Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Mar 31 | tinyurl.com/2pmnlq
Set by moderator on 17 March 2008.
PerlJam Hrm. looks like the recent digest changes have broken building on cygwin. 00:27
00:38 purl joined 00:42 contingencyplan joined
kid51 PerlJam: Can you send [BUG] report on that? 00:44
PerlJam sure. I was going to try to figure it out for myself and fix it though (since it would be a good way for me to learn about cygwin from a systems level I think) 00:46
But I haven't the heart to do that I now realize.
kid51 k 00:50
PerlJam odd. I updated cygwin and ran "git svn rebase" in my parrot dir and it appears to be modifying every file all of a sudden 01:11
01:22 AndyA joined
kid51 Coke: which RFC were you talking about in your msg? 01:28
Coke digs through his gmail inbox... 01:29
#39142
kid51 k
01:35 petdance joined 01:36 integral joined
Infinoid ok, I've boiled down RT #51790 about as much as I can, but a lowlevel pmc/runloop expert needs to look at it. 01:42
I'm gone for the night. happy "little green men" day, everyone! 01:43
diakopter martians day? 01:44
Infinoid martians == leprechauns
cotto_home does src/pmc/multiarray.pmc have any purpose? 01:45
it doesn't seem to be either tested or referred to anywhere else in Parrot 01:46
nopaste "kid51" at 70.107.6.159 pasted "multiarray instances" (16 lines) at nopaste.snit.ch/12532 01:48
cotto_home nm. premature senior moment 01:50
kid51 I couldn't tell you what it's *for* -- don't know much about the pmcs -- but it's certainly *there* :-) 01:56
cotto_home I agree 01:58
kid51++ for the sanity check 02:09
02:11 TimToady joined 03:18 diakopter joined 03:28 jrt4 joined 03:38 camgirl29 joined
cotto_home anyone want to squish a pmc2c.pl bug? 04:38
nm. found it 04:56
05:21 Eevee joined
Infinoid you know, it would be really nice if RT could get around to sending the replies I posted 10 hours ago, to the list 05:31
would have saved c the trouble of debugging to the same point I got to...
cotto_home is this another "c"? 05:36
Infinoid chromatic, sorry :) 05:37
goodnight
purl Goodnight Gracie!
cotto_home I though so
"cotto" doesn't need abbreviation as much
svnbotl r26459 | chromatic++ | trunk: 05:41
: [dynpmc] Fixed crypto PMCs so that they don't accidentally mark non-GCable
: memory.
diff: parrotvm.org/svn/parrot/revision/?rev=26459
r26460 | chromatic++ | trunk: 06:13
: [src] Enabled calling the init() vtable entry on dynpmcs. They could use it
: now and then. (Reported by Bernhard Schmalhofer in RT #51790).
diff: parrotvm.org/svn/parrot/revision/?rev=26460
07:07 IllvilJa joined 07:08 uniejo joined
svnbotl r26461 | fperrad++ | trunk: 07:15
: [languages]
: update LANGUAGES_STATUS.pod from languages-smoke results
r26462 | fperrad++ | trunk:
: [Lua]
: - fix md5.tostring()
diff: parrotvm.org/svn/parrot/revision/?rev=26462
r26463 | fperrad++ | trunk: 07:22
: [scheme]
: fix all tests on Win32
r26464 | fperrad++ | trunk:
: [WMLScript]
r26465 | chromatic++ | trunk: 07:35
: [src] Documented the use of interp->lo_var_ptr, so it's more obvious why it's
: there. A rename might be in order; any suggestions?
: Reported by Jesse Taylor in RT #51566.
r26466 | fperrad++ | trunk:
07:40 contingencyplan joined 08:18 ewilhelm joined 08:29 AndyA joined
svnbotl r26467 | fperrad++ | trunk: 08:37
: [PLATORMS]
: - update MinGW32 gcc 3.4.2
diff: parrotvm.org/svn/parrot/revision/?rev=26467
08:48 cognominal joined 08:57 mire joined 09:09 kj joined 10:33 ruoso joined 10:41 kid51 joined
svnbotl r26468 | rblasch++ | trunk: 11:30
: Windows PLATFORMS updates.
diff: parrotvm.org/svn/parrot/revision/?rev=26468
11:52 ask_ joined
svnbotl r26469 | coke++ | trunk: 12:26
: [distro]
: track changes from r26466
diff: parrotvm.org/svn/parrot/revision/?rev=26469
12:34 askie joined 12:49 skids joined 12:53 petdance joined 13:02 gryphon joined, bgeron joined
svnbotl r26470 | coke++ | trunk: 13:16
: [tcl]
: Update some comments regarding RT tickets that shouldn't have been.
diff: parrotvm.org/svn/parrot/revision/?rev=26470
r26471 | fperrad++ | trunk: 13:23
: [WMLScript]
: - fix ppd17pmc (parseNumber logic)
diff: parrotvm.org/svn/parrot/revision/?rev=26471
13:44 jrt4 joined 13:51 spinclad joined
Coke Anyone who is bored, some of those new tcl TODO items are probably pretty easier fixable. 14:30
"easily". (basically, if the test is in t/cmd_foo.t; change the diagnostic error output in runtime/builtin/foo.t
svnbotl r26472 | coke++ | trunk: 14:31
: [tcl]
: Ran all the tests through tcl 8.5.1; some behavior (mostly diagnostic output)
: has changed; update the tests, TODO them for parrot where necessary. Delete
: a few invalid tests that partcl happens to pass, but shouldn't.
diff: parrotvm.org/svn/parrot/revision/?rev=26472
14:56 rdice joined
kj for other bored people: episode 4 of the Parrot Compiler Tools Tutorial is just published. 15:29
15:32 skids joined 15:44 Leonid joined
Leonid Hi all. I want to contribute code to the parrot project. Where do I start ? 15:44
Coke Hi, Leonid. What languages do you know? 15:45
Leonid c++
perl
Coke ok. 15:46
there are a ton of perl modules used to build and test parrot; the VM itself is written in C, not C++: do you have some comfort there?
Leonid I am a little bit newbie to the open source contribution. I will be glad a more "in depth" explanation :)
Yeah C is good. 15:47
btw why C and not C++ ?
any special reason ?
Coke Yup. moment.
kj I think it's in the FAQ...
www.parrotcode.org/faq/ 15:48
Coke www.parrotcode.org/faq/#What_langua...ritten_in?
shorten Coke's url is at xrl.us/bhuwr
askie :D
Coke not an in depth answer, but there you go. 15:49
Leonid ok I will read it... I am at work now :)
So what my futher steps ?
Eevee echoing Leonid, although I was going to lurkmore for a while first 15:50
Coke Most of our "todo" work is listed in the ticketing system, but it can be hard to find something to work on.
rt.perl.org/rt3 15:51
kj Leonid: why does Parrot interest you? (this might help in deciding what you want to do)
Coke Looking for new/open tickets with [TODO] or [C] or [Perl] in the system is a good first pass, and pretty much what I was going to do.
Leonid Because I love perl and the Idea of VM for it ..... wow :) 15:52
Coke reading through docs/ (after you build) and src/ to look around if you just want to jump in the deep end.
There is the core parrot stuff to work on, or various HLL's that need work. If there's one of those you're interested in, that might be a better place to start. 15:53
Leonid Do I need to get an UserName or something ???
Coke No.
(you can browse rt as a guest)
(that may be the wrong the url for that. :) 15:54
kj Leonid: (love perl and VM for it) that's pretty general :-P
Leonid OK after I will get around .... How I inform you ppl I am started to working on one of the tasks ? 15:56
kj you probably have questions by then ;-)
Leonid after I will resolve all the questions ? I just want to understand the procedure . 15:59
16:00 sjansen joined, skids joined
kj well, depending on what you do; for instance, suppose you want to do some code-cleanup (layout, maybe you found a bug), then you can create a patch and send it to the mailing list (you'd need to subscribe to that beforehand) 16:00
if the patch is sane it'll get applied by someone with commit rights 16:01
Coke if your patch is for an existing ticket, you can add it via the web interface, also.
get enough patches applied, you can eventually get a commit bit.
Leonid ohh cool :) 16:03
The link you sent me is the bugs page ... do you have any TODO list too ? or is it the same ?
kj it's an issue tracker
"issue" is a better word, as it contains not only bugs, but also proposals and other... ehm.. issues :-) 16:04
but to answer your question: yes, that's the todo list (is the same)
Leonid The best way to start is to download the code and compile it first ... right ? 16:10
kj yes
and run make test
Leonid understand how to work with it and then start to add my own nasty things :)
16:17 Theory joined 16:23 xiaoyafeng joined 16:34 parrot-poke joined 16:37 Leonid left 16:38 barney joined
barney chromatic++ for fixing RT#51790 16:46
barney running 'make fulltest' 16:50
Coke barney: Any issues so far? 17:04
17:11 davidfetter joined
Coke seen ambs? 17:13
purl ambs was last seen on #parrot 5 days and 20 hours ago, saying: no, different architecture, and different messages in the test accordingly to it [Mar 12 13:59:37 2008]
barney Coke: no issues yet, but Linux is a bad platform for 'make fulltest'
Coke ORLY?
purl YA RLY.
Coke I recommend doing it on feather if you have an account.
(and getting one if you don't)
unless you have a beeeeg box. 17:14
Juerd If you want an account fast, mail me the request within 15 minutes :) 17:15
barney It's not the speed, it's that don't catch many errors
Juerd has a tiny tuit :) 17:16
barney Juerd: I'll come back to yout in 5,5 months
Juerd Okay 17:17
Coke I don't understand. there are errors that are *not* being caught by fulltest? (in which case, how would you know about them?) 17:19
jmcadams++ # new TPF PR chair. 17:21
Juerd I wonder if we'll see activity in TPF PR
barney I'm talking about config and platform dependent errors. E.g. the MDx failure on Cygwin. 17:22
Coke not much you can do about that. 17:23
that's why we have PLATFORMS 17:24
barney and smoke.parrotcode.org/smoke/ 17:25
purl smoke.parrotcode.org/smoke/ is too short
17:28 skv joined
Coke so, yah, check what you can, but don't worry too much. 17:31
*** hey, everybody, run make fulltest and update PLATFORMS so barney doesn't feel guilty! ***
barney created release branch svn.perl.org/parrot/branches/RELEASE_0_6_0 17:35
Coke barney: ... don't do that.
most of our releases so far have not been branch-style, but trunk-style; let's keep them uniform for now. 17:36
barney Why? It's only a temporary branch. Following 'Appendix 1' in release_managers_guide.pod 17:37
Coke I think that was the one that was recently fleshed out by Bob, who did his release unlike other recent releases, no? 17:39
seen allison? 17:40
purl allison was last seen on #parrot 6 days and 22 hours ago, saying: jonathan: yes, it's there now [Mar 11 12:56:49 2008]
17:40 ambs joined
davidfetter oi, ambs 17:40
cotto_work Coke, do you mind checking my commit bits? Allison said they were set during the last parrotsketch, but I couldn't commit a fix last night. 17:42
Coke cotto_work: cotto, aren't you the guy from MS? 17:43
barney Coke: yes that's the notes from Bob Rogers 17:44
Coke barney: It's my understanding that allison would rather we not do it that way.
But, obviously, the docs were never updated. Proceed as you see fit. =-)
barney k 17:45
cotto_work yes, although that fix was written on my own time
(I won't be committing anything written at work until that issue gets settled out.) 17:46
s/settled/sorted/ 17:47
Coke but it's my understanding that things were stalled until allison & MS completed their talks. 17:48
cotto_work that's what I thought until chromatic said otherwise: 17:50
<particle> is cotto a committer yet?
<chromatic> Yes.
Coke Then you have a question for chromatic during PS today. =-)
cotto_work ok
17:51 svnbotl joined 17:58 cout joined
svnbotl r26474 | bernhard++ | RELEASE_0_6_0: 18:10
: Prepare for the release of Parrot 0.6.0
diff: parrotvm.org/svn/parrot/revision/?rev=26474
18:15 Psyche^ joined
barney is running 'make fulltest' again 18:16
svnbotl r26475 | bernhard++ | RELEASE_0_6_0: 18:17
: [docs]
: Recommend to use 'make distro_tests'
r26476 | bernhard++ | RELEASE_0_6_0:
: Regenerate native PBC for t/native_pbc/*.t
diff: parrotvm.org/svn/parrot/revision/?rev=26476
ambs waves
Hi
18:17 ambs left 18:18 ambs joined
ambs waves 18:18
(ok, working now)
18:25 chromatic joined
svnbotl r26477 | bernhard++ | RELEASE_0_6_0: 18:26
: [docs]
: Suggest to move the release branch.
diff: parrotvm.org/svn/parrot/revision/?rev=26477
chromatic #ps in 2 18:28
ambs bernhard++ 18:30
Go, 0.6.0, go!
barney /topic #parrot Parrot 0.6.0 release is on the way | parrotcode.org/ 18:39
chromatic We need a t/distro/prefix_space.t. 18:40
Coke ... which would check what? 18:41
chromatic That barney really did change the topic.
Coke ... doh 18:43
moderator #parrot Parrot 0.6.0 release is on the way | Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Mar 31 | tinyurl.com/2pmnlq 18:44
barney Just following the docs 18:46
18:46 ambs joined
moderator Parrot 0.6.0 release is on the way | Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Mar 31 | tinyurl.com/2pmnlq
Infinoid chromatic: to enforce the "Indentation must consist only of spaces." rule of pdd07? 18:47
thought tabs.t did that 18:48
chromatic I was making a joke. 18:49
Coke In my defense, I got it when you explained it above. ^_^ I don't know what oo-oids excuse is. =-) 18:50
chromatic A better sense of humor than both of us combined, perhaps. 18:51
ambs This Colloquy is strange... some strange smiles rendered.
Infinoid heh. my excuse is sheer idiocy :)
Eevee chromatic: don't worry, I got it 18:52
chromatic Coke's using the I-want-to-be-Japanese-kitty-girl smiley set.
Coke neko-o kudasai!
ambs then, that line includes a strange smile with black hair? :) 18:53
chromatic hai, so desu-ka
Coke wishes he had hair. :P 18:54
ambs laughs
ambs still has some
Infinoid just had most of his cut off
Eevee I was going to spew kana but I forget the scim incantation 18:55
chromatic reimplements cotto's patch in RT #51832. 18:57
barney created parrot-0.6.0.tar.gz 19:00
Coke Wow, what an awesome hacker you iz.
chromatic: after your post about 25%, I checked out my stats on ohloh; I can tell you that if I'm #2, going by number of commits isn't telling much. =-) 19:01
chromatic It just means you're old, old man.
You'll turn senile soon, and then I'll catch up. 19:02
Coke get off my lawn, you scamp.
ambs will parrot 0.6.0 pass all tests on my Mac? :D 19:03
chromatic x86 or ppc? 19:04
ambs x86
it was failing something related with pcre
chromatic It will probably continue to fail then. I don't remember seeing any commits about that. 19:05
ambs I don't remember any as well 19:06
chromatic I suggest bribery. 19:07
ambs waits for barney's URL :D 19:08
barney Commits can commence. Release takes place in a branch. 19:09
chromatic Good.
Coke chromatic: good you like branches, or good you like to commit? 19:10
chromatic Good I just committed. 19:12
svnbotl r26478 | chromatic++ | trunk:
: [lib] Fixed #line directives in PMCs (based on a patch from Christoph Otto in
: RT #51832).
diff: parrotvm.org/svn/parrot/revision/?rev=26478
Coke peers at the diff between the two diffs. 19:13
chromatic The order of atoms in a character class is almost always immaterial. 19:14
Coke ... almost!?
chromatic ^
ambs svn ups 19:15
Coke chromatic: cheater. :P
chromatic I prefer to think of this as brilliance. 19:16
Infinoid allright. I don't see any reason not to do the codingstd stuff iteratively, so I'm checking in what I have, and will continue work with commits to individual tests 19:17
chromatic One warning: the macro-wrapping test probably won't work in the large. 19:18
svnbotl r26479 | bernhard++ | RELEASE_0_6_0:
: [docs]
: Mention the the news is for 0.6.0
diff: parrotvm.org/svn/parrot/revision/?rev=26479
Infinoid I don't think I'm going to get each and every codingstd test working under this framework 19:19
they don't all duplicate the code I'm consolidating. but that isn't a bad thing
Coke that's fine. 19:20
chromatic ... then someone could look at all of the duplicate code in the config/ modules and reduce that. 19:22
Coke or src/pmc! 19:23
TclList has a 'reverse' method. Should this get moved up to RPA? A?
Infinoid Larry said something about removing all of this whitespace we've been duplicating :)
chromatic Isn't that just a different comperator to our quicksort? 19:24
svnbotl r26480 | infinoid++ | trunk:
: [t/codingstd]
: start consolidating codingstd tests per RT#51718:
: add Parrot::Test::Util::Runloop
: patch c_cuddled_else, check_isxxx, trailing_space, c_cppcomments and c_code_coda to use it.
diff: parrotvm.org/svn/parrot/revision/?rev=26480
Coke chromatic: having 'reverse' means I don't have to write twice as many comparators.
(but no, I can't think of a use for 'reverse' on an unsorted array.) 19:25
19:28 ambs_ joined
Infinoid lots of trailing spaces in the .pir files 19:35
# That's 219 failed files out of 1496 files total.
Tene commits a new HLL in the middle of the release. 19:41
;)
cotto_work those always work correctly the first time 19:42
barney too late
chromatic Once again, that ruthless Bavarian efficiency prevents us from screwing up. 19:43
ambs :D
barney++
Tene Curses, vfoiled again!
barney Is t/manifest/*.t supposed to fail in a dist ? 19:44
chromatic Shouldn't.
purl shouldn't is such a funny word.
chromatic Infinoid, no need for "scalar" in that patch.
ambs purl: forget shouldn't
purl ambs: I forgot shouldn't
svnbotl r26481 | infinoid++ | trunk: 19:58
: [t/codingstd]
: [12:44] <@chromatic> Infinoid, no need for "scalar" in that patch.
: Normally I like my numifications to be explicit, but he's right, they're a bit ugly here.
diff: parrotvm.org/svn/parrot/revision/?rev=26481
20:10 peeps[work] joined
Coke in general, replacing PIR by C should improve speed, neh? 20:11
(where C == custom ops, PMC methods/vtables, etc.) 20:12
barney: have you cut the release yet? 20:15
Infinoid wow, removing trailing whitespace from a quine is interesting 20:16
barney Currently retesting the tarfile
svnbotl r26482 | bernhard++ | RELEASE_0_6_0:
: #51860: Failures in t/manifest/*.t in a release
: Comment out the 5 failing test cases.
diff: parrotvm.org/svn/parrot/revision/?rev=26482
Coke Infinoid: I hate those quines. hate them hate them hate them.
I had to update the copyright in them a while back. =-)
Infinoid heh
Coke barney: I've never seen manifest test failures in a release before. odd.
Infinoid I'm just removing all the 32's that appear before 10's
Coke Infinoid: isn't there a tool to regen that portion?
barney I added manifest_tests to 'make fulltest' 20:17
Coke barney: ... so it's your own fault then, ok. =-)
chromatic Coke, if control flow crosses the C/PIR boundary too often, that could slow you down.
Coke chromatic: ... ah. analagous to SQL + PL/SQL.
doesn't each opcode invocation cross that boundary? 20:18
chromatic I'm thinking more specifically of marshalling/demarshalling into and out of PCC.
Coke (ignoring JIT)
PerlJam chromatic: you said something about profiling PIR earlier ... do we have a way to profile PMCs? 20:19
chromatic Profile them how?
Coke the only profiling parrot has right now is with -p, which, as I recall, is opcode level.
chromatic I use Valgrind/Callgrind and that's pretty good at the C level, but connecting that to the PIR level is not always easy.
PerlJam chromatic: well, profiling (from the parrot perspective) how much time is spent in each PMC's methods (for instance) 20:20
chromatic Callgrind works pretty well for that.
PerlJam okay, I'll add that to the big-list-of-things-to-learn :)
chromatic It's fairly easy. 20:22
alias cgp='valgrind --tool=callgrind --dump-instr=yes --trace-jump=yes ./parrot'
tewk seen parrot-poke: 20:24
purl I haven't seen 'parrot-poke:', tewk
Coke wonders if that works on osx. 20:25
chromatic I *think* there's a Valgrind that works there.
Coke eh. feather has one installed. 20:26
20:26 slightlyoff joined
Coke Is there a nice analizer for the resulting callgrind file? 20:28
chromatic callgrind_annotate
If you want to profile Tcl, dump the Tcl code to a PIR file and profile that. 20:29
Otherwise you'll spend a lot of time in TGE/PGE.
Coke chromatic: callgrind shows # of invocations, not time spent? 20:31
chromatic RIght. 20:33
Number of virtual processor cycles.
Coke ... ok, that sounds closer to time than # of invocations. 20:34
chromatic Yeah, it's more accurate across runs than counting u-secs.
Coke ok. 20:35
of course, I suspect I'm going to run this and find I'm spending lots of time in RPA:get_int_keyed_pmc or something. 20:36
moderator Parrot 0.6.0 "P&P" released | Please mentor for SoC | parrotcode.org/ | YAPC::NA talks deadline is Mar 31 | tinyurl.com/2pmnlq 20:37
chromatic ... which is why you need to optimize the PIR. I think. 20:37
Coke how much longer can I expect something to take with callgrind?
ambs Plug and Pray?
chromatic A couple of orders of magnitude. 20:38
Coke perhaps I will re-do this on something smaller. =-)
looots of Parrot_assert 20:40
chromatic Yep, that one's a bit spendy. 20:41
I think that's the source of some of our recent slowdowns.
20:41 svnbotl joined
Coke ... wow. after that, the next 13 functions are memory/GC related. 20:42
chromatic The fewer PMCs you create, the faster we go. 20:47
Coke I presume that includes strings? 20:48
(lowercase)
ah, I can avoid the overhead of parrot_asert with -DNDEBUG
chromatic STRINGs yes. 20:50
We can probably fix the overhead of parrot_assert with judicious macroing.
The function call overhead there is a bit of a drain.
Coke chromatic: -NDEBUG avoids it. 20:51
you lose all your asserting, of course.
Infinoid awesome, trailing_space.t now passes for pir files! 10 thousand line diff, 219 .pir files modified, all tests pass. should I put it up for review, or just check it in?
Coke er, -DNDEBUG, i said it rightt the first time.
Infinoid: do all tests pass? 20:52
Infinoid yep
Coke even in languages? =-)
Infinoid that's a good idea
Coke (do a before/after to avoid false negatives)
barney Parrot 0.6.0 "P&P" is now on CPAN, propably not indexed yet 20:58
ambs barney: P&P? 20:59
20:59 amoore joined
Coke ewilhelm++ 21:00
barney++ 21:02
barney the meaning of P&P can be chosen without any restrictions
cotto_work how postmodern of you 21:03
Eevee possibly latin for "ran out of names"? 21:04
ambs lol 21:06
Coke here's a bug for someone to document/fix: smokeserv-server.pl contains the pir code in the __DATA__ segment, which gets included in the HTML on smoke.parrotcode.org/smoke/ 21:10
need to eliminate the code, and then make the perl code test pass in the least annoying fashion.
er, "eliminate the coda from the HTML"
21:25 ambs joined
Infinoid ok, no differences in languages-test 21:37
barney has no Hacker beer for celebrating around. Drinking Augustiner instead. 21:45
21:45 Limbic_Region joined
Infinoid barney: thanks! hooray for 0.6.0 21:46
Infinoid breaks everything
davidfetter what's hacker beer? 21:54
davidfetter vaguely recalls hacker-pschorr(sp?) from münchen
svnbotl r26485 | infinoid++ | trunk: 21:56
: [t/codingstd] check for trailing whitespace in .pir files
: - fix up 219 .pir files which had trailing whitespace.
: - fix up ordinal-encoded trailing whitespace in examples/pir/quine_ord.pir (fun)
: - fix up the expected output of examples/streams/ParrotIO.pir
: (its formatting is sensitive to whitespace changes)
diff: parrotvm.org/svn/parrot/revision/?rev=26485
barney my favorite: www.hacker-pschorr.de/unternehmen/c...no1417.php 22:03
shorten barney's url is at xrl.us/bhvyy
22:35 skids joined 23:00 rblackwe joined 23:08 cotto[work] joined 23:09 Andy joined 23:14 peepsalot joined 23:15 cotto_home joined 23:19 jrockway joined 23:23 marmic joined 23:28 Theory joined
Coke summons particle 23:29
23:29 particle joined
particle Coke: got a few to update parrotcode.org? 23:31
23:40 marmic joined 23:49 xiaoyafeng joined 23:54 CrawfordComeaux joined