Parrot 2.10.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Onward and upward with Google Code-In | Please test rakudo with bleeding edge parrot!
Set by moderator on 11 December 2010.
dalek rrot/embed_api2: c31e7a5 | Whiteknight++ | src/global_setup.c:
make sure we initialize the library paths array before the config hash. The library paths array has two parts: part that is static and part that comes from the config hash. We want to make sure we setup the static parts first
 00:05
rrot/embed_api2: 8b915b2 | Whiteknight++ | / (5 files):
Merge branch 'embed_api2' of github.com:parrot/parrot into embed_api2
whiteknight Coke: ping 00:07
dalek rrot/bulgarian_readme: 51f05a3 | dukeleto++ | README.български:
[doc] Bulgarian translation of our README, from a GCI student
 00:13
kid51 dukeleto: ping 00:14
dukeleto My stance on translations is that we can add bad/unedited translations in branches and get them improved.
kid51: pong
kid51 Yes. Could we get luben to look at that? (I have some coworkers who might, as well.) 00:15
dukeleto kid51: delegate as you see fit, just not to me. 00:17
dukeleto is on vacation
our dynops have horrific test coverage 00:18
Coke whiteknight: pong.
updating... 00:19
whiteknight Coke: I've pushed those changes. should reduce some warnings
although looking back at your make log, I think there are more warnings to fix still
Those warnings about interp getting clobbered aren't fixed
Coke I'll post a fresh log in about 4m. 00:20
dalek rrot: f123b30 | jkeenan++ | / (7 files):
Complete work on trac.parrot.org/parrot/ticket/532. Change
rrot: 5b8fb37 | jkeenan++ | / (4 files):
Merge branch 'master' of git@github.com:parrot/parrot
plobsing dukeleto: dynop coverage is poor because of our bias against using anything not deemed "core" in tests 00:21
dukeleto plobsing: no. It is poor because we are lazy. 00:22
plobsing: no other reason
Laziness isn't necessarily a bad thing, either. 00:23
nopaste "coke" at 192.168.1.3 pasted "latest build failure for whiteknight++" (1311 lines) at nopaste.snit.ch/27055
whiteknight blarg
I'm fixing that warning with clobbered interp first, I'll fix that failure next 00:26
Actually, looking at that log I thought I had already fixed that particular problem 00:31
so that's weird
kid51 tests embed_api2 on darwin/ppc (10.4) 00:33
whiteknight kid51: I have some fixes coming soon
kid51: what compiler do you use there, gcc? 00:34
kid51 gcc as compiler; g++ for link, etc.
Same routine which Coke taught me lo these many years ago! 00:35
echo "MACOSX_DEPLOYMENT_TARGET is $MACOSX_DEPLOYMENT_TARGET" 00:36
CC="/usr/bin/gcc"
CX="/usr/bin/g++"
--ld="$CX" \\
--configure_trace \\
$@
better:
CC="/usr/bin/gcc";CX="/usr/bin/g++";/usr/local/bin/perl Configure.pl --cc="$CC" --cxx="$CX" --link="$CX" --ld="$CX"
00:38 kid51 is now known as kid51_at_dinner
dalek rrot/embed_api2: e38be63 | Whiteknight++ | src/embed/embed_private.h:
attempted fix for a series of warnings Coke++ is seeing
 00:39
rrot/embed_api2: 8dac9d1 | Whiteknight++ | / (2 files):
fix function which defines a return value but never uses return
rrot/embed_api2: 4fdb89b | Whiteknight++ | config/gen/makefiles/root.in:
fix makefile to properly build installable parrot
TT #532 closed by jkeenan++: Finish headerizer refactor
TT #532: trac.parrot.org/parrot/ticket/532
whiteknight okay, fixes are here. Coke, kid51: if you're able to test these I would appreciate it
updating embed_api2 to trunk is going to be messy. I did a mock run earlier today 00:45
I want to make sure things are working well and I fix all existing problems first
00:59 fbrito joined
whiteknight okay, I'm signing off for thenight. Thanks for the testing 01:26
01:26 whiteknight left 01:28 dmalcolm left
kid51_at_dinner Hmm, those latest commits came after I started my test runs; will have to reconfigure, etc. 01:44
01:44 kid51_at_dinner is now known as kid51
fbrito off-topic: are you guys aware of the meteor shower that is going to happen in the next few hours? 01:56
kid51 sky is overcast here, so no meteors for kid51 02:11
plobsing same here. cloud and snow obstruct my view. 02:18
fbrito NASA is talking about 100 meteors/hour 02:25
02:25 mikehh left
Coke snow-- 02:25
plobsing Coke: when you deal with heat and sun as poorly as I do, you tend to accept the other seasons 02:28
can't hate the weather all the time
Coke oh, I had a chance to live in TX, but moved back to upstate NY. I like snow. I'm just mad I'm going to miss the meteors. 02:30
msg whiteknight - build worked. 02:32
kid51 Coke: try 'make testr' 02:33
nopaste "kid51" at 192.168.1.3 pasted "embed_api2 branch: 'make testr' errors, analyzed" (140 lines) at nopaste.snit.ch/27058 02:36
02:37 mikehh joined
mikehh kid51: just replied to youe email, as I have said before, I think we need to keep the requirement that all functions need documentation 02:39
Coke t/tools/pbc_dump.t fails for me. 02:41
kid51 Curiously, that passes for me: $ perl t/harness -r t/tools/pbc_dump.t 02:43
t/tools/pbc_dump.t .. ok
All tests successful.
Coke kid51: and yes, tons of failures in testr. 02:44
kid51 But the kind of errors in make testr is fairly consistent, as apparent from my last paste. 02:58
Well, maybe whiteknight will figure something out.
dukeleto: ping 03:17
kid51 must sleep 03:20
03:20 kid51 left 04:51 Matt221 left
sorear so nobody knows anything about the PIR LALR(1) parser generator? bummer 05:03
05:16 fbrito left 05:18 fbrito joined 05:29 fbrito left, rurban_ joined 05:30 fbrito joined 05:32 rurban left, rurban_ is now known as rurban 05:35 fbrito left, fbrito joined 05:43 bluescreen left 06:09 eternaleye_ joined, eternaleye left 06:29 fbrito left 07:27 wagle left 07:40 rfw left, rfw joined 08:01 theory left 08:07 wagle joined 08:37 silug left 08:45 fperrad joined 08:56 contingencyplan left 09:10 cosimo joined 09:25 rfw left 10:19 bluescreen joined 12:15 bluescreen_ joined 12:42 lucian joined 12:54 mtk joined 12:57 fbrito joined
fbrito hi guys 13:00
tadzik hi 13:01
fbrito my "make cover" is generating something like this: tapir2.ro.vutbr.cz/cover/cover-resu...2/c_cover/ but only with 4 or 5 files... what should I do? did I forget to install something?
tadzik oh, I have a bug in mk_language_shell.pl – wklej.org/id/437820/ It's using the dot as a languagu name 13:06
13:22 bluescreen_ left 13:29 rurban_ joined 13:32 rurban left, rurban_ is now known as rurban 13:38 bluescreen_ joined 13:57 whiteknight joined
whiteknight good morning #parrot 13:58
still no aloha, apparently 13:59
atrodo it's a shame, really 14:00
dalek rrot/embed_api2: 09cde93 | Whiteknight++ | / (8 files):
Merge branch 'embed_api2' of github.com:parrot/parrot into embed_api2
 14:01
14:04 sECuRE left
Coke has anyone msg'd bacek about it? 14:07
14:08 sECuRE joined
whiteknight aloha msg bacek aloha is broken! 14:09
Coke ... I'll tweet him. 14:10
whiteknight :)
fbrito whiteknight: good morning :D 14:13
ah, I have a question... my "make cover" is generating something like this: tapir2.ro.vutbr.cz/cover/cover-resu...2/c_cover/ but only with 4 or 5 files... what should I do? did I forget to install something?
tadzik hi whiteknight
whiteknight good morning fbrito, tadzik 14:17
fbrito: I think make cover may be broken or something 14:18
at least, it was a few weeks ago and I didn't fix it
somebody else may have
those coverage numbers look good to me. What do you think is missing? 14:19
fbrito those are fine, but when I run it locally, it has only 5 files 14:20
whiteknight oh, okay. I see
yeah, so it's probably still broken
but wait, another student was doing some coverage-based test tasks 14:21
we probably need to wait for cotto to get back so we can figure out what is going on
fbrito ah, ok 14:26
hm, I think I found something wrong with acosh of complex numbers with negative image 14:30
not ok 529 - acosh of 0-1i: # Have: -0.881374+1.570796i 14:31
but www.wolframalpha.com/input/?i=acosh(0-1i) has a result with opposite signs 14:32
whiteknight weird
is that a bug in Parrot, or a bug in the acosh library function? 14:33
fbrito hm 14:38
whiteknight because I know that there are several problems relating to those trancendental functions on various platforms. I don't think they are well-tested 14:41
fbrito whiteknight: and there are also lots of functions returning -0.0 14:48
whiteknight fbrito: negative zero is okay in some cases
in fact, we have some tests that expect negative zero
tadzik docs.parrot.org/parrot/latest/html/...4.pod.html -- where are those in Parrot tree? 14:52
whiteknight tadzik: Not in the Parrot tree anymore
tadzik whiteknight: where are they then?
whiteknight looks like they are at github.com/ekiru/squaak-tutorial 14:53
although I suspect they should also be mirrored under the parrot organization too
tadzik „A repo I used to update the Squaak tutorial. This repo is now out-of-date.”
neh, weird
whiteknight okay, I've forked it here: github.com/parrot/squaak-tutorial 14:54
so it probably needs to be brought back up to date
tadzik I'll be giving a Parrot talk on friday on a local LUG meeting and I'd like to show some compiler fun, but I have to learn it first :) 14:56
whiteknight oh, that's fun. Where's the LUG? 14:57
tadzik on my Uni. It's not any officiacl LUG actually
Warsaw, Poland, if that's what you mean 14:58
whiteknight oh, okay
tadzik ...too far? :) 14:59
whiteknight much too far
atrodo whiteknight> What happened to your private jet? ;) 15:04
15:04 bluescreen left
whiteknight tadzik: I knew a person a long time ago who went to University in Warsaw 15:07
atrodo: my imaginary jet is in the imaginary shop for repairs 15:09
it's reality is broken
atrodo i hate it when reality is broken
tadzik whiteknight: remember the name?
whiteknight tadzik: only the screenname. Derbeth 15:10
tadzik btw, I created the new language with mk_language_shell. Then I changed rule statementlist { [ <statement> | <?> ] ** ';' } to rule statementlist { [ <statement> | <?> ] ** \\n }, but the code 'print("foo")\\nprint("bar")' does not even parse. Am I missing something? 15:11
dalek rrot/embed_api2: d4a7f12 | Whiteknight++ | compilers/imcc/debug.c:
fix most testr failures. IMCC, and anybody else who isn't libparrot, should never call Parrot_x_exit directly. Use Parrot_x_jump_out_error instead
 15:12
rrot/embed_api2: 2e9b4e6 | Whiteknight++ | t/op/exit.t:
the test that I un-todo'd turns out to be bad and needs to be removed. Failures in loading/running pbc shouldn't exit with status 0 by default. that's stupid. Failures in program execution should signal that to the OS. I've removed the test entirely instead of re-TODOing it
whiteknight tadzik: I have no idea. That's probably a question for some NQP folks
pmichaud, jnthn, or moritz would probably know immediately
tadzik nqp folks around?
probably
whiteknight they would say that "LOL ur parse is t3h failz" 15:13
tadzik hmm, I can check the LOLCODE code
whiteknight Coke, kid51: ping
does LOLCODE work now? 15:14
tadzik yep
fails some tests, but works
it uses some old nqp too
whiteknight oh 15:16
tadzik hm? 15:18
btw, I'm actually on WUT (Warsaw University of Technology), not University of Warsaw :) 15:19
Coke whiteknight: yes? 15:22
retest embed_api2? 15:23
when I run "make cover", I get a ton of files. 15:24
$ ack -fa cover_db/ | wc -l 313
tadzik out 15:25
15:26 cosimo left
Coke whiteknight: it still builds. 15:28
nopaste "coke" at 192.168.1.3 pasted "embed_api2 for whiteknight++" (1979 lines) at nopaste.snit.ch/27078 15:30
Coke latest warnings posted.
cotto will be back to normal in a few hours.
Coke thanks for the warning. 15:31
whiteknight Coke: if you get a chance can you make testr? 15:33
Coke passes.
running fulltest.
dukeleto ~~ 15:34
whiteknight oh, awesome. Coke++
I'm sure fulltest will freak. There are lots of codetests that are going to die painfully
I haven't gotten to that point yet
Coke "make library_tests" fails. 15:37
t/tools/pbc_dump.t fails all tests.
(and ISTR there are warnings on that build.)
15:40 Patterner left
dukeleto whiteknight: have you seen this submitted work? socghop.appspot.com/gci/task/show/g...9130699951 15:40
15:40 bluescreen_ left 15:41 smash joined
smash hello everyone 15:41
Coke hio 15:42
15:43 sECuRE left
whiteknight hello smash 15:43
15:48 sECuRE joined 15:50 Psyche^ joined, Psyche^ is now known as Patterner 15:52 bluescreen_ joined, dmalcolm joined 15:53 theory joined
whiteknight ha, I just read the "release notes" from the attack on Gawker, and they are hilarious. 16:00
it's just a comedy of errors on the part of the Gawker admins. At every step they used the worst possible security, and then continued the badmouth the people who were attacking them 16:01
the hackers kept getting angrier, and kept doing worse and worse things
went into the DB and dropped all tables, sudo rm -rf, etc 16:02
atrodo reminds himself not to be a douche 16:05
16:09 bluescreen_ left, bluescreen_ joined
whiteknight no, this is nothing to capitalize on. It's just funny that the security at Gawker would be so low, and they would continue to taunt the hackers who were logged into their system at the time 16:09
a little bit of humilty might have saved them from the worst-case scenario 16:10
atrodo Exactly
whiteknight a little bit of good security definitely would have
atrodo Never hurts. But being aware of what to do in the event of security breach is just as important 16:11
whiteknight right
16:11 bluescreen joined
whiteknight not using the same password on all sites you visit is a good thing. Using OpenID with a provider you trust is even better 16:11
atrodo Aye. Looks like I'll be seriously considering doing exactly that 16:12
whiteknight I share passwords on some sites, but I break it up. So bank websites use a different password from my parrot stuff, which uses a different password from my email address, which uses a different password from unimportant crap 16:17
and that's probably still not good enough 16:18
dalek rrot: 07df1ca | NotFound++ | / (3 files):
add the throws_type function to Test;More and use it in a new test for Namespace
 16:19
fbrito whiteknight: so, about my GCI task (increase code coverage of Complex PMC): what should I do with the asech "bugs" that I found? create a bug ticket and mark the test as TODO? and since I couldn't run "make cover", I don't know if I increased the coverage more than 10% 16:22
whiteknight fbrito: what system are you on? Coke said he was able to make cover this morning with no problems 16:27
fbrito: for your question, create tickets and mark the failing tests TODO 16:29
fbrito I will try "make cover" again... going to lunch right now. be right back 16:31
whiteknight okay
16:37 contingencyplan joined
NotFound Exception type 0 is EXCEPTION_BAD_BUFFER_SIZE, but a lot of places use 0 as "I don't know or don't care". So I propose to set type 0 as EXCEPTION_UNKNOWN_TYPE 17:14
Or maybe just move EXCEPTION_INVALID_OPERATION to 0 17:16
dalek rrot: 07d9124 | NotFound++ | src/pmc/namespace.pmc:
use appropiate exception types in Namespace.export_to method
 17:17
bluescreen_ whiteknight: is make library_tests failing for you 17:19
its passing in my computer
dalek rrot: 33a5288 | NotFound++ | t/pmc/namespace.t:
more tests for Namespace methods
17:20 davidfetter joined
cotto_work2 ~~ 17:21
17:21 cotto_work2 is now known as cotto_work
whiteknight bluescreen: I haven't tried it 17:28
bluescreen_ ok... the only thing failing in the fulltests are the coding stds
dalek TT #1891 created by fbrito++: Problems with acosh, asech, atanh and acoth in Complex PMC 17:29
TT #1891: trac.parrot.org/parrot/ticket/1891
rrot: b9e4e73 | NotFound++ | runtime/parrot/library/Test/More.pir:
fix typos from 07df1ca
mikehh hmmnn, I commited the same changes locally, how do I revert? - in my local repo 17:36
or is it easier just to clone it again? 17:38
NotFound mikehh: create a branch, delete master branch, create branch tracking remote master, delete the other branch 17:40
Maybe is easier to clone again X-) 17:41
mikehh when I did a git pull it says - Falling back to patching base and 3-way merge..., No changes -- Patch already applied. 17:42
NotFound Looks like it fixed himself 17:43
mikehh anyway I shall probably just clone again, just in case
moritz you can always do a git reset --hard origin/master
whiteknight fbrito: there is a comment in the source of src/pmc/complex.pmc that says all those inverse hyperbolics produce incorrect results 17:46
fbrito: as part of your work, can you link to your new ticket from that source file so people can keep track? 17:47
cotto_work whiteknight: in general it's fine to use anything I post online as you see fit. I posted the Lorito braindump notes so that the process wouldn't be stuck between then and when I had time to do a proper writeup. 17:51
plobsing the internals of complex.pmc are just nasty. why are we re-implementing (using PCC methods) things most libm implementations provide optimized versions of? 17:52
Coke "why do we even have a complex PMC?"
plobsing yeah. everyone knows we should use Quaternion PMCs in stead. 17:53
cotto_work Coke: +1
plobsing +1 to deprecate + remove. if anyone cares about complex, they will implement it and stand a better chance of doing it right (because they actually care) 18:03
PerlJam Coke: complex numbers are a common requirement of most language run time environments, therefore, if we are to be successful as a language platform, we *must* have it ;-) 18:04
Coke PerlJam: is rakudo using parrot's Complex PMC?
PerlJam plobsing: +1
Coke: no idea. 18:05
atrodo cotto_work> Any ideas on the extended version of the braindump?
cotto_work atrodo: I'll have time to do it tonight
atrodo Exciting! 18:06
PerlJam Coke: looks like "no"
cotto_work atrodo: do you have a blog? 18:07
atrodo cotto_work> I should. I have a site, but no content 18:08
cotto_work Go blog then!
atrodo tempting...
Oh, right. I stopped working on the blog because I couldn't get the logo working like i wanted... 18:10
(lame)
Coke atrodo: here you go: www.google.com/images?rlz=1C1CHMP_e...mp;bih=879 18:13
atrodo :D Actually, I have a logo, which was the disapoointing part. I'm too married to the one I made 18:14
Coke humbly suggests to blog anyway with the logo you have, and fix it up later if it bugs you.
atrodo That's what I'm about to do 18:16
cotto_work atrodo++ 18:17
anyone seen bacek recently?
Coke atrod++ 18:18
cotto_work: I tweeted him about aloha.
cotto_work Coke: thanks
Coke someone else can try email. ;)
dalek TT #1892 created by plobsing++: [DEPRECATED] Complex PMC
TT #1892: trac.parrot.org/parrot/ticket/1892
rrot: aaf01d1 | plobsing++ | DEPRECATED.pod:
add Complex PMC deprecation
rrot: c3f2c98 | plobsing++ | runtime/parrot/library/Test/More.pir:
Merge branch 'master' of github.com:parrot/parrot
cotto_work whiteknight: as I think more about self-hosted languages, it's starting to make sense to me that we shift our focus from interoperable languages built with npq-rx to interoperable, mostly self-hosted language implementations sharing a common bytecode and MOP.. 18:21
NotFound Just "language implementations sharing a common bytecode and MOP" isn't enough? 18:22
plobsing \\o/ parrot as more than just a VM to run nqp-rx
dalek rrot/embed_api2: 9f2ff29 | Coke++ | / (2 files):
Add missing include to "pbc_dump"
 18:23
cotto_work NotFound: for me there's no point to Parrot if they're not interoperable.
NotFound cotto_work: I mean, we shouldn't care about how the compilers are implemented.
Coke cringes at the possibility of having to rewrite tcl YA. 18:24
cotto_work Coke: there's no reason you couldn't continue to use nqp-rx. I just don't want it to be the only viable option.
and if Rakudo's using it, it's not going anywhere.
Coke cotto_work: Be nice to have one /complete/, viable option first, but ok. 18:25
NotFound In fact, I didn't care, I wrote winxed the way I liked without following any recomendation.
18:25 muixirt joined
Coke NotFound++ 18:25
plobsing winxed or Ωη;)XD (when finished) are viable alternatives to nqp-rx 18:26
cotto_work "Ωη;)XD"? 18:27
Coke Hey, before putting stuff in deprecated, can we get some kind of vote or something?
whiteknight PLA relies on the Complex PMC, but there's no reason why I couldn't either merge Complex into that project, or create a new project for it and use that as a dependency
NotFound That's not entirely true, I followed recommendations for language designers from Stroustrup's book "The Design And Evolution Of C++"
atrodo And draak. Possibly
whiteknight that ticket should probably be marked RFC first. Build consensus, then deprecate
we can vote on it today at #ps
cotto_work Coke and whiteknight, yes
plobsing cotto_work: ohm-eta-wink-kzd is my port of OMeta to winxed
NotFound plobsing: there is some prize for the project with the most bizarre name? 18:28
plobsing whiteknight: whats the point of an RFC to deprecate ticket on top of a deprecation ticket? all tickets inherently carry the option to reject.
whiteknight plobsing: because "deprecation" tickets have traditionally been used to track an approved deprecation. Using it for something else will cause confusion at least 18:29
plobsing NotFound: OMeta/Winxed sounded too boring, it's 2010, and I'd like to see what breaks
github, for example, rejected the name (probably in an attempt to avoid SQL-injection) 18:30
NotFound resist to link little Bobbie Tables again
cotto_work NotFound: be strong 18:31
NotFound <strong>
plobsing it's 2010. if you aren't using parameterized queries, you're doing it WRONG
atrodo xkcd.com/327/ 18:32
NotFound </strong>
atrodo not so strong
whiteknight Coke++
plobsing whiteknight: I've always interpreted a deprecation ticket as an "option to remove". Nothing says that we need to exercise that option, nothing says we cannot relinquish that option. 18:34
18:34 mtk left
cotto_work That uncertainty isn't necessarily a good thing. 18:35
plobsing it is inherent in volunteer based projects
the alternatives is to apply the same criteria as our roadmap. that boarders on micro management. 18:36
Coke I don't think we need a rigorous process, but not everyone follows the ticket list. bringing it up in #ps is fine. 18:37
NotFound Uncertainity principle in open projects: if you know that it moves, you don't know its direction.
Coke should git ignore *.gcno, *.gcda, and *.gcov ?
(#ps) even ex-post-facto. 18:38
cotto_work Coke: yes. Those are artifacts from make cover. 18:39
Coke k. they are unignored ATM. 18:40
(at least in embed_api2)
18:41 mtk joined 18:42 ligne left
NotFound plobsing: what's the github project name? 18:44
Found: github.com/plobsing/ohm-eta-wink-kzd 18:45
plobsing not much to see yet. most of the action is taking place under src/bootstrap 18:46
the plan is to bootstrap from OMeta/JS using node.js
both js and winxed versions of the bootstrap compilers come from the same source (with cpp bridging the slight differences between the languages) 18:47
NotFound Amazing 18:48
plobsing the end result will be an overlay of OMeta grammars in winxed 18:49
github.com/plobsing/ohm-eta-wink-k...4%A6%CE%B7 is what I'm aiming for
whiteknight plobsing: that will be pretty awesome methinks 18:52
plobsing that's the dream. for now I'm happy stage 0 compiles null toplevels fine. 18:54
atrodo: do you have any draak samples? 18:56
atrodo github.com/atrodo/draak/blob/master/gmr/pl0.gmr 18:57
Sadly, I was working on getting code generation to work when I switched over to working on Lorito
s/to work/started/
plobsing atrodo: what parsing algo is it based on? LR? rec-descent? 19:00
atrodo LL, recursive descent 19:01
It was simpler to write and maintain, and since I parse the grammar file everytime, i suspect going to LALR would result in a speed loss. 19:03
plobsing why do you parse the grammar file every time? seems like that's your problem. 19:04
atrodo Because that was the goal. Have one file that describes the grammar and code generation that is used as the source of compilation. 19:06
plobsing so you're writing a compiler interpreter in stead of a compiler compiler? 19:07
atrodo More or less, yes
plobsing cool 19:09
19:10 hercynium joined
NotFound Sounds a bit like a Knuth Parse Machine. 19:13
19:25 rfw joined
Coke t/manifest/02-regenerate_file.t also fails on embed_api2 19:32
ah, and rightly so. 19:33
whiteknight ?
19:38 ok22 left, ok22 joined 19:42 fbrito left
whiteknight I think we could definitely have embed_api2 mergable shortly after 2.11 19:48
and I look forward to that
because shortly thereafter I want to open deprecations for several things 19:49
and I want to rearrange/clean a lot of code
19:56 chromatic joined
Coke (rightly so) because you added files but didn't regen the manifest. 20:01
whiteknight ah, okay 20:02
cotto_work #ps in 28 20:03
atrodo NotFound> Do you have a linky for "Knuth Parse Machine"? 20:11
20:17 tcurtis joined
NotFound atrodo: no, only a 1988 Spanish book. I think it's from a paper about LR parsers. 20:18
It may be "Top down syntax analysis" but I'm not sure, the book has lot of erratas in the biblio references, 20:26
sorear NotFound: ISTR reading something about a Parrot LALR engine which emitted PIR. Do *you* know what I'm thinking of?
20:26 fbrito joined
chromatic Was that the one Allison wrote for Pynie? 20:27
NotFound sorear: looks it's a common subject these days. 20:28
plobsing chromatic: didn't know allison wrote a LALR engine for parrot. I know she wrote spix, a PEG engine. 20:31
chromatic I can't keep track of anything but the difference between LL and LR anyhow. 20:32
fperrad seen kapace 20:33
cotto_work fperrad: no bot
plobsing chromatic: PEG is just a pretty syntax over recursive descent (possibly with some smarts, eg: about cacheing) 20:34
chromatic Somehow I'd never realized it was rec descent, but I also hate parsing. 20:35
fperrad seen kapace_ 20:36
atrodo fperrad> EALOHASTILLMIA 20:37
fperrad ping kapace_
sorear EBACEKSTILLMIA 20:39
Coke NotFound: want privs to the google calendar? 20:43
chromatic What happened to the topic? 20:44
20:44 kid51 joined
sorear It's still there 20:44
Coke 15:45 -!- No topic set for #parrot 20:45
sorear my IRC server begs to differ.
moderator Parrot 2.10.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Onward and upward with Google Code-In | Please test rakudo with bleeding edge parrot! 20:45
sorear I guess some subset of servers lost the topic 20:45
#perl just had the same problem 20:46
NotFound Coke: I've never used it other than looking at already made calendars.
moderator Parrot 2.10.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Onward and upward with Google Code-In | Please test rakudo with bleeding edge parrot! | Remove deprecations | close tickets | merge html_cleanup and embed_api2 20:47
cotto_work dukeleto: ping 20:51
atrodo hmmm, it appears that I've actually got two blog posts here. 20:52
20:53 allison joined
kid51 whiteknight: embed_api2 PASS make testr on 2 different linux boxes; will try darwin at $home tonight 20:57
whiteknight kid51++ Thanks! 20:58
now I need codestd fixes, documentation, and lots of testing
21:01 lucian left 21:03 haypo joined
fbrito tadzik: I need your feedback on the GCI task that you are mentoring: www.google-melange.com/gci/task/sho...9193205977 21:04
tadzik: when you have some time, please take a look
Coke twitter.com/#!/bacek/status/14787890802335745 - how to grab aloha. 21:05
whiteknight yes, we need aloha back 21:06
21:07 bluescreen_ left
cotto_work That second to last tweet from bacek doesn't look good. 21:07
atrodo no, not really 21:08
21:08 lucian joined
whiteknight I think what we need to do is take some pressure off bacek, so he can focus on other things if he needs to 21:16
we need to get Aloha going ourselves, and we need to kick the GC into high gear again
cotto_work feather seems like it would be a good place to host aloha 21:17
whiteknight I wonder what we would need to do for OSU to hook us up with another VM for tools like that?
Coke parrotvm is also an option.
we have a vm. just use that one.
21:19 rfw left
sorear why does parrotvm exist? 21:19
I talked with Juerd earlier, feather is very explicitly for Parrot too
21:20 rfw joined
whiteknight we can use feather too 21:20
Parrot relies on many tools and wants more. I would like to explore ways to expand our capacity
NotFound Idea for GSoC: create an irc bot engine for parrot. 21:21
whiteknight Notfound: yes, that is a good idea
an event-based IO engine is better, and an IRC library on top of that 21:22
NotFound Yeah
Coke parrotvm exists for the same reason parrot.org exists.
NotFound Coke: For fun?
Coke (as oppposed to the perl.org hosted parrotcode.org)
because someone thought it was a good idea for parrot to not be hosted on perl.org and related services. 21:23
whiteknight it is a good idea
and OSU has been a wonderful host
21:25 fbrito left
Coke I question the cost-benefit, but it's sunk. 21:26
21:28 whiteknight left
cotto_work Where's dukeleto been? 21:29
21:29 rurban_ joined
chromatic Vacation. 21:30
cotto_work Ah. Good for him.
21:32 rurban left, rurban_ is now known as rurban, kid51 left
atrodo is aloha's data available? 21:38
mikehh forgot to report - All tests PASS (pre/post-config, make corevm/make coretest, smoke (#1621) fulltest) at 8d73990 - Kubuntu 10.10 amd64 (g++-4.5 with --optimize)
presumably on bacek's server (in postgres I think) 21:39
we shoul;d also update the karma from moritz's logs 21:40
bah again, this new keyboard keeps moving keys slightly from what I is used to :-} 21:41
moritz mikehh: I already have a script that extracts karma from the logs; if you remind me tomorrow I can run it and mail you the results 21:48
21:50 mtk left 21:51 tcurtis left
cotto_work moritz: great 21:54
moritz: you could just nopaste the data or stick them into a gist 21:56
NotFound My experiments with JS-alike prototype based method look up gave a dead-end. How and when to invalidate method cache? 22:01
Silly idea: add a flag to Class to don't cache method lookup. 22:03
chromatic Keep pointers to children (ugh) or re-use GC mark information to implement some sort of invalidation flow or use a global counter of potential invalidations and invalidate all caches when their numbers do not match that generation.
Try to guess which one Perl 5 uses!
NotFound The (ugh) one?
atrodo None of the above? Pure magic?
chromatic The global variable. 22:04
sorear Javascript needs to maybe invalidate method caches every time you poke a value into an object 22:05
NotFound sorear: yes, that's even worse than the prototype problem.
And even if you don't poke it. Just if at least two objects have different versions of one method. 22:06
cotto_work chromatic: is there any reason not to make Lorito bytecode immutable? 22:08
NotFound Overly expensive solution that may work without touching parrot: for each instance generate its own class.
chromatic Immutable seems right. 22:09
NotFound I think I've heard some JS engine uses a less simplistic and optimized version of that. 22:10
cotto_work then that sounds like it'd fit into the interp rather than the context 22:11
NotFound But for a short experiment is too much.
Coke (global counter of potential invalidations) so does tcl.
cotto_work well, bytecode segments. I'd think that we'll want to be able to load (and possibly unload) bytecode segments, though not modify them 22:13
sorear supporting self-modifying code is insane. +1 to not doing it 22:14
anyone who wants to create code at runtime should announce their intenion, with a "create new bytecode segment" call (you might spell it flush_icache) 22:15
NotFound So I must discard my old idea of an INTERCAL compiler that implements ABSTAIN by self-modifying code ;) 22:16
cotto_work What are the issues with allowing bytecode segments to be unloaded at runtime, given that we can easily figure out where all contexts' PCs are pointing?
NotFound If we don't allow discarding segments we'll never solve memory leaks in 'eval' 22:17
sorear IMO we shouldn't provide "discard segment" as an op 22:18
Segments should just be ordinary GCable objects
NotFound Aye
sorear If all the subs that point to a segment are dead, and no contexts are referencing it - bye!
22:24 haypo left
plobsing if we make a segment a PMC, swapping one in place of another is simply a become: call. 22:24
you would of course, need subs to cooperate with you or things would fail pretty fast 22:25
NotFound Did we have some usage for anonymous classes? Every time I try to use one, I get some NULL PMC access. 22:32
atrodo cotto_work> Just in time to leave for the day: creative.atrodo.org/2010/12/my-litt...world.html 22:45
cotto_work atrodo: I look forward to reading it! 23:00
23:02 dmalcolm left
cotto_work atrodo: thanks for the post. I'm only at the first point and I already see something I wasn't clear about communicating. 23:08
chromatic Yeah, me too. 23:13
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#1663) fulltest) at c3f2c98 - Ubuntu 10.10 i386 (g++-4.5 with --optimize)
chromatic Also the "tight coupling of the MOP" 23:14
But then again, I looked at dukeleto's prototype ops and said "Why do we have call as an M0 op?"
23:22 whiteknight joined 23:23 lucian left, smash left
whiteknight good evening, #parrot 23:29
cotto_work hio whiteknight 23:30
whiteknight hey cotto_work
cotto_work whiteknight: creative.atrodo.org/2010/12/my-litt...world.html
whiteknight atrodo++ 23:31
cotto_work atrodo++ indeed
atrodo: ping
23:39 allison left 23:50 kid51 joined 23:52 lucian joined
kid51 atrodo: Can we have your blog posts picked up by planet.parrotcode.org? 23:53
23:53 hercynium left 23:57 Yuki`N joined
Yuki`N Time to tackle. 23:58
BigNum.
dukeleto Yuki`N++ 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!