Parrot 1.7.0 "African Grey" is out! | Testing hackathon November 14 and 15 -- improve opcode coverage | find out what's up with the slice opcode | Latest modified TT's: icanhaz.com/parrotbugs
Set by moderator on 10 November 2009.
japhb #perl6++ # Yet again with the sane discussion of contentious topics 00:01
darbelo karma #perl6 00:02
purl #perl6 has karma of 1
darbelo karma perl6
purl perl6 has karma of 21
darbelo karma perl5
purl perl5 has karma of 8
darbelo karma #perl5
purl #perl5 has neutral karma
darbelo karma #perl
purl #perl's karma is greater that the sum of its participants'
Whiteknight darbelo: just added a CharMatrix2D type 00:16
dalek rrot-linear-algebra: bb83303 | Whiteknight++ | .gitignore:
add a .gitignore file to clean things up a little bit
 00:17
rrot-linear-algebra: 6e7f275 | Whiteknight++ | config/Makefile.in:
fix makefile to generate proper install paths
rrot-linear-algebra: 6c97c90 | Whiteknight++ | src/pmc/nummatrix2d.pmc:
NumMatrix2D now does matrix
rrot-linear-algebra: f0a8920 | Whiteknight++ | (2 files):
pmcmatrix2d now does matrix
rrot-linear-algebra: 6b96bac | Whiteknight++ | (3 files):
add a new CharMatrix2D type. It's a bare-bones copy+paste from NumMatrix2D right now
darbelo Whiteknight++ 00:18
does it eat kittens?
Whiteknight not yet 00:19
it doesn't do much of anything
darbelo nopaste.snit.ch/18665
pmc2c--
cotto_work darbelo++ for making a compelling case for getting back to work on opsc/pmcc 00:22
Infinoid diakopter: What was the dalek breakage? If I break it now, will it get in your way? :) 00:23
00:23 mikehh joined
diakopter Infinoid: it seemed to fix itself 00:23
Infinoid when?
I installed some stuff from cpan to make it happy, yesterday. 00:24
(and just now I installed one more thing to make the test suite happy)
darbelo cotto_work: Last I looked at pmcc it looked like it could be mostly coerced to perform pmc2c's current job, even if no syntax for lorito solidyfies. 00:25
cotto_work That's step 1 of the plan. 00:26
00:26 dalek joined 00:30 dalek joined
darbelo So, all that needs to be done is copy the contents of compilers/pmcc to a new branch, plug it into the makefile and start hacking? 00:30
cotto_work darbelo, yes. I'd recommend starting with ops2c, though, since that project is expected to be smaller. 00:31
Infinoid msg moritz I pulled from your dalek-plugins fork, but I'm not sure it's all quite happy. Botnix doesn't seem to like it... gets confused over the package name. You wouldn't happen to have a test case, would you? 00:33
purl Message for moritz stored.
cotto_work but work on either would be great
Infinoid msg moritz Perhaps we should pass in a list of secondary branches, and handle all of the branches from one timer? 00:42
purl Message for moritz stored.
dalek rrot: r42468 | jkeenan++ | branches/configtests (3 files):
Convert one Parrot::Configure attribute from get_p5 to get(x_provisional).
chromatic Non-portable exit(): perl5.git.perl.org/perl.git/commitdiff/81693ff9 00:49
dalek rrot: r42469 | jkeenan++ | branches/configtests/config (3 files):
Convert P::C attribute 'scriptdirexp' from get_p5 to _provisional.
rrot: r42470 | jkeenan++ | branches/configtests/config (2 files):
Convert P::C attribute 'sig_name' from get_p5 to _provisional.
 00:52
nopaste "plobsing" at 76.67.61.178 pasted "yet another libjit x64 fix attempt" (20 lines) at nopaste.snit.ch/18666 00:56
plobsing hi btw
anyone on x86_64 want to try it out? 00:57
cotto_work why not. It'll be fun.
dalek rrot-plumage: bb5d225 | japhb++ | :
[CORE,TESTS] Work around platform-specific spawnw differences (improve d...
00:57 abqar joined
dalek rrot-linear-algebra: 16b0f1d | Whiteknight++ | src/pmc/charmatrix2d.pmc:
fix indexing on CharMatrix2D so it now does what I think it should do
 00:59
rrot-linear-algebra: 707396b | Whiteknight++ | src/pmc/charmatrix2d.pmc:
add a set_string_keyed_int method so we can set a row at a time
cotto_work plobsing, it looks like that worked. I'm double-checking. 01:00
japhb plobsing++ # perseverance 01:02
plobsing (horrible hacky workaround)--
cotto_work yup. it works 01:04
plobsing++
plobsing woohoo!
can someone commit that patch to the libjit_framebuilder branch?
01:05 patspam joined
kid51 plobsing: I applied it to branch (though I haven't tested it.) 01:13
plobsing kid51++ 01:14
dalek rrot: r42471 | jkeenan++ | branches/libjit_framebuilder/config/gen/libjit/frame_builder_libjit_c.in:
applying patch by plobsing++
 01:15
rrot: r42472 | jkeenan++ | branches/configtests/config (2 files):
Convert P::C attribute 'd_socklen_t' from get_p5 to _provisional.
 01:22
rrot-plumage: 1ce4085 | japhb++ | :
[plumage] call_flattened() -> new | operator
 01:30
rrot-plumage: ed3d54b | japhb++ | :
[CORE] Glue.pir: call_flattened() is no longer needed; remove it
01:35 mikehh joined
cotto_work pmichaud, does nqp-rx support mmd without diving down into pir? 01:35
alternately, s/does/will/ 01:36
japhb I'm ready to make use of PerlJam's new statement modifier snytax for NQP in Plumage ... can someone do the honors of updating the parrot copy? 01:42
cotto_work chromatic, an interesting extension to your realias script might be to use $PROMPT_COMMAND in bash to reread ~/.aliases every time you run something from the bash prompt.
cotto_work predicts that this will become a common dance
japhb cotto_work: If so, that will indicate much progress in NQP features. I for one welcome my dancing overlords. 01:43
chromatic That seems like a lot of extra work, cotto_work. I don't add new aliases that often. 01:44
cotto_work It's not much work. just put PROMPT_COMMAND='source ~/.aliases' in your .bashrc
chromatic I meant for every command to run. 01:49
japhb cotto_work, I'm guessing he means making bash (even) slower
cotto_work technically yes, but I wouldn't expect it to be perceptible
01:52 mj41_ joined
chromatic I don't change aliases frequently enough that that's a problem. 01:52
cotto_work I usually have a couple dozen shells open in various screen sessions, so for me it's worthwhile. To each his own. 01:53
japhb Yeah, it's the pain of updating every open shell when you *do* change aliases that's the killer. 01:54
cotto_work exactly
dalek rrot-linear-algebra: 472bfdd | Whiteknight++ | src/pmc/charmatrix2d.pmc:
add a new method to initialize a row in CharMatrix2D from a grabbag of types
 01:58
rrot: r42473 | jkeenan++ | branches/configtests (13 files):
Convert more P::C attributes away from get_p5. t/steps/auto/alignptrs-02.t still has failures.
02:03 mikehh joined 02:09 petdance joined 02:13 eternaleye joined 02:15 eternaleye joined
japhb pmichaud, would you mind refreshing parrot's nqp-rx, please? Thanks! 02:42
02:49 tetragon joined 02:56 jsut_ joined
dalek rrot: r42474 | jkeenan++ | branches/configtests/t/steps/auto/alignptrs-02.t:
Fix test so that difficult-to-reach branch is tested.
 03:35
kid51 plobsing: You've got hard tabs around line 529 of config/gen/libjit/frame_builder_libjit_c.in, causing codingstd errors in branch. 03:38
Otherwise libjit_framebuilder passes make fulltest on Linux/i386. 03:40
dalek rrot: r42475 | jkeenan++ | branches/libjit_framebuilder/config/gen/libjit/frame_builder_libjit_c.in:
Eliminate leading hard tabs, which broke both tabs and c_indent codingstd tests.
 03:41
03:45 hercynium joined
dalek rrot: r42476 | jkeenan++ | branches/configtests/t/steps/auto/arch-01.t:
Eliminate one superseded test.
 03:45
03:48 janus joined
pmichaud cotto_work: yes, I expect that nqp will support mmd w/o pir. I'm still figuring out how I want to do that. 04:29
japhb: will refresh nqp-rx, yes.
japhb Thanks! 04:45
05:36 theory joined
dukeleto 'ello 05:43
diakopter 'i
dukeleto: sry. I nearly resisted. 05:44
echo 'sub a(:$x,:$y) { $x - $y }; say(a(1,6))' |./nqp --target=past | v8 past.js
I'm actually not sure how/why that is working. 05:45
5
oh. now I see why (not). :) 05:46
(because say(a(y:(1),x:(1))) produces the same result, so it's a fluke; whew.) 05:47
dukeleto diakopter: what have i missed in the nqp-rx world?
nqp-rx is in parrot core, i hear
diakopter SUB([stuff]) dtrt, now... duff/perlpilot added trailing if/unless 05:48
I'm working on a new backend for its --target=past output, obviously
diakopter watches github.com/perl6/nqp-rx/commits/master/ 05:50
dukeleto diakopter: i just do git pull :) 05:56
diakopter pmichaud: typo in 12-logical.t logicl # I haz lurnd git; I can haz commitbit plz?
I do pull, except when it won't merge over my changes even when I've committed them. :/ 05:57
still learning.
purl hmmm... learning is not wasting time
diakopter learning.
purl learning is, like, not wasting time
diakopter purl: die in a flood of your own blood
purl diakopter: i'm not following you...
diakopter needs to remember not to feed the troll bot 05:58
diakopter chokes purl with a cookie 05:59
diakopter implementes pirop def_or
06:11 dukeleto left, dukeleto joined
dukeleto 'ello 06:11
dalek rrot: r42477 | dukeleto++ | trunk/NEWS:
[cage] Update NEWS for the upcoming release
 06:49
07:05 zak_ joined
dukeleto i will ask again: i want to document RTEMS work in the parrot repo. where should i put this document? 07:07
cotto docs/dev might be a good place 07:18
dalek rrot: r42478 | dukeleto++ | trunk/t/op/bitwise.t:
[t][TT #1119] Convert t/op/bitwise.t to PIR
cotto putting it in the wrong place makes it easy to move it to the right place 07:20
moritz good morning 07:26
07:27 uniejo joined 07:35 fperrad joined 07:53 davidfetter joined
fperrad ping japhb 07:54
purl I can't find japhb in the DNS.
08:12 mikehh joined 08:18 iblechbot joined
dalek rrot: r42479 | mikehh++ | trunk/MANIFEST.SKIP:
regenerate MANIFEST.SKIP
 08:28
rrot: r42480 | mikehh++ | trunk/ext/nqp-rx/src/stage0 (2 files):
fix codetest failures - Pod syntax
 08:51
dukeleto 'ello 09:20
mgrimes++ for t/op/bitwise.t conversion
a patch status of "partially applied" would be REALLY useful on trac.parrot.org 09:23
moritz Infinoid: sorry, I don't have a test case, I don't even know how to run dalek locally. And while I agree that a list of branches to try_link is cleaner, I don't understand the setup well enough to do that change 09:26
dalek rrot: r42481 | dukeleto++ | trunk/t/op/sprintf2.t:
[t][TT #1122] Convert t/op/sprintf2.t to PIR, mgrimes++
 09:34
09:34 payload joined
dukeleto darn, i put the wrong tt in the commit msg 09:35
o well
moritz dukeleto: the best you can do is to reply to the commit mail (to parrot-dev)
09:37 payload1 joined
dukeleto moritz: thanks, that is a good idea 09:37
moritz not my own; I think Coke++ once told me after I wrote a wrong commit message :-) 09:39
09:39 payload joined
dalek TT #1119 closed by dukeleto++: [PATCH] Converted tests from perl to pir 09:40
rrot: r42482 | dukeleto++ | trunk/t/op/number.t:
[t][TT #1122] Convert t/op/numbert.t to PIR, mgrimes++
 09:44
09:48 payload joined 09:58 payload joined
dalek TT #1122 closed by dukeleto++: [PATCH] Convert more tests from perl to pir 10:20
rrot: r42483 | dukeleto++ | trunk (2 files):
[t][TT #1122] Convert t/op/literal.t to PIR and keep old PASM tests in t/op/literal-old.t, mgrimes++
Infinoid moritz: I don't run dalek locally, either. I have a fake dalek module that gets run for "make test" 10:24
moritz: I'll try to find some tuits for secondary branches, but no guarantees
moritz Infinoid: I have another patch attempt coming up, but so far I didn't have the tuits to set up all the dependencies for testing 10:26
Infinoid testing should (hopefully) be standalone 10:27
I think it needs a couple of XML modules from CPAN, and probably Perl6::Export::Attrs from CPAN too. Otherwise it should Just Work.
I can help with that if you're running into any issues with it 10:29
10:30 payload joined
moritz github.com/moritz/dalek-plugins/com...127473948c 10:31
my attempt to handle multiple branches per github repo gracefully 10:32
not yet tested
Infinoid I think I want to tweak that a bit to pass in the branch name so it can adjust the output prefix, but yeah, that's what I was thinking of 10:33
moritz feel free to steal from it whatever you like
Infinoid thanks! 10:34
I think it would be nice for the bot to output "rakudo: blah blah" for master commits, and "rakudo ng: blah blah" for ng branch commits
moritz Infinoid: after installing the required modules all tests fail like this: 10:37
t/gitorious.t ........ Can't locate object method "fetch_url" via package "modules::local::karmalog" at /scratch/mlenz/dalek-plugins/modules/local/karmalog.pm line 63.
I tried an older versions (without modifications by me), still the same
Infinoid checks whether he's hacking PERL5LIB somehow, or something like that 10:38
I think it's relying on "use lib getcwd();" to add /scratch/mlenz/dalek-plugins to your lib path
moritz + urls => \\@@rss_links, 10:40
that should just be one @, of course
Infinoid Ok, I mis-read your error message. It's obviously finding the module, but somehow it's running one method in the package and failing to find another method in the same package? 10:46
What rev of perl5?
moritz 5.10
Infinoid ok, that's what I'm testing on 10:47
nopaste "Infinoid" at 173.75.243.182 pasted "moritz: Does simple module loading work in your checkout?" (4 lines) at nopaste.snit.ch/18669 10:53
moritz Infinoid: nope, dies with the same error message 10:54
Infinoid ok. I'm hoping to find out why :)
perl -e 'use lib "."; use YAML; do "modules/local/karmalog.pm"; print(Dump(\\*modules::local::karmalog::))' 10:57
nopaste "moritz" at 132.187.31.74 pasted "dumping for Infinoid++" (13 lines) at nopaste.snit.ch/18670 10:59
moritz anyway, off to lunch now, bbiaw 11:00
Infinoid moritz: When you get back, I think it will work if you remove "use YAML::Syck;" from the beginning of emit_ticket_karma()
Infinoid is once again reminded of how "do" is evil
Infinoid pushes that 11:03
11:31 payload1 joined 11:33 payload1 joined 12:43 joeri joined 12:47 bluescreen joined 12:58 mariano__ joined 12:59 payload joined 13:06 plobsing joined
plobsing hi #parrot! 13:06
13:13 whiteknight joined
whiteknight hello #parrot 13:15
plobsing hi whiteknight 13:16
13:18 szabgab joined 13:19 bacek joined 13:24 sothil joined 13:33 Zak joined 13:49 Austin_away joined 13:59 davidfetter joined 14:11 davidfetter joined
mj41 TapTinder - build status is fast now ... tt.ro.vutbr.cz/buildstatus/pr-Parrot/rp-trunk 14:19
14:44 mikehh joined 15:17 Psyche^ joined 15:23 iblechbot joined 15:26 patspam joined 15:39 sothil_ joined
japhb fperrad, I have to run, but I'll be back in half an hour or so ... was there anything other than your two Plumage-related emails last night that I need to know about? 16:02
16:03 Andy_ joined
dalek p-rx: 9555dc3 | pmichaud++ | src/cheats/ (2 files):
Move <ternary> subrule into HLL::Grammar.
 16:25
p-rx: b6a6b03 | pmichaud++ | src/cheats/hll-grammar.pir:
Fix bug in preclim handling of EXPR() subrule.
p-rx: cf48f55 | pmichaud++ | src/Regex/Cursor.pir:
Allow Grammar.parse() to take the name of a subrule as well as a rule.
p-rx: 130a2bb | pmichaud++ | src/Regex/Cursor.pir:
Change rule argument in Cursor.parse to be a named argument.
purl dalek: that doesn't look right
p-rx: 344775f | pmichaud++ | src/stage0/ (4 files):
Update bootstrap versions.
16:25 theory joined
cotto_work < purl-- > 16:25
also, good morning
16:28 darbelo joined
whiteknight good morning cotto_work 16:29
dalek rrot: r42484 | pmichaud++ | failed to fetch changeset:
[nqp]: Update NQP to have statement modifiers, bugfixes.
 16:33
pmichaud I'm getting failures in trunk... known? 16:36
whiteknight not that I am aware of 16:37
building now (will take a long time, am on windows)
pmichaud t/op/bitwise.t (Wstat: 0 Tests: 67 Failed: 0) Parse errors: Bad plan. You planned 68 tests but ran 67. 16:38
t/op/number.t (Wstat: 0 Tests: 173 Failed: 1) Failed test: 172
whiteknight urg
mikehh I don't get any on Ubuntu 9.10 amd64
pmichaud I'm on Kubuntu 9.04 x86 16:39
I'm guessing someone might not have updated a test properly.
looks like those tests were converted to PIR in r42482 and r42478 16:41
uh oh 16:42
oh, nm
cotto_work those can be tricky to verify
pmichaud it might be related to availability of libgmp
NotFound I've found a problem that may be the cause of some random failures... wait a few moments... 16:43
16:43 mikehh joined
nopaste "cotto_work" at 131.107.0.74 pasted "Configure.pl failures on win7/mingw_gcc" (97 lines) at nopaste.snit.ch/18672 16:46
whiteknight t/op/number.t:131 failed on Win32/strawberry 16:49
a couple other TODO tests are passing 16:50
131 is a stupid -0 test
it appears that we want -0, but windows is giving us regular 0 instead. Any word on whether this is right or not? Seems wrong to me 16:51
darbelo Everyone else gives -0, windows has it wrong.
whiteknight so we *want* -0? 16:52
particle whitenight: msvc? what version?
whiteknight that seems absolutely absurd to me
particle iirc msvc 7 had a bug
whiteknight particle: not msvc. Strawberry on Windows
particle ah
whiteknight I need to try to remember the incantation to build here with msvc 16:53
I used to know it
particle ITS NOT WINDOWS, IT'S THE COMPILER
whiteknight whatever
I'm on windows. Have test failure
dalek rrot: r42485 | NotFound++ | trunk/src/pmc/object.pmc:
[pmc] call to overrided object.invoke lacked a check for 0 positional arguments
darbelo particle: Actually the math library, but yes.
particle yeah, the math lib that ships with the compiler
msvc7 had different behavior than all other msvc's. i can't tell you about strawberry, though 16:54
i suggest you hop over to #win32 and ask 16:55
17:08 jan joined 17:26 Ingmar joined
whiteknight I'm still flabbergasted that -0 is a behavior that we want and encourage 17:31
particle it's ieee-specified 17:36
falls out naturally from the storage format for floats, and it's used in some limited cases (weather, physics) 17:37
japhb I actually had as a prof one of the guys that defined that behavior. (And also wrote the microcode for the 8087 and 80387, for that matter)
He explained at length why -0 matters. 17:38
The shortest version of it is: If you don't have -0, your fluid dynamics simulations are wrong, and you design crappy wings, and your planes don't fly right. 17:39
cotto_work Getting my planes to fly will take a lot more than -0. 17:41
whiteknight I know plenty about IEEE754, I've done work on floating-point hardware before. So I know *why* -0 exists in the technical sense 17:44
I'm also not swayed by the vague "I know a guy who knew a guy who claimed it mattered once"
japhb whiteknight, I can (if I dig hard enough in my brain) actually explain the details. But it's been nearly two decades, so a summary is a tad easier. :-) 17:46
whiteknight but I have never seen a concrete case where a correct solution to a problem required the use of -0, or the rule "0 != -0" was important
japhb And W. Kahan is the kind of guy that after several years of classes with him, even skeptics begin to just trust him. He spends all of his free time proving why slight incompatibilities with the spec result in horribly wrong answers. ;-) 17:47
17:47 bubaflub joined
PerlJam whiteknight: it matters when you need to propagate the correct sign to the result of a modulus 17:48
dalek TT #1267 created by jkeenan++: t/op/bitwise.t: Test failure subsequent to skip BigInt
whiteknight PerlJam: how so? Example?
PerlJam (okay, okay, I just made that up :)
whiteknight ok
PerlJam I've never seen a case where -0 matters either. 17:49
whiteknight I've seen plenty of cases where -0 causes unexpected problems, and plenty of places where people have to check for and correct that value to "normal zero" to prevent problems
Even wikipedia doesn't have more then vague explanations for why it might be useful in some strange cases to a very limited number of people 17:52
japhb whiteknight, here's one example, related to the summary I said above. Generally, airfoil fluid dynamics simulations (at least during that time period, I'm not up on the latest algorithms) would follow a discrete number of individual flow surfaces. When a flow hit a surface perfectly perpendicularly, the code would split the flow into two new flows, flowing "up" and "down". When -0 gets folded to 0, that would result in both flows contri
buting to "up", and none contributing "down". After that point, all of the calculations go to hell.
whiteknight japhb: well that's just a case of using a very lazy numerical algorithm to compute the lift contributions 17:54
japhb It's related to why we "round to even" ... if we don't, there's a tiny bias across bazillions of random roundings, and in fact this becomes terribly obvious on the scale of the NYSE or a major national chain like Target or Walmart.
darbelo lazy numerical algorithms are faster than the hard-working kind. 17:55
japhb: goolgle "telco benchmark" for related numerical fun.
Eh, google.
PerlJam japhb: cite a whitepaper that explains this :) 17:56
japhb Kahan was endlessly bitter about people optimizing their chip to do something faster than the strict IEEE spec allowed, because they didn't understand the edge cases. Similar to the infamous Pentium and Excel math bugs, *most of the time* the error doesn't matter. When it does, though, your answer is completely wrong. 17:58
darbelo PerlJam: We don't need any whitepapers. We can just refer to the standard, the standard is perfect, that's why it's the standard.
japhb PerlJam, I've got binders of the stuff on dead trees. I don't know how much actually made it onto the internet.
darbelo, :=P
Will look, however.
PerlJam then I shall continue to blithely assume that people smarter than me have figured it all out and made sensible decisions regarding numbers on computers until then :) 17:59
japhb PerlJam, www.cs.berkeley.edu/~wkahan/ 18:00
Prepare to have your eyes cross.
darbelo "Much Ado About Nothing's Sign Bit" 18:02
:)
japhb www.cs.berkeley.edu/~wkahan/ieee754status/ and www.cs.berkeley.edu/~wkahan/ieee754...ieee754.ps for IEEE spec specific discussion, if I'm skimming this right.
darbelo, and there you have it. :-) 18:03
Like I said, a fair portion of this is still on dead trees. He was constantly posting papers printed on reams of dot matrix printout outside his office.
"How JAVA's Floating-Point Hurts Everyone Everywhere" 18:08
Not much for pulling punches, either.
bubaflub not to interrupt the discussion, but make -j is broken on trunk 18:09
darbelo Didi'n I fix that already this week? 18:10
japhb I had another professor nicknamed "Trivial" Hilfinger, because he thought everything was trivial and couldn't understand why his students didn't just know everything off the top of their heads. Kahan was the only person I've ever seen leave Trivial looking dumbfounded. Best day of my college career. 18:11
darbelo Heh, I know people like that. 18:12
"If I found that trivial, I wouldn't be on this classroom, okay?" 18:13
japhb YUP
dukeleto 'ello 18:14
japhb: mathematicians use the word trivial to be lazy 18:15
japhb Ooh, Superghost! Uh, I pick 'h' ... oh DAMN!
dukeleto japhb: and or: there are only two types of problems in the world: trivial and unsolved.
whoa, you guys studied fluid dynamics as well? scary. 18:16
dukeleto attempts to backlog
if you read ieee754-2008, they have a lot to say about -0 :) 18:17
CLA?
purl CLA is Contributor License Agreement or www.perlfoundation.org/contributor_..._agreement or www.parrot.org/foundation/legal or www.parrot.org/files/parrot_cla.pdf or www.lowcarbfriends.com/bbs/main-low...-acne.html
japhb dukeleto, Trivial really thought so. Our final exams were supposed to be three hours. His always had to be the last ones of the day, because it generally took people twice that long -- the first person to finish in my huge class was done at like the 3.5 hour mark (and I wasn't anywhere close, sigh). When pushed on this, he said "I did the test myself in 20 minutes!" and seemed to honestly believe that meant the test should have taken th
e students not much longer.
dukeleto japhb: i had the same experience in grad skool 18:18
japhb: most people took 5-6 hours to finish the "party"
japhb heh
dukeleto japhb: i had a professor that would *only* refer to tests as "parties" and homework as "treats" 18:19
exclusively
he was a turbulence plasma physicist
and i think he gave some of the hardest/trickiest problems i have ever had the pleasure to bash my head against
japhb :-) 18:20
18:28 bubaflub_ joined
whiteknight I had a reputation in college as a fast test-taker. I don't think there was a single final exam in my entire undergraduate career that I didn't walk out of first 18:28
I knew two things really well: First, I always knew which problems I was and was not able to solve, and I also knew where the curve was going to be 18:29
dukeleto bubaflub_: salutations
why doesn't Configure.pl --ask ask about wanting GMP or not? 18:30
japhb whiteknight, nice technique
whiteknight a lot of our tests were open-notes, so many students would spend lots of valuable minutes digging through old paperwork. I wouldn't bring notes, because I knew I wouldn't be able to find anything useful in a reasonable amount of time 18:32
darbelo I do a similar thing, except I don't really take notes. 18:36
whiteknight haha, I rocked that strategy in a few classes too :)
"english literature", "art and aesthetics", "the history of europe", etc 18:37
particle yawns
japhb particle, damn you! 18:38
darbelo "the history of europe"? I thought you had a tech degree.
japhb Not like I'm not already tired
particle :)
whiteknight darbelo: Yeah, EE. But I had to take a bunch of "general education" classes
18:38 bubaflub joined
whiteknight all the tech classes were interesting, and I took notes 18:39
dukeleto open notes tests are way harder
that is the kick in the pants 18:40
18:40 AndyA joined
dukeleto who broke -j again? 18:40
dukeleto needs a good punishment for breaking -j
japhb Open notes tests were the professor's way of saying "It doesn't matter how many facts you know. These problems will still kill you."
dukeleto, use 1 core for standard make. Use the other N-1 for loading dozens of tabs from a Firefox folder while you wait for the build to finish. 18:41
darbelo dukeleto: Works for me, -j5 just passed here. If you nopaste the error I'll see about fixing it. 18:42
Could be a BSD vs GNU make difference.
dukeleto darbelo: me and bubaflub++ are both seeing -j as broke on OS X 18:43
bubaflub: can you nopaste the opengl -j build error? 18:44
bubaflub yep, i reported that about 40 minutes ago
sure, one sec
dukeleto i am going to put the parrot testing hackathon on calagator.org/ 18:45
japhb wakes up slightly ... opengl fail?
dukeleto japhb: opengl and -j don't like each other in the build
japhb: at least on OS X
japhb Hmmm. It looks like OpenGL is just generally unhappy ... 18:46
$ parrot examples/opengl/triangle.pir
invalid signature string element v!
Oy. I may need to take a nap before facing this morning.
dukeleto japhb: opengl examples have never worked for me on OS X 18:48
japhb As a matter of fact, I'm going to. dukeleto, just configure --without-opengl for now (IIRC)
dukeleto japhb: thanks, that is a reasonable workaround for now
japhb dukeleto, sigh. Things to look at later. :-(
dukeleto japhb: delegation! learn to love it.
japhb dukeleto, I seem to be the only one who cares to expend any energy on it. :-/ 18:49
dalek p-rx: 8d6dfe2 | duff++ | (4 files):
[nqp] Naive while/until loop modifiers
dukeleto japhb: bubaflub++ seems interested in opengl, we might be able to get him to throw some tuits at it 18:50
japhb bubaflub, you are more than welcome to do so. ;-)
bubaflub righteous. 18:51
purl stick it to the man!
bubaflub i'm about to nopaste it
japhb bubaflub, sorry to walk away, but I've really got to stop this headache. I'll take a look later if no one beats me to it. 18:52
bubaflub hey, no problem japhb feel free to email me or whatever
i.e. to get me to help on OpenGL stuff 18:53
dukeleto bubaflub: japhb is the man to talk to about opengl issues, but we really need more people hacking on our opengl bindings
bubaflub: if you can just verify that the opengl examples work/don't on your box, that is a start
bubaflub: but don't hope for much :)
bubaflub *sigh* 18:54
dukeleto and why don't the tests for the examples fail?
maybe the opengl examples have no tests?
darbelo dukeleto: "does this triangle look right to you?" is hard to put into a test :) 18:56
bubaflub unless it's missing a vertex 18:57
dukeleto darbelo: "does this triangle segfault" is easy to test, though :)
dalek tracwiki: v7 | dukeleto++ | AllHackathons 19:02
tracwiki: trac.parrot.org/parrot/wiki/AllHac...ction=diff
tracwiki: v8 | dukeleto++ | AllHackathons
tracwiki: trac.parrot.org/parrot/wiki/AllHac...ction=diff
dukeleto why can't a make a link to a new freakin' page on the wiki? 19:03
i write SomeJunkLikeThis and i don't get a link to make a page
it's like, the one sole purpose of a flippin' wiki, to do that, right?
i hate our wiki 19:04
whiteknight yeah, the trac wiki is far from full-featured
19:05 chromatic joined
dukeleto whiteknight: what about just having one feature? i would like that, right now 19:06
whiteknight yeah, that would be nice too
dukeleto why doesn't SomeWikiJunk make a page link? 19:07
whiteknight the ability to use a little bit of HTML would be killer
dukeleto what do I have to do to make a new page? this has been blocking me for a while and now it is making me angy
angry, even
bubaflub japhb: got some logs for ya, going to nopaste em in the channel
dukeleto whiteknight: we should use something like ikiwiki.info/
nopaste "bubaflub" at 76.199.4.75 pasted "mac os x configure with opengl warnings" (97 lines) at nopaste.snit.ch/18673 19:08
"bubaflub" at 76.199.4.75 pasted "mac os x make -j 4 with opengl failure" (190 lines) at nopaste.snit.ch/18674
bubaflub japhb: line 152 is the relevant line in the make failure, line 72 in the configure warnings 19:09
19:10 mikehh joined
dukeleto calagator.org/events/1250458003 <-- testing hackathon 19:13
19:14 joeri joined
dukeleto www.parrotcode.org/docs/tests.html is all wrong 19:14
darbelo parrotcode.org is obsolete. 19:15
whiteknight kill it with fire
darbelo It should be made into a redirect to the right place.
dukeleto: If you want to create a page, just navigate to the nonexistant page. There'll be a "Create this page" button there. 19:16
whiteknight redirect++ 19:26
dukeleto darbelo: that is just all kinds of wrong in the wiki world
19:26 mokurai joined
dukeleto darbelo: parrotcode.org comes up before anything else in google searches 19:27
why is it there?
who controls it?
darbelo dukeleto: Yes. On both counts.
It's there becouse nobody removed it, don't know who owns it. Somebody on the PaFo board shouold know. 19:28
19:29 soxet joined
dukeleto so if i yell PLEASE FREAKIN' DELETE PARROTCODE.ORG in this channel, someone should hear me? 19:33
darbelo Only if they backlog, so yell it often ;)
... or DDoS the domain or something. 19:34
dalek tracwiki: v1 | dukeleto++ | OpcodeTestingHackathon 19:39
tracwiki: trac.parrot.org/parrot/wiki/Opcode...ction=diff
cotto_work dukeleto, it looks like it points at parrot.org now 19:45
s/points at/redirects to/
dalek tracwiki: v2 | dukeleto++ | OpcodeTestingHackathon 19:46
tracwiki: trac.parrot.org/parrot/wiki/Opcode...ction=diff
19:46 soxet joined
dukeleto cotto_work: only the index page redirects. direct links do not 19:46
dalek tracwiki: v3 | dukeleto++ | OpcodeTestingHackathon 19:49
tracwiki: trac.parrot.org/parrot/wiki/Opcode...ction=diff
tracwiki: v4 | dukeleto++ | OpcodeTestingHackathon
tracwiki: trac.parrot.org/parrot/wiki/Opcode...ction=diff 19:50
moderator Parrot 1.7.0 "African Grey" is out! | Testing hackathon November 14 and 15 -- trac.parrot.org/parrot/wiki/Opcode...gHackathon | find out what's up with the slice opcode | Latest modified TT's: icanhaz.com/parrotbugs 19:52
darbelo "find out what's up with the slice opcode"? 19:56
chromatic We talked about that in a #ps some two or three weeks ago. That's all I remember. 19:58
19:59 Austin joined
Austin effing chatzilla 19:59
purl++
Should there be a has_attribute vtable function? 20:00
Wow. I've been chosen to receive a UN grant of $1,000,000. 20:01
I can stop wasting my time on this open source stuff, and concentrate exclusively on cocaine and strippers!
Woo-hoo.
dukeleto darbelo: i don't know what is up with the slice opcode. but we should probably add more tests to it
Austin Slice is dead, I think. 20:02
Removed a while back, with just some leftover references to confuse everyone.
(per Chromatic)
dukeleto then someone update the frickin' /topic!
Austin (or was it allison?)
dukeleto i think allison and chromatic both try to confuse us in the appropriate ways ;) 20:03
Austin Good idea, Duke. There's never a committer around when you need one.
Maybe I should say "donuts"
See if one shows up.
chromatic A donut or a committer? 20:04
Austin Works every time.
dukeleto does the slice opcode still exist?
is there something "up" with it?
or can we change the /topic ?
Austin Was it you, chromatic, that pronounced the death knell of the slice opcode?
*death knell = doom 20:05
*(or, pronounced = pealed, your choice.)
chromatic I don't think I did. 20:09
There's no slice opcode, but there is a slice VTABLE entry. 20:10
It's unused. 20:12
slicey slice
dukeleto so our topic is just totally nonsense? 20:13
20:13 theory joined
pmichaud slice has not really worked for at least a couple of years 20:15
chromatic Let's yank the VTABLE then.
Austin So what does one do with a vtable entry that lacks a corresponding opcode?
pmichaud +1
purl 1
Austin Other than call it from C, I guess?
pmichaud imo, slice makes far more sense as a method call than a vtable 20:16
dalek rrot: r42486 | fperrad++ | trunk/runtime/parrot/library/distutils.pir:
[distutils] add get_nqp()
 20:22
whiteknight remove the slice VTABLE. if nobody is using it, it's dead code 20:23
cotto_work slice it out 20:25
dukeleto SLICE OUT, DAWG 20:27
dalek rrot: r42487 | fperrad++ | trunk/runtime/parrot/library/distutils.pir:
[distutils] add options pir_pge_flags & pir_nqp_flags
 20:35
cotto_work dukeleto, don't you have a commit bit? 20:40
chromatic I can slice them out now. 20:42
dukeleto cotto_work: yes, i do. that doesn't mean i know what i am doing. 20:45
whiteknight yo dawn, i herd you liek slice. So I put some slice in yo slice so you can slice while you rip it out 20:46
s/dawn/dawg/ 20:47
japhb is bak, reinforced with: short nap resulting in dream of beating a frat boy with a brick of cheese; breakfast for lunch (mmm, meusli), and chocolate. 20:59
awww, looks like I missed bubaflub. 21:00
particle don't worry, he'll be back when you leave
21:01 bacek joined
dukeleto japhb: bubaflub was my gsoc student this year, i can relay a msg if you need 21:01
japhb msg bubaflub the make failure looks like someone goofed while changing headers around.
purl Message for bubaflub stored.
dukeleto japhb: he is up for hacking on all kinds of stuff 21:02
japhb dukeleto, excellent
msg bubaflub I can fix the configure problems, but I need a tarball of your gl/glu/glut headers 21:03
purl Message for bubaflub stored.
chromatic Realclean necessary. 21:06
japhb chromatic, from a new push, or in relation to the problems that bubaflub posted? 21:07
dalek rrot: r42488 | chromatic++ | trunk (8 files):
[PMC] Removed slice VTABLE and related detritus. All tests pass.
 21:08
rrot: r42489 | chromatic++ | trunk (3 files):
[include] Removed now unnecessary parrot/slice.h header.
chromatic ^
japhb fair enough
darbelo You mean it isn't necessary all the time? 21:09
japhb darbelo, make reconfig FTW
chromatic I rarely realclean.
21:09 mariano__ joined
darbelo FTW? 21:12
purl i think FTW is a unix file-finding library or Fuck The World or for teh win or for the win or FONDLE TEH WINNAR or lame
dukeleto darbelo: for the win! 21:14
21:15 masak joined
dukeleto masak: salutations 21:15
masak good evenin'.
dukeleto masak: what are you hacking on these days? 21:16
japhb dukeleto, our culture
masak dukeleto: just about to wrap up the Web.pm grant. trying to get proto floating again. diddling with November.
oh, and the culture thing. :) 21:17
Tene masak: did you get parrot-mysql working with squerl?
dalek rrot: r42490 | japhb++ | trunk/config/gen/opengl.pm:
[OpenGL] Should improve bubaflub++'s OpenGL configure, but not a complete fix
 21:18
masak Tene: still haven't tried.
Tene masak: you can install it with plumage. ;)
japhb msg bubaflub see parrot r42490 for some minor relief of configure dispepsia
purl Message for bubaflub stored.
dukeleto hmmmmm, configure dispepsia
masak Tene: cool! :) 21:19
japhb Can someone who knows what's been going on with moving headers around fix nopaste.snit.ch/18674 please 21:20
japhb has not been paying attention to any of that. :-/
Tene svn mv include/* /tmp
dalek rrot: r42491 | jkeenan++ | trunk/t/op/bitwise.t:
Applying bitwise_t.patch submitted by bubaflub++ in trac.parrot.org/parrot/ticket/1267.
 21:21
japhb Tene, the problem appears to be in a header several levels of include below anything I directly touch, so likely to have side effects if I go mucking around down there 21:22
dukeleto *always* does a realclean 21:26
Tene japhb: I have no idea what's been going on lately. :) 21:27
21:35 TiMBuS joined
darbelo japhb: adding the missing include file to the src/glut_callbacks.c deps should fix that. 21:36
21:38 bubaflub joined
dukeleto i just love 90 line stack traces, how did you know that was exactly what I wanted? 21:38
chromatic You *were* using Java, a language with an emesis fetish. 21:41
darbelo japhb: The pmc/pmc_context.h header is a generated file, there should be an explicit makefile dep on every file that includes it.
japhb: same with all other pmc_*.h files you #include. 21:42
japhb darbelo, But I'm confused here: why do the direct dependencies of src/glut_callbacks.c have anything to do with it? The file that's breaking is way below the one that I do include, which is just parrot/parrot.h. Or are you saying that parrot/parrot.h needs to be listed as an explicit dependency now?
dukeleto 90 line stack trace + deep multiple inheritance + abstract methods = fail 21:43
japhb darbelo: I only include parrot/parrot.h, and a glut header, nothing else.
darbelo japhb: Ugh. That is some bitter header soup. 21:44
japhb Yeah. 21:45
darbelo The actual 'splode is in context.h, which pokes into the context PMC's internals via that header. 21:46
The confusing part is that opengl is getting built before the PMCs, which is not very same at all. 21:47
japhb agreed
bubaflub i'm hacking on some t/op/*.t stuff
is there a list to see what people are already working on? 21:48
or should i just claim a test and convert it PIR and throw the patch in the ticketing system
darbelo Maybe you should add a dep on some late-built file, to make sure it gets built after the generated headers are there.
japhb bubaflub, purl probably already assaulted you, but ... don't forget to send me a tarball of your GL/GLU/GLUT headers 21:49
bubaflub ah, that's what it was. ok.
the stuff that parrot generates?
21:50 MoC joined
japhb bubaflub, no, I need your system headers. 21:50
bubaflub ah
hmmm
japhb So I can update the master typemap
darbelo japhb: That goes beyond ugly. 21:51
japhb darbelo, can you think of a reliably "late-built" file to depend on?
darbelo, what, the typemap?
chromatic libparrot
purl i heard libparrot was not included in the link for libnci_test.dll
darbelo japhb: what chromatic said.
dalek tracwiki: v5 | chromatic++ | OpcodeTestingHackathon 21:52
tracwiki: added notes about improving op coverage
tracwiki: trac.parrot.org/parrot/wiki/Opcode...ction=diff
japhb ./config/gen/makefiles/root.in:$(LIBGLUTCB_SO): $(LIBPARROT) $(SRC_DIR)/glut_callbacks$(O)
So clearly I just need to move that requirement "earlier"
21:55 theory joined
darbelo japhb: $(SRC_DIR)/glut_callbacks$(O): $(LIBPARROT) 21:55
japhb darbelo, Actually, it may work to just depend on $(GENERAL_H_FILES)
Which includes $(GEN_HEADERS) 21:56
darbelo That can work too.
And is actually less inaccurate
japhb Ok, hacking that fix in 21:57
bubaflub japhb: which headers specifically do you need? 22:02
(from me, that is)
just gl.h / glu.h / glut.h ? 22:04
japhb bubaflub, easiest just to tar up the directories those are in. 22:06
(because every platform includes different files there, and some of them are nothing but "compatibility names" that just include some other private .h, etc.) 22:07
bubaflub japhb: k, i'll tar / bzip em to you the headers from /System/Library/Frameworks/GLUT.framework/Headers and /System/Library/Frameworks/OpenGL.framework/Headers
japhb perfect 22:08
dalek rrot: r42492 | japhb++ | trunk/config/gen/makefiles/root.in:
[OpenGL] make -j users, try this. Makes sure generated H files exist before building libglutcb.
 22:11
bubaflub japhb: i got a bzip for ya... where should i send it? 22:15
darbelo purl: japhb? 22:17
purl japhb is probably Geoffrey Broadwell, mailto:geoff@broadwell.org
Tene purl: tene? 22:18
purl you are Stephen Weeks or a madman or blogs.gurulabs.com/stephen/
darbelo guesses the mail stored in purl's factoid.
Tene purl: darbelo? 22:19
purl darbelo is Daniel Arbelo Arrocha <mailto:arbelo@gmail.com> or Daniel "The Wrecking Ball" Arbelo Arrocha
bubaflub japhb: sent. 22:20
japhb bubaflub, thanks
darbelo purl: bubaflub?
purl bubaflub is, like, mailto:bobjkuo@gmail.com
japhb bubaflub, is your configure/make better (if not perfect) as of r42492?
bubaflub lemme make realclean real quick 22:21
one sec
darbelo tries make realquick
japhb WANT 22:22
darbelo AOL 22:23
bubaflub japhb: Configure works, i'll try make -j now
japhb bubaflub, you should still have this warning in configure: 22:24
In OpenGL header '/System/Library/Frameworks/OpenGL.framework/Headers/CGLIOSurface.h', prototype 'CGLTexImageIOSurface2D', can't handle type 'IOSurfaceRef'; original prototype:
extern CGLError CGLTexImageIOSurface2D(CGLContextObj ctx, GLenum target, GLenum internal_format, GLsizei width, GLsizei height, GLenum format, GLenum type, IOSurfaceRef ioSurface, GLuint plane);
correct?
bubaflub yep 22:25
still there
but -j works again. and there was much rejoicing
darbelo japhb: You *really* wanted to stress NCI when you started OpenGL. 22:26
japhb darbelo, yup, guilty as charged. :-) 22:28
I believe that if you really want a subsystem to be production grade, you need to really push it hard.
darbelo japhb++
bubaflub japhb++ for fixing -j as well. 22:30
japhb Auuggh! I want to shake someone in Cupertino, violently. 22:33
darbelo doen't know where Cupertino is, but aproves of violent shaking. 22:34
bubaflub shaken cupertino syndrome
japhb darbelo, Cupertino is home of '1 infinite loop' 22:35
darbelo I prefer the Induced Combustion Syndrome, for cases such as those.
22:37 PacoLinux joined
japhb bubaflub, try 42493. 22:41
masak Tene: I might get to parrot-mysql/Squerl in the weekend. 22:42
bubaflub japhb++: nice. configure didn't barf at all 22:43
make is fine
japhb wheee
bubaflub i'm hacking on some stuff now, but will make realclean soon
japhb And now I get to figure out how/when someone broke the actual implementation. SIGH.
dalek rrot: r42493 | japhb++ | trunk/config/gen/opengl.pm:
[OpenGL] IOSurfaceRef, you just earned yourself a trip to 'void*', Little Missy.
 22:44
22:44 whiteknight joined
whiteknight good evening, #parrot 22:45
japhb That should be done in a Vincent Price voice, I think.
cotto_work good evening whiteknight
whiteknight hello cotto_work 22:46
whiteknight logged on to his windows vista partition tonight, for the first time in months 22:52
*lots* of things to be updated
bubaflub whiteknight: you should consider updating to windows 7. from what i hear it's much better
as being shot only once is much better than being shot twice 22:53
whiteknight yeah, let me just scrape together the money for the upgrade
bubaflub ah, if you have a .edu email address you can get it for $30
whiteknight nope, haven't had one of those since I graduated
bubaflub bummer.
whiteknight More likely, I'm going to log out of Vista later tonight and never, ever use windows again
:)
japhb whiteknight, perhaps you can get it via the MS open source lab guys, if you tell them you're using it to keep Parrot working on Win 7? 22:54
oh, well, then.
whiteknight I'vebeen having troubles getting my webcam working on Ubuntu (though I think I figured out the problem now), and I have a video conference tonight
so that's the only reason I'm here
darbelo I tried the 7 RTM that was available via the ACM subscription I got from SoC. The installer didn't boot on my computer. 22:55
whiteknight oh, that stinks 22:56
darbelo I've come to think of that as a feature.
whiteknight what computer do you have that's such a problem?
darbelo I tried it on a Asus 4-core AMD box IIRC, nothing nonstadard about it. 22:57
nopaste "japhb" at 76.191.190.8 pasted "void -> void NCI call broken by PCC? How do I fix this?" (20 lines) at nopaste.snit.ch/18679
whiteknight urg 22:58
darbelo It actually booted, to be fair, displayed a nice wavy background and let me move the mouse pointer.
But want more from an OS installer than a mouse pointer and a pastel background. 22:59
Mybe I'm picky, but I want it to, at least, install the OS.
whiteknight see? you're just being needy 23:00
japhb That's only what you *thought* you wanted.
whiteknight like all those people who want parrot not to segfault. That's just an absurd, unreasonable demand
japhb Anyone got an idea on my nopaste above? chromatic, allison, bacek, ...?
whiteknight japhb: looking at it. Can you get a backtrace? 23:01
23:01 plobsing joined
darbelo Actually, my favourite installer has neither a pastel background nor a mouse pointer. 23:01
whiteknight I suspect an NCI signature is being fed into the PCC mechanism
darbelo is a console freak.
japhb whiteknight, yeah, I'd believe that analysis
I'm a little rusty, but I'll give the bt a try
whiteknight w00t. 23:02
23:02 jan joined 23:03 kiwichris joined
whiteknight gdb --args parrot examples/opengl/triangle.pir 23:03
nopaste "japhb" at 76.191.190.8 pasted "BT for whiteknight++" (23 lines) at nopaste.snit.ch/18680 23:08
japhb Oooh, looks like a callback fail! 23:09
bubaflub just a heads up, t/codingstd/pod_syntax.t is failing 23:10
plobsing you are calling Parrot_pcc_invoke_sub_from_c_args wrong 23:11
whiteknight oh great, a callback 23:13
messy backtrace
plobsing the bug is at src/glut_callbacks.c:251
whiteknight plobsing++
plobsing s/"v"/"->"/
japhb plobsing, to be fair, I don't think it was me. svn blame shows that the section of config/gen/opengl.pm that generates src/glut_callbacks.c hasn't been touched by me in like 15 thousand revs.
plobsing should fix
sorry for first person. unintended accusation. 23:14
dalek TT #1268 created by bubaflub++: [PATCH] convert t/op/time.t to PIR 23:15
whiteknight bubaflub++
japhb plobsing, ... but I'm happy to try to fix it, if someone can point me at the docs mapping the old NCI signatures to the new ones.
whiteknight getting started on the hackathon early
bubaflub haha, it's the weekend for me
i work part time 9 - 2 on Fridays
plobsing japhb: there are no new nci signatures yet
bubaflub my girlfriend works 5 p.m. to 3 a.m.
so i'm good to hack
whiteknight japhb: old sigs are <return(1char)><args>
new sigs are <args>-><returns>
plobsing japhb: but that call takes a pcc signature
japhb whiteknight, plobsing: so ... is the call being done with the wrong function, or is the signature in need of fix? And if the latter, into what format, (if NCI sigs != PCC sigs) 23:16
whiteknight the signature string is wrong 23:17
japhb OK.
whiteknight is "v", should be "->"
japhb whiteknight, so no longer a need to list a return value even if there isn't one? Good!
japhb braces for epic hackery
whiteknight EPIC! 23:18
whiteknight has basically no hacking time tonight 23:22
23:25 ascent joined
japhb Sigh, shell quoting is going to ruin the joke in my commit message. :-( 23:34
plobsing you've fixed it already? 23:35
thats fast
cotto_work That's what you get for committing with -m 23:36
dalek rrot: r42494 | japhb++ | trunk/config/gen/opengl.pm:
[OpenGL] Fixing one line of code: . Knowing which line to fix: ,999.
 23:37
TT #1269 created by bubaflub++: [PATCH] convert t/op/comp.t to PIR 23:47
japhb Sigh, shell quoting is going to ruin the joke in my commit message. :-( 23:49
purl i already had it that way, japhb.
japhb BAH
Sigh, it looks like more than just order of return and arguments changed with new sigs.
Where is the doc? Searching for '->' and '-E<gt>' in docs/ finds nothing relevant to PCC call signatures. 23:50
plobsing japhb: tools/build/nativecall.pl converts NCI sigs to PCC sigs
that might be a useful resource 23:51
japhb plobsing, thank you.
But there should be a doc.
If something humongous like PCC landed *without documentation*, I will be miffed. MIFFED, I SAY.
darbelo japhb: In theory, as we have not passed a deprecation boundary. That should still work. 23:52
japhb darbelo, "that"?
darbelo Old style signatures. 23:53
japhb (honest question ... I'm not sure which antecedent you are referring to.)
Oh. I expect they wouldn't work on the _pcc_ routines, which is what someone changed all the glut_callbacks code to use.
I'm happy to try to get it working using the _pcc_ calls, but since I didn't make the change, I'd like a doc to how to properly fix it. :-) 23:54
The fix I committed above works for simple cases, and gets the spinning triangle example working, but nothing advanced.
darbelo We are taking a 2x slowdown in acount of PCC 'compatibility' code. It bloody better be *compatible* dammit.
japhb darbelo, if the _pcc_ routines are supposed to accept old signatures ... they don't. Pretty much definitively. ;-) 23:55
darbelo Hmm. I might be thinking of a different set of functions. 23:56
23:58 bubaflub left
darbelo japhb: Does trac.parrot.org/parrot/wiki/PCCMigrationNotes refer to your particular case at all? 23:59

Powered by Cro and the Raku® Programming Language.

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

Thank you!