say that. 18:46
whiteknight I completely nuked my repo and checked out a new one 18:47
Coke darbelo++ gets a cookie. pcc branch failing several partcl tests. 18:48
this means that parrot is missing tests.
darbelo Told ya. 18:49
nopaste "coke" at 72.228.52.192 pasted "partcl on pcc_reapply" (21 lines) at nopaste.snit.ch/18383
whiteknight Parrot is definitely missing tests
chromatic Tests welcome from languages.
Coke hurm. I appear to have access to create a mailing list via the web.
darbelo And we need to run them more often. Look at the win32 stuff happening right now. 18:50
Coke darbelo: we have failing smolder tests that no one looks at. =-)
whiteknight this month was hard because so much focus pointed away from trunk
Coke having more people test things so we can ignore them probably won't help.
whiteknight I don't ignore them
darbelo Coke: true.
Coke (so we need to fix something else. Like making smolder not being green a big deal.)
whiteknight I just don't immediately fix them all eiher
Coke whiteknight: how is that different? =-) 18:51
(except for in the future)
whiteknight it's a matter of intent
don't ignore, "prioritize"
Coke ok. if you'd like to word-kibbitz, then they are priortized too low considering the agita they are now causing at release time. 18:52
whiteknight that's better
darbelo If we're fixing them on release day, our prioritues are hosed, IMHO.
Tene Didn't we have someone on the ml a little while ago offering to set up smolder reporting on a wide range of systems?
dukeleto whiteknight: any status on win32 on trunk?
whiteknight re-retesting now 18:53
got a fresh checkout
dukeleto whiteknight++ # go go gadget!
Coke SMOLDER FAILURES:
smolder.plusthree.com/app/public_pr...port/29232
shorten Coke's url is at xrl.us/bftavj
Coke smolder.plusthree.com/app/public_pr...port/29231
shorten Coke's url is at xrl.us/bftavm
Coke smolder.plusthree.com/app/public_pr...port/29230
shorten Coke's url is at xrl.us/bftavo
Coke smolder.plusthree.com/app/public_pr...port/29229
shorten Coke's url is at xrl.us/bftavq
dukeleto whoa nelly
Coke (current 4 failurs from the RSS feed.)
whiteknight okay, same list of failures
Coke If we have a failure on a platform reported multiple times, we should probably TODO it. (and open a ticket.) 18:54
whiteknight t/op/arithmetics.t stuff is all the weird -0 shit
dukeleto ok, netbsd it not officially supported yet, so that is ok 18:56
he What? :)
(The latest failures have to do with me enabling shared library support, and testing it)
dukeleto one failing test on linux/amd64: smolder.plusthree.com/app/public_pr...st_failure 18:57
shorten dukeleto's url is at xrl.us/bftawi
whiteknight okay, somewhere Parrot is silently exiting in t/pmc/env.t
dukeleto that looks fishy. others would have surely noticed, right? or is no one else on that exact linux platform?
whiteknight it doesn't like the line "delete $P0['PARROT_TMP']" 18:58
he dukeleto: that one is pretty similar to what I see on NetBSD.
dukeleto he: all your work is appreciated, but I am trying to get a release out the door and passing tests on our officially supported platforms
he: interesting
he dukeleto: I understand; I was joking. 18:59
dukeleto whiteknight: is $P0 null?
he: no worries :) Slightly tense since I have some plans for this afternoon and i don't know what is going on right now. tests are all of a sudden failing on win32 and linux
but not on my boxen
trunk passes 'make test' on FreeBSD 6.2 for me 19:00
whiteknight dukeleto: no worries. If you need to go we can take over
19:01 nopaste joined
he Re-test after dealing with rpath issues in C source tests on NetBSD/macppc 5.0: smolder.plusthree.com/app/public_pr...ails/29236 19:01
shorten he's url is at xrl.us/bftax8
dukeleto whiteknight: i have another 1/1.5 hours, I just am totally useless right now, because the tests aren't failing on any of my boxen
whiteknight gotcha.
brb
he The only failing test now is t/pmc/threads.t
Coke he: that was failing for me on amd64, but nowhere else. 19:02
dukeleto is going out to lunch
Coke (I just marked it as /1 in the PLATFORMS and ignored it)
dukeleto please have everything fixed when I get back
he I tried to nopaste my "use rpath if it's set", it's at nopaste.snit.ch/18385 19:04
(For some reason the nopaste didn't end up here.)
19:12 Rhy0lite joined
moderator Parrot 1.7.0 "African Grey" is coming soon, no non-release-related commits to trunk, please | Improve PCC testing this week or the release gets it!
cotto_work Is plumage's long-term plan to live outside Parrot's repo long-term and only be bundled with Parrot releases? 19:12
s/repo long-term/repo/ 19:13
japhb cotto_work, previous to today it was to eventually merge back. But it sounds like the general feeling is to keep it outside and snapshot it.
Coke chromatic: so. I have partcl failures against pcc_reapply. Would you say that's a show stopper to merging back to trunk or no?
chromatic Depends on the failure. 19:14
japhb I don't have a particularly strong feeling about this, other than the general Perl 5-ish fear of long-term dual-life modules. But as I understand it, this is supposedly largely mitigated by Parrot's aggressive release cycles.
darbelo japhb: It looks to me like parrot is going the small core, lots of modules way. 19:15
japhb darbelo, yes indeed. But I was given to understand (again, before today), that Plumage and whatever it needed where going to be the outer shell of that small core. 19:16
darbelo Opposite from the perl 5 "pull it into the core" approach.
Rhy0lite Hi. Is there any information about the benefits that Parrot achieves or will achieve by using Continuation Passing Style? I have read Squawks of the Parrot blog and other information, but I am trying to understand the advantage, other than simplifying support for languages that provide continuations as first-class object.
Coke ah, this one's probably easy: tcl_group.so: undefined symbol: Parrot_run_meth_fromc_args_reti
chromatic Coke, you should be able to rewrite that one into a Parrot_pcc_invoke_from_c_args variant. 19:17
Coke but I didn't write that code. =-)
it's coming from the pmc2c code. 19:18
chromatic Now you have my attention.
Coke bah.
cotto_work Coke: from current pmc2c code?
Coke I might have been lying. moment.
yes. lying. it's in the .pmc 19:19
(*&#$
chromatic You need to submit a Smolder report for your wardrobe because your pants are on FIRE.
darbelo chromatic++
Coke (hey trac, can you go slower? thanks!) 19:20
cotto_work Coke, don't taunt it.
Coke chromatic: it would be helpful if that upgrade info was here: trac.parrot.org/parrot/wiki/Deprecation
(and will it work against trunk also?) 19:21
darbelo Coke: it should. 19:22
Coke ok. pointers on that wiki page will be helpful to more than just me, presumably.
19:26 mokurai joined 19:27 joeri joined
Coke chromatic: Parrot_pcc_invoke_from_c_args is not in the pcc_reapply branch. 19:28
chromatic It should be instead Parrot_pcc_invoke_sub_from_c_args(). 19:29
Coke got it. danke. 19:30
dalek p-rx: b179ffb | pmichaud++ | src/ (4 files):
More quote_EXPR and quoting improvements.
 19:32
shorten dalek's url is at xrl.us/bfta45
Coke can I just use Parrot_PCCINVOKE ? 19:33
(this is for a method, not a sub, btw, so using _method_)
whiteknight do not use Parrot_PCCINVOKE
Coke also, I am using PMC * const method = Parrot_oo_find_vtable_override_for_class(interp, _class, get_bool); 19:35
... but Parrot_pcc_invoke_method_from_c_args seems to just take a STRING for the method name.
how can I reconcile?
whiteknight Coke: you can't make a STRING?
Coke I could. does Parrot_pcc_invoke_method_from_c_args respect vtable overrides ?
whiteknight or, let me rephrase: what do you need exactly? 19:36
Coke whiteknight: to invoke a vtable override if it exists.
whiteknight how do you do it currently?
Coke github.com/partcl/partcl/blob/maste...ng.pmc#L38
shorten Coke's url is at xrl.us/bfta5j
Coke using the two methods I just mentioned here. 19:37
dalek rrot-plumage: dbd4325 | darbelo++ | :
Add metadata for parrot-linear-algebra.
shorten dalek's url is at xrl.us/bfta5m
Coke ( and that fails against pcc_reapply)
japhb darbelo, Is it wotking now? 19:38
chromatic Given a Sub and an invocant, you want to invoke that Sub as a method with that invocant and pass parameters and receive return values.
Coke and apparently Parrot_pcc_invoke_method_from_c_args is not the way to do that. yes? 19:39
19:39 nopaste joined
chromatic There's a different function somewhere I'm sure I saw that is the right thing for you, but I haven't found it yet. 19:39
Coke k. 19:40
darbelo japhb: not yet. But that's a parrot-linear-algebra issue.
chromatic Maybe you have to build your own CallSignature. Gah.
Coke Parrot_pcc_invoke_from_sig_object ? 19:41
whiteknight Parrot_pcc_invoke_sub_from_c_args
Coke whiteknight: it's not a sub.
unless I'm misreading the difference between the _sub_ and _method_ variants there.
whiteknight yes, Parrot_pcc_invoke_from_sig_object 19:42
Coke chromatic: that would suck a bit.
whiteknight and Parrot_pcc_build_sig_object_from_c_rgs
Parrot_pcc_build_sig_object_from_c_args
Coke so I do need to be build my own CS?
whiteknight yes
darbelo That sucks.
purl The rock is now off.
chromatic Parrot_call_method_ret_int() 19:43
in src/extend.c
cotto_work whiteknight: s/c_args/varargs/ ?
whiteknight maybe, I'm looking at it now 19:44
Coke chromatic: Parrot_String method
whiteknight Coke: PMC * call_sig = Parrot_pcc_build_sig_object_from_varargs(INTERP, SELF, "->I", &result);
Coke method is still a method name there.
(not a method obj.)
whiteknight no, sorry: PMC * call_sig = Parrot_pcc_build_sig_object_from_varargs(INTERP, SELF, "Pi->I", &result) I think 19:45
chromatic Coke, read the code.
method is *unused* in favor of the Sub passed in. 19:46
Coke chromatic: ... *boggle*
chromatic Yes, the documentation is Bad.
cotto_work Has anyone run make cover on pcc_reapply? 19:47
That'd be a good idea before we dive into writing tests.
allison Coke: a method can be invoked with the *sub* variants
Coke: the only unique thing is when you need to look up the method by name 19:48
Coke: (when you don't already have a code object for the method)
Coke that would be a nice note for the docs. 19:50
19:50 ruoso joined
whiteknight we do need lots of examples of this stuff 19:50
by the way, do we have any other win32 users here today?
we could really use some more hands on this
cotto_work I'm on win7 but I've been seeing a number of failing tests anyway. 19:51
Coke return (INTVAL)Parrot_call_method_ret_int(interp, method, SELF, get_bool, "->I"); 19:52
?
chromatic Looks reasonable. 19:53
Coke k. something is still barfing. digging.
cotto_work I'll see if I can get something useful. 19:54
Coke segfaults : nopaste.snit.ch/18386 19:55
(and our nopaste bot is gone.)
19:55 nopaste joined
darbelo <value optimized out>? 19:56
Coke: can you build unoptimized?
chromatic Hm, the Pi>I looks wrong. Where's the ->?
Coke darbelo: I could, but no! =-) 19:58
chromatic Coke, can you remove line 1316 from src/extend.c and rebuild Parrot?
Hm, on second thought. 19:59
Remove the +1 from line 1317. 20:00
whiteknight win32 is infuriating 20:01
I love how parrot just exits silently for no apparent reason 20:02
like, I assume it's a segfault
nopaste "cotto_work" at 131.107.0.77 pasted "test results on win7 with strawberry toolchain" (38 lines) at nopaste.snit.ch/18387
whiteknight but it could say something
Cotto: I'm seeing the same env.t failures here
cotto_work and io?
purl i guess io is "input/output". or "indirect object" or a core module or one of Jupiter's moons or a small, prototype-based programming language at www.iolanguage.com/about/
whiteknight so they're probably related and if you can figure them out it would help everybody
io.t is evil on windows, fails intermittently, and is mostly a problem with the test that nobody has fixed yet 20:03
on windows, the fourth test just doesn't run instead of being TODOs
nopaste "coke" at 72.228.52.192 pasted "better" (29 lines) at nopaste.snit.ch/18388 20:04
Coke chromatic: so now we have Pi->I
... and then it still segfaults in clone_key_arg. =-)
20:05 nbrown joined
whiteknight are you doing this on pcc_reapply? 20:07
Coke yes.
whiteknight I thought clone_key_arg was deleted in that branch
or, at least no longer needed 20:08
Coke whiteknight: apparently not?
cotto_work whiteknight I'll take a shot when I can (a little later today), but my win32-fu isn't strong.
Coke (not deleted, anyway)
chromatic: looks like that method in extend.c also has a c-string off by one error (2, should be 3.)
whiteknight cotto_work: I may have just solved it
chromatic 2 is fine, Coke. 20:09
nopaste "coke" at 72.228.52.192 pasted "pcc_reapply patch." (20 lines) at nopaste.snit.ch/18389
chromatic The strcat confused me too.
cotto_work whiteknight, schweet!
whiteknight how do I make a string constant in platforms.c? CONST_STRING and CONST_STRING_GEN don't work there
Coke chromatic: oh. because we only need one trailing \\0 ?
chromatic Exactly. 20:10
Parrot_str_from_literal(), whiteknight?
Coke chromatic: ewwww. =-)
whiteknight chromatic: okay, I just tried Parrot_str_new and it's compiling
chromatic I can't say I'm a fan of this code in src/extend.c.
20:11 kurahaupo joined
cotto_work whiteknight, patch? 20:11
dukeleto is back
kurahaupo good $localtime all.
whiteknight cotto_work: no patch, just a quick commit 20:12
20:12 Rhy0lite left
whiteknight r41961 20:13
dalek rrot: r41961 | whiteknight++ | trunk/config/gen/platform/win32/env.c:
[win32] fix a weird error in the environment variable code for win32 where we were passing a char* instead of a STRING* like we should ahve been
 20:14
whiteknight has to go home now. Will be back online in an hour to sort out the rest of this nonsense
we'll get this all fixed tonight
dalek kudo: af3872d | moritz++ | docs/announce/2009-10:
[docs] initial release annoce for 2009-10
 20:15
kudo: 7e34de6 | moritz++ | docs/ChangeLog:
mention new signature binder in ChangeLog
shorten dalek's url is at xrl.us/bftbar
dalek's url is at xrl.us/bftbat
whiteknight I don't know what we're going to do about arithmetics.t. We might just have to ignore that for now
later 20:16
20:18 particle1 joined
dalek kudo: 95d2f43 | moritz++ | lib/Safe.pm:
disallow IO::Socket in Safe.pm
 20:20
rrot: r41962 | chromatic++ | branches/pcc_reapply/src/extend.c:
[extend] Fixed fencepost errors when prepending invocant signature to

want to delight HLL developers.
Coke chromatic: the vtable of the key that is getting cloned is at 0x1
shorten dalek's url is at xrl.us/bftbbu
chromatic It's an invalid key then.
Coke, delete line 1326 from src/extend.c and try again. I have an idea, but it's just an idea. 20:22
20:23 hercynium joined 20:24 nopaste joined
Coke nope. 20:26
(still segfaults) 20:27
chromatic Alright, time to make a test case we can put in t/src/extend.t. 20:28
cotto_work The env.t failure is fixed on win32. 20:29
Coke chromatic: feel free to use code from tclstring.pmc for that. 20:30
(that code also requires something like:)
github.com/partcl/partcl/blob/maste...g.pir#L199 20:31
shorten Coke's url is at xrl.us/bftbdm
Coke (the class override to the PMC)
chromatic If the problem is where I think it is, I don't think the code needs to get that detailed. 20:32
I think it may be that all of these wrapper functions have a subtle bug.
dalek kudo: 26bf858 | (Kyle Hasselbacher)++ | docs/announce/2009-10:
[announce/2009-10] typo fixes
 20:37
shorten dalek's url is at xrl.us/bftbgi
Coke seen kid51? 20:41
purl kid51 was last seen on purl 21 hours, 29 minutes and 31 seconds ago, saying: <private message>
Coke ack -a Parrot_pcc t => <> 20:48
(there is some Parrot_call, though, which is more appropriate.)
dalek kudo: ec7e91e | moritz++ | t/spectest.data:
add S05-grammar/methods.t to spectest.data
 20:49
shorten dalek's url is at xrl.us/bftbhq
dukeleto is writing the release announcement for parrot.org 20:50
darbelo dukeleto++
Coke dukeleto: did whiteknight's windows problem get addressed? 20:51
dukeleto Coke: Util is working on it now
Util++
darbelo He fixed some of them, I don0t know if he got them all.
cotto_work Coke, env.t passes and io.t fails
dalek trixy: 21f2eff | darbelo++ | Configure.pl:
Make Configure.pl report the actual command used to 'make' Matrixy.
 20:52
shorten dalek's url is at xrl.us/bftbh4
dalek trixy: c3ffb60 | darbelo++ | :
Merge branch 'master' of github.com:Whiteknight/matrixy
shorten dalek's url is at xrl.us/bftbh6
dalek trixy: 1966941 | darbelo++ | config/makefiles/root.in:
Remov some stray ':=' uses from the makefile template.
shorten dalek's url is at xrl.us/bftbh8
21:01 bacek joined
cotto_work hi bacek 21:02
bacek G'Day cotto_work
21:04 he_ joined
dukeleto release announcement review (not published to main page): www.parrot.org/news/parrot-1.7.0-a...y-released 21:06
shorten dukeleto's url is at xrl.us/bftbkk
kj good evening 21:07
purl Ah, evening. The tumultuous mind tarries and contemplates, reveling in the silence afforded by the diurnal proletariat. Good evening, indeed.
dukeleto kj: hola
purl hola, dukeleto.
kj I see that PARROT_MAX_ARGS has been increased to 16
chromatic 11 wasn't loud enough. 21:08
kj I'm not 100% sure, but I think for each Parrot op there is a struct instance
(eh, sorry, I was raised object-oriented)
and each of those instances has a number of arrays
dukeleto Util++ Util++ Util++
cotto_work kj, that sounds right.
kj sized MAX_ARGS
dukeleto kj: yes, it has been increased to 16
cotto_work dukeleto did he fix it yet?
kj which means, that if there's 1200 ops
and 3 arrays i believe 21:09
eh, so 3 * 1200 * 8 * 4bytes
dukeleto cotto_work: he is very close to committing stuff that properly TODOs stuff for win32
cotto_work good enough
darbelo dukeleto: no quote?
cotto_work we need to kick that release out the door and get back to business
dukeleto darbelo: I NEED A QUOTE? /me panics
kj eh, anyway, that's alot of bytes extra in memory...
just a thought. 21:10
dukeleto darbelo: there was no quote on www.parrot.org/news/parrot-1.6.0-ha...-released, but now I want a quote!
darbelo dukeleto: Don't panic. They aren't mandatory, just cool.
dukeleto darbelo: I loves me some quotes
darbelo Other thatn that, it looks good to me. 21:11
kj isn't it a concern that there's 115KB extra in memory just to allow 16 operands, which doesn't really happen a lot
as in, none so far I think 21:12
dukeleto darbelo: any thoughts on what kind of quote, what the "tone" of the quote should be? pure ridonkulousness?
jonathan kj: Yes, but the fact we allocate any fixed size seems odd, given that most of our ops are only 2 or 3 in size.
darbelo We had a relese named "TEH PARROTZ", that's a pretty low bar.
kj jonathan: I agree that it's "odd", but I can also see that it's the easiest way to do it 21:13
oh btw, they're chars, not ints, so divide by 4 there...
still, 30KB or so
ah well, I guess it's not much on a PC
cotto_work kj, that's nothing given all the stuff parrot does. ;) 21:14
21:14 bluescreen joined
kj I guess so yes. 21:14
dalek TT #1125 created by pmichaud++: [BUG] imcc handles backslashes improperly in sub names 21:15
pmichaud chromatic: if you could work your usual magic on TT #1125, I'd much appreciate it. :-)
darbelo dukeleto: also, there's www.parrot.org/category/news/releases to look at previous entries for quote tone. 21:16
jonathan pmichaud: You discover the best bugs. :-) 21:17
kj: Sure, there's probably bigger fish to fry for now, though. :-)
But I agree it could be optimized. 30KB isn't so bad though. 21:18
kj jonathan: yeah right. if i think about it now it's nothing. I just had to think about the big numbers
jonathan They're only big if you put bytes in front rather then kb :-) 21:19
kj it could be optimized for sure; 1 array only needs 1 bit information, and the other 2 array needs 2 bits, so it could all be encoded in 1 char
anyway, i'm prematurily optimzing here i think 21:21
cotto_work premature optimization?
purl it has been said that premature optimization is the root of all evil or when you make the mistake of optimising for speed when really you should optimise for maintainability and correctness, in the first instance. or unnecessary optimization is premature by definition. or hacking with tequila in preference to wine
pmichaud jonathan: iwbni I could discover the bugs without spending lots of times in debuggers and tracing first :-| 21:22
dalek rrot: r41963 | Util++ | trunk/t/op (2 files):
[t] TODOed failing Win32 tests for 1.6 release - pipe writes and negating -0.0
cotto_work t/op/arithmetics.t looks fine for me 21:23
jonathan pmichaud: I've managed to trip over the same (already ticketed) Parrot bug like 3 times in the sig binder re-write and lost a bit of time to it each time... :-| 21:24
Then felt silly for doing so.
(the "don't call something that tailcalls from C" one... 21:25
)
Util On Win32, both the Neg(-0.0) issue in arithmetics.t, and the pipe writes in io.t, only occur with the MSVC compiler; MinGW runs them as `TODO PASSED`. If there is a good way (ala .SYSINFO_PARROT_OS) to detect the compiler name, then we can refine the TODOs. 21:26
cotto_work Util, with minw gcc I get a pipe failure but an arithmetics.t pass 21:29
21:30 Whiteknight joined
dukeleto whoever can test r41963 on your box, I would highly appreciate it 21:30
darbelo: www.parrot.org/news/parrot-1.7.0-a...y-released , better?
shorten dukeleto's url is at xrl.us/bftbkk
cotto_work dukeleto, access DENIED 21:31
darbelo dukeleto++
SHIP IT! 21:32
cotto_work dukeleto, I'd shorten it a bit
(which is ironic if you look at the 1.4 announcement)
darbelo But... But... It has tensor algebra! 21:33
cotto_work now apparently the TODO'd tests pass with mingw gcc 21:34
Util cotto_work: Perhaps it is a versioning difference; I am using GCC 4.4.0-TDM-1, and pass both tests
Whiteknight I'm loading up my test machine right now
chromatic pmichaud, I probably can't get to it until tomorrow, but I'll work on it. 21:36
cotto_work tdm?
Apparently I'm on gcc 3.4.5. 21:37
pmichaud chromatic: no big worries, I figured out an easy workaround for now. 21:38
chromatic I *think* the fix is easy too, but I don't want to start before I have a solid block of time in which to fix it if things go kablooey. 21:39
kj Who is interested in PIRC documentation?
kj is drawing some attention
cotto_work kj, writing or reading?
kj cotto_work: reading
I'll do the writing
darbelo I am. 21:40
cotto_work I'll have an hour or so when I get home tonight. I'd be happy to do any proofreading/whatever then. 21:41
kj, should I just check commits from you when I get back? 21:42
kj well, the reason I'm asking is, I'd like to know what else is needed. There is a wiki page, and the source is documented, but I need input on "I don't understand how XXX works in PIRC"
darbelo I hate IMCC so much that I'm willing to *read documentation* in order to kill it.
kj eh, to finish my sentence: I need input on "how is X done", so I can write that. I need to understand what other people are missing, so I can fill that gap 21:43
Whiteknight Im interested in pirc docs
darbelo Not exactly what you asked, but: What features are missing? 21:44
kj Whiteknight: i started a page on the wiki (PIRCDevelopment) and there's the source documentation
Util cotto_work: www.tdragon.net/recentgcc/ - *much* more aggressive about getting the latest stable GCC ready for Win32 than the official MinGW team. Released 4.4.1 this month.
kj darbelo: the major thing that needs fixing is bytecode generation. related to that is lexicals working in generated bytecode; i looked/copied from IMCC, but i dont' think I got it right. 21:45
other stuff includes critically reviewing the macro layer.
but that's a syntax thing, really, and if need be, that could be isolated in a separate preprocessor to keep it clean 21:46
darbelo So, parsing (outside of macros) works but generates the wrong bytecode?
kj and theres' some rough edges, but most of them are not difficult to fix, just takes time
darbelo: not the wrong bytecode, just bytecode that crashes
Whiteknight a separate macro preprocessor might be the way to go, since the spec calls for .macro_local variables
kj Whiteknight: I already implemented those I think 21:47
darbelo Correctly crashing bytecode. Got it ;)
kj ...it's been a while. they worked at some point I think
Whiteknight oh, okay. nevermind then
darbelo are the generated .c files checked into the repo?
kj darbelo: simple stuff just using integer values, no problem. FLOATVALs and STRINGs: no good.
darbelo: are your referring to parser/lexer generated? Yes they are in the repos 21:48
Whiteknight: what's more difficult is nested macro expansions.
anyway, before listing all the bugs, the way to get people working on it, I hope, is to provide eye-candy-sweet documentation 21:49
darbelo The makefile must be wonky, then. I'm getting errors in an .l file. 21:50
kj errors??
purl i think errors is buggy? funny
kj warnings, yes
but errors no good
darbelo For some reason When I make pirc it tries to lex src/hdocprep.l 21:51
And that fails.
purl that fails are Nicholas's fault
darbelo purl forget And that fails.
purl darbelo, I didn't have anything matching and that fails
darbelo purl forget And that fails
purl darbelo, I didn't have anything matching and that fails
darbelo purl forget that fails 21:52
purl darbelo: I forgot that fails
kj darbelo: what version of flex do you use?
there's a version that has problems
darbelo 2.5.4
kj yes i think that's the one with prolbms
I seem to remember.... not 100% sure 21:53
darbelo It shouldn't try to run lex at all if the .c files are there.
kj what are the errors?
purl the errors are part of the game. Alcohol can help the setup. or nothing we talk about in fight club.
kj the Makefile could be not completely sane, but in general shouldn't really matter
darbelo "src/hdocprep.l", line 1: unrecognized '%' directive 21:54
"src/hdocprep.l", line 16: bad character: #
"src/hdocprep.l", line 16: unknown error processing section 1
etc
Coke kj: I am interested in pirc docs, but am more interested in getting a --pirc=pirc option for config. =-)
kj Coke: .. but PIRC is not really stable yet... 21:55
Coke kj: the best way to make it stable is to /run it/
purl Hmm. No matches for that, Coke. 21:56
Coke (and then fix it. =-)
kj well you can run it already. But I see what you mean. perl Configure.pl --pirc=pirc will link in PIRC instead of IMCC 21:57
Coke ooh, that's already in?
kj oh sorry no
dalek kudo: 0f6203f | jonathan++ | src/ (5 files):
Implement attributive parameters.
shorten dalek's url is at xrl.us/bftbtk
kj i mean, that's what it should do
Coke =-)
right.
the default can be --pirc=imcc, and at some point we can change the default, and at some point we can remove imcc.
cotto_work we need argument processing to be smarter before that can happen though 21:58
kj who could fix that?
Coke cotto_work: don't we already have config options?
(how hard can adding another one be?)
chromatic Processed by IMCC.
kj cotto_work: it's for perl configure, not parrot
cotto_work kj, right, but currently imcc does all argument processing instead of something compiler-independent. 21:59
kj cotto_work: what i (we?) mean is that it's an option that you pass to Configure.pl
not to the parrot executable
Coke I think his point is we can't not link in IMCC. 22:00
cotto_work Coke, yes
kj ah yes
that's true
but that's step 2
step 1 is adding the configure option :-)
the main Makefile should be fixed for that I guess 22:01
that IMCC will be compiled into pirc.lib or whatever
eh, pirc.o
Whiteknight dukeleto: all tests pass (some TODO passed) on win32 22:03
ship it
kj darbelo: you should update your flex to version 2.5.33; 2.5.4 is no good
Coke there. partcl now has a single makefile and no longer rebuilds things if you do make && make 22:04
darbelo kj: But I shouldn't have to. If the the generated files are there, there's no reason for me to regenerate them.
dalek rtcl: 15ff4b9 | coke++ | (7 files):
[build] Eliminate ops/pmc Makefiles (resolve issue #111)
shorten dalek's url is at xrl.us/bftbu4
dalek kudo: 073d5ed | jonathan++ | docs/ChangeLog:
Additions to ChangeLog.
 22:09
shorten dalek's url is at xrl.us/bftbvi
Coke so, public git branches seem about as merge-friendly as public svn branches, since you can't safely rebase them. neh? 22:16
chromatic I don't see the connection between rebasing and merging. 22:17
moritz Coke: if you rebase and publish as another branch, you still see all the commits 22:18
22:18 fperrad_ joined
moritz as opposed to svn, where you do the rebase as a big diff, and lose all the history for log/blame/whatever 22:18
and afaict git doesn't have a huge problem if you merge trunk into branch, and then later on branch into trunk 22:19
(please correct me if I'm wrong here, git gurus)
chromatic I can't imagine how it would. 22:20
Coke chromatic: if I merge, then all the master commits come in as a single commit rather than as they did in master.
jonathan I had a rebase of something I'd pushed go horribly pear-hsaped at the weekend.
I dunno what I did.
Coke if I rebase (which is what I want), then I have to destroy the branch and start over. 22:21
darbelo Whiteknight: ping
Coke (which is easier to do in git than it is in svn, sure.)
but if anyone was following that public branch, they're screwed when I rebase.
chromatic I don't understand "I have to destroy the branch".
Coke rewriting the history of a public branch is bad juju.
chromatic How often are you going to merge this branch back to master?
Coke just once when I'm done.
chromatic I don't see the problem with rebasing then. 22:22
Coke but if I rebase from master to branch often, I have to kill-and-redo each time.
chromatic Sure, but you can rebase once before you merge back as long as you don't push that rebase.
Coke chromatic: for long lived branches, that doesn't help. 22:23
jonathan You shouldn't push after rebasing?
dalek rrot-linear-algebra: e1dcb2d | darbelo++ | (6 files):
Move soirce files into src/ directory.
shorten dalek's url is at xrl.us/bftbxi
dalek rrot-linear-algebra: 6c2756e | darbelo++ | Configure.pl:
Add a Configure.pl script.
shorten dalek's url is at xrl.us/bftbxk
jonathan So *that's* what I did wrong...
dalek rrot-linear-algebra: 84b9740 | darbelo++ | config/Makefile.in:
Add dummy makefile template. It doesn't do anything, but adding it now makes future development easier.
shorten dalek's url is at xrl.us/bftbxp
dalek rrot-linear-algebra: 8d547d1 | darbelo++ | :
Merge branch 'master' of github.com:Whiteknight/parrot-linear-algebra
shorten dalek's url is at xrl.us/bftbxr
chromatic I'm not sure anything helps long-lived branches.
Coke git was supposed to. =-) 22:24
moritz maybe it does, if you merge instead of rebase
chromatic Git makes merging suck less.
moritz and git-rerere can help you record conflicts, thus making long-lived branches less pain to merge/rebase
s/conflicts/conflicts and resolutions/ 22:25
Coke but if I merge, how does that work? I end up with two different histories, which doesn't bother me so much, but when I merge that back to master, is git smart enough to avoid the merge commits from the other direction?
or do I then lose the branch history and get a single merge commit in master? 22:26
chromatic I *think* that depends if you use --squash, but I'm not sure.
moritz in general you don't get a single merge commit
unless you use --squash 22:27
but I'm not sure if git is smart enough to handle the merge conflicts
I hope it is ;-)
Coke will `git branch -d origin/tclint' do what I think? 22:28
moritz nope 22:29
Coke how does one delete an upstream branch?
chromatic It will refuse to delete that branch if you have unmerged changes between it and your current branch, as I understand it.
moritz github.com/guides/remove-a-remote-branch
github has lots of small guides for such questions
Coke ... wow, I never would have guessed that. 22:30
moritz git push origin :tclint
Coke moritz++ 22:33
moritz speaking of git woes...
Coke does dalek report on anything from github other than master?
moritz git status says: modified: S06-signature/named-parameters.t
but git stash says "No local changes to save" 22:34
wtf?
Coke: no
Coke moritz: did TimToady just freeze?
moritz excuse me?
darbelo Coke: dalek only reports what the RSS feed shows him.
Coke odd. after removing some branches, git branch -a still shows them. 22:40
dalek p-rx: 725c893 | pmichaud++ | src/ (3 files):
Clean up quote handling a bit. Work around TT #1125 by avoiding
 22:50
shorten dalek's url is at xrl.us/bftb3n
dalek p-rx: 86d5495 | pmichaud++ | src/HLL/ (2 files):
More quoting escapes, better handling of quote constructs.
shorten dalek's url is at xrl.us/bftb3p
dalek p-rx: 64c960f | pmichaud++ | src/NQP/ (2 files):
Add Q quoting form to NQP.
shorten dalek's url is at xrl.us/bftb3r
dalek p-rx: b917679 | pmichaud++ | src/NQP/ (2 files):
Add single angle bracket quoting to NQP (non-splitting for the moment).
shorten dalek's url is at xrl.us/bftb3t
22:56 plobsing joined 23:03 jsut_ joined
plobsing seen kid51? 23:08
purl kid51 was last seen on purl 23 hours, 57 minutes and 6 seconds ago, saying: <private message>
darbelo "last seen on purl" ... 23:09
plobsing s/on/by/ ??
darbelo I don't really wan to know that much. 23:10
dalek rrot-plumage: 61fade4 | japhb++ | :
[META] Many TODO additions
shorten dalek's url is at xrl.us/bftb64
dalek rrot-plumage: dd234b0 | japhb++ | :
[TOOLS] import_proto.p6: Calculate authority based on repo home
shorten dalek's url is at xrl.us/bftb66
Whiteknight did the release land? 23:14
darbelo Whiteknight: Not yet. 23:15
Whiteknight does the duke need any more help with it?
darbelo Whiteknight: I pushed the start of a build infreastructure to parrot-linear-algebra, can you give it a look?
dalek rrot-plumage: 48e28b6 | japhb++ | :
[TOOLS] import_proto.p6: Refactor make_meta_file()
 23:16
shorten dalek's url is at xrl.us/bftb7x
23:17 mikehh joined
Whiteknight darbelo: yeah, I will. lots to do tonight but I'll check it out 23:18
darbelo It doesn't do anything yet, but I wanted to run it past you to see if you agreed with the general shape. 23:19
Whiteknight okay, how do I get your changes from the server? "git pull" gives me an error 23:20
darbelo what error?
purl darbelo: Operation not permitted
darbelo purl: jump off a bridge.
purl darbelo: huh?
darbelo purl: die
purl die is DEAD! Don't you read the PDDs? Loser.
Whiteknight fatal: 'master': unable to chdir or not a git archive 23:21
darbelo wtfbbq?
plobsing try git pull origin maybe 23:22
Whiteknight You asked me to pull without telling me which branch you 23:23
want to merge with, and 'branch.master.merge' in
your configuration file does not tell me either.
do I need to set something in my config? 23:24
darbelo Whiteknight: Are you using branches in your local repo?
Whiteknight darbelo: no branches. I had a repo, I pushed it to github, and I want to get changes 23:25
darbelo Hmm. Can you try to clone a new version of the remote repo? 23:26
git clone git@github.com:Whiteknight/parrot-linear-algebra.git 23:27
23:27 jsut joined
darbelo in another place 23:27
Whiteknight that does it 23:29
darbelo I'm guessing that your other repo, by virtue of having started localy, is lacking some config or another to correctly talk to github. 23:30
Whiteknight yeah, I'm sure I'm missing some kind of configuration detail somewhere 23:31
darbelo You could comparing the .git/config files for the started-local and upstream-clone repo. 23:32
s/comparing/compare/ 23:33
plobsing why bother, you probably have a perfectly good repo now
darbelo academic curiosity ;)
23:35 cconstantine joined
darbelo tries to recover what little FORTRAN knowledge he once had. 23:37
Whiteknight darbelo: makefile stuff is good, yes
but, maybe we should write our Configure in Parrot instead of Perl5
darbelo Whiteknight: Excelent!
Whiteknight that's more of a nit-pick though, nothing we need to do now
darbelo Want pir on nqp?
darbelo has examples of both. 23:38
;)
Whiteknight Anything that ships with Parrot
so NQP if it's easy
darbelo NQP is a 'cp' away. Plumage has a NQP Configure, the only catch is that it doesn't support #IF() statements in the makefile templates. 23:40
But it allows you to define your own @vars@ to make up for it. 23:41
japhb And I'd be happy to make it smarter if really needed.
23:42 szabgab joined
darbelo japhb: Nah, it's fine as it is. The ability to add to the config hash at runtime is much more powerful than #IF() in the templates. 23:42
mikehh All tests PASS (pre/post-config, smoke (#29241), fulltest) at r41963 - Ubuntu 9.10 (beta updated) amd64 23:43
Whiteknight we should probably add an NQP-based configure for matrixy too 23:44
lots to do!
23:45 patspam joined
japhb It's now in the Plumage TODO to make a proto-like "make a new project skeleton" command, and I'll probably use the Plumage's own Configure.nqp (or a close relative) in the skeleton 23:46
darbelo japhb++
Whiteknight ah, the test suite runs! Awesome 23:54
darbelo where?
Whiteknight matrixy
I might have to make a commit first 23:55
23:55 Limbic_Region joined
darbelo Oh yeah. All it needs infrastructure-wise now is to install. 23:56
After that it's all language and library work.
cotto_work plumage?
purl plumage is the future Parrot module ecosystem. It will include tools to search metadata, handle dependencies, install modules, and so forth. The repository is at gitorious.org/parrot-plumage/parrot-plumage and the design docs are at trac.parrot.org/parrot/wiki/ModuleEcosystem

Powered by Cro and the Raku® Programming Language.

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

Thank you!