Parrot 2.9.1 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | GC tuning | remove deprecations |
Set by moderator on 21 October 2010.
00:00 dngor left 00:20 mikehh joined
dukeleto kid51: how do we fix that? 00:21
kid51 well, I'm not sure I want to fix that, at least in the short run
dukeleto kid51: that is fine, but it sounds like you are bored. We don't want that ;) 00:22
kid51 I have a lot else to do, particularly in the project
dukeleto: I wonder if you could think about writing up either a blog post or a post to parrot-users about PLparrot (or whatever it is called) 00:23
As I mentioned last week, the fact that you said you had 4-5 people making contributions to it means that it qualifies as Parrot's second 'major user' (after P6) 00:24
I want to understand that better, and I want all our contributors to start understanding that as well. 00:25
And it's the sort of topic I'd like to see on parrot-users, which is under-utilized
dukeleto kid51: sure. I have written a few blog posts about it, but some things have changed. You *do* read my blog posts, right? 00:26
00:26 luben left, hudnix left
kid51 Reading your post on the google OS blog now, I realize that I did read that one ... 00:27
... but to be honest I can't say that I recall others.
dukeleto kid51: leto.net/dukeleto.pl/postgres/
kid51: that will give you more background about PL/Parrot and PL/Perl6 00:28
kid51 Ah, so the two postgres-related ones were in April and June -- no longer in my short-term memory. ;-) 00:29
00:30 hudnix joined
kid51 So, perhaps an update on their current status is in order, starting with ... 00:31
... do they still work, or have we in Parrot broken them? ;-) 00:32
If they're working, what will it take for us (Parrot + Postgres people) to turn this work into a promotable product? ...
... which will then attract more developers to the product and the project(s)? 00:33
Let's pose this more concretely.
What combination of usable product and promotional information about the product (web sites, PDFs, slideshows, etc.) would be needed to enable me to walk into a meeting of Postgres developers in NYC, talk about the product, and not get laughed out of the room? 00:34
(... especially given that I've never used postgres :-) ) 00:35
00:36 hudnix left
dukeleto kid51: good things to think about 00:38
kid51: i have been thinking about those things lately, along with davidfetter
kid51 We need to get to the point where we can say, "This code is cool and you can use it now!" 00:41
00:53 luben joined
kid51 dukeleto: So how did your "Dynamic Language Interop session" go? 01:05
01:21 dngor_ left 01:23 dngor joined
dukeleto kid51: we can get to that point faster with your help ;) 01:32
kid51: well, i sent notes to parrot-dev 01:33
kid51 dukeleto: yes i've been reading that mail 01:35
01:36 cotto left 01:45 preflex left, whiteknight left, szabgab left, jhelwig left, contingencyplan left, he left, jnthn_ left, bacek_at_work left, aloha left, plobsing left, Maddingue left, Khisanth left, dukeleto left, pjcj left, integral left, pmichaud left, Tene left, KatrinaTheLamia left, autark left, sECuRE left, moritz left, Infinoid left, confound left 01:49 preflex joined, whiteknight joined, szabgab joined, jhelwig joined, contingencyplan joined, he joined, jnthn_ joined, bacek_at_work joined, aloha joined, plobsing joined, Maddingue joined, Khisanth joined, dukeleto joined, pjcj joined, integral joined, pmichaud joined, Tene joined, KatrinaTheLamia joined, moritz joined, sECuRE joined, autark joined, confound joined, Infinoid joined
whiteknight I'd be very happy to hear more details about some of the things that happened at the GSoC mentor summit 01:52
some of these emails are very intriguing, if cryptic
kid51 whiteknight #mentorsummit on irc.freenode.net 01:55
whiteknight thanks
kid51 but everyone is probably at dinner now
02:01 dngor_ joined 02:04 dngor left
whiteknight blah. Dinner is for the weak 02:05
02:28 dngor_ is now known as dngor 02:30 whiteknight left 02:50 kid51 left 02:54 theory left
davidfetter dukeleto, ? 03:24
03:43 theory joined 03:44 pjcj left
bacek aloha, humans 04:10
dalek rrot: r49640 | bacek++ | branches/generational_gc/src/ops (2 files):
Fix "op copy" to avoid messing with PMC guts which breaks under
 04:12
rrot: r49641 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Don't propagate string from current generation into current generation.
bacek seen kid51 04:15
aloha kid51 was last seen in msg 2 hours 19 mins ago <private message>.
bacek msg kid51 Only gen_gc atm 04:16
aloha OK. I'll deliver the message.
04:20 Drossel joined 04:23 Kulag left
dalek rrot: r49642 | bacek++ | branches/generational_gc/src/pmc/scheduler.pmc:
Fix bad assumption in Scheduler.destroy.

  $P0 = new 'Scheduler'
null $P0 sweep 1 Then we have NULL interp->scheduler.		 04:58
davidfetter msg kid51 if you don't find me on irc, contact me at david@fetter.org re: NYC PostgreSQL Users' Group 05:44
aloha OK. I'll deliver the message.
bacek NICE.... 06:20
Who introduced Parrot_pmc_new_noinit??? 06:21
plobsing why is it a problem? 06:26
06:36 cotto joined
cotto ~~ 06:37
plobsing |ĀÆ|_|ĀÆ|_ 06:44
cotto Don't be a ā–”. 06:45
plobsing I tried /\\/\\/ before, but ppl mistake it for a regex 06:48
cotto That would send the wrong signal. 06:58
07:55 theory left 08:13 fperrad joined
bacek ~~ 09:00
plobsing, yes. new_noinit doesn't initialize custom_mark and custom_destroy flags. For example thawed Hash will not mark stored keys/values...
sorear then thaw is broken 09:03
the contract of new_noinit is that the caller gets to deal with that mess
bacek sorear, custom_mark, etc incapsulated inside PMC. caller of new_noinit shouldn't worry about it. 09:22
afk again
09:28 bacek left 09:51 bacek joined, Hunger left 09:52 Hunger joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#779) fulltest) at r49642 - Kubuntu 10.10 amd64 (g++-4.5) 10:10
10:13 contingencyplan left
dalek rrot: r49643 | bacek++ | branches/generational_gc/src/pmc (2 files):
Initialize custom mark and destroy flags in thawed objects. It's required because ImageIO uses pmc_new_noinit which doesn't initialize these flags
 10:21
rrot: r49644 | mikehh++ | branches/generational_gc/src/gc/gc_ms2.c:
[generational_gc] add casts to get g++ to build
 10:36
10:54 whiteknight joined 11:00 M_o_C joined
whiteknight good morning, #parrot 11:02
tadzik good morning whiteknight
about PCT, I accidentally asked that on #perl6. Can compilers using PCT be written in something else than NQP, or is there a perspective for that?
(will be back soon) 11:10
11:10 tadzik left 11:23 kj joined 11:25 kj is now known as kjs 11:32 tadzik joined 11:42 lucian left
dalek rrot: r49645 | mikehh++ | failed to fetch changeset:
[html_cleanup] merge latest trunk (r49644) into the branch
 12:25
mikehh hoping this stuff is going to be a lot easier with git 12:26
12:44 kjs left 12:46 fperrad left 12:55 kid51 joined, fperrad joined 12:58 tadzik left 13:07 M_o_C left 13:09 tadzik joined 13:16 whiteknight left 13:37 kid51 left 14:00 pjcj joined 14:09 jan left 14:46 kid51 joined
nopaste "kid51" at 192.168.1.3 pasted "generational_gc branch: down to 2 FAILs at r49644" (20 lines) at nopaste.snit.ch/24792 14:48
14:52 dngor left 15:02 M_o_C joined 15:04 kid51 left, M_o_C_ joined 15:09 M_o_C left 15:23 pjcj left
dukeleto 'ello 15:58
16:26 jsut joined 16:31 jsut_ left 16:32 theory joined
davidfetter hai dukeleto 16:51
17:23 masak joined 17:33 mikehh left
dukeleto 'ello 17:33
cotto ~~ 17:35
masak Ā«greetĀ« @people 17:37
tadzik or should it be: Ā«.greetĀ«?
masak no, you can't put method calls inside metaoperators. 17:38
tadzik oh 17:39
masak only operators. 17:40
plobsing isn't . an operator?
tadzik [1,2,3]Ā».say works for me
masak plobsing: yes. ish.
tadzik: right, but then '.' acts as the operator, as plobsing pointed out. 17:41
tadzik mhm
masak whereas in Ā«.greetĀ«, what you've actually surrounded is a method call, not an operator.
in STD nomenclature, I think both '.' and 'Ā».' count as variations on the 'dotty' rule. 17:42
(and in that sense, they are not operators, they're dotties.) 17:43
plobsing what about junctions?
plobsing . greet(all(@people)) 17:44
masak that's fine.
but due to TTIAR, you can't have white space around the dot.
unspace is fine, though. 17:45
plobsing ttiar?
masak "Two Terms In A Row". a restriction on the syntax of Perl 6.
to a first approximation, terms and operators must alternate. 17:46
(except when they mustn't)
s/mustn't/don't have to/ 17:47
plobsing heh. consistent like a natural language. good thing I can get by in those most of the time. 17:48
tadzik hmm, what about @people ==> tadzik.greet(*)? 17:50
masak not sure you need the (*)
though 'tadzik' needs a sigil, unless it's a sub call. 17:51
plobsing wait. are methods and attributes not in different namespaces?
masak oh, entirely.
why do you ask? 17:52
plobsing so how would it differentiate tadzik.greet the attr access from tadzik.greet the method curry?
masak the dot always means the method call. 17:53
in the case of public attribute (those with a dot twigil), an access method has been automatically generated for your class, unless you wrote one yourself. 17:54
but (barring invisible optimizations), the dot always goes through a method. 17:55
tadzik tadzik could be a class name. Ha
plobsing lowercase class names? seems magical. what would that mean? 17:56
masak obviously it's the low-level tadzik :P 17:57
tadzik :)
18:03 whiteknight joined 18:11 contingencyplan joined 18:28 tadzik left 18:59 M_o_C_ left 19:28 mikehh joined 19:46 pjcj joined 19:51 pjcj left 19:53 jan joined, pjcj joined 20:00 kid51 joined
kid51 whiteknight: ping 20:04
whiteknight pong
kid51 whiteknight: I think this sentence in your latest blog post needs clarification:
"Also, the kinds of documentation that users need is often far different from the kinds of documentation that users need."
whiteknight ha
kid51 We seem to be in a lull as far as actual code contributions go right now ... 20:05
... but perhaps we could make lemonade out of that lemon by using the lull to focus on design and APIs. 20:06
whiteknight fixed
thanks
masak kid51: I think it's a wonderful phrasing of a fundamental problem with user documentation. :)
kid51 The notion that libparrot is our primary product is certainly a startlingly new (to me, at least) interpretation. 20:07
If that concept proves persuasive, then to me it suggests the starting point for a rewrite of the PDDs. 20:11
But I would not one thing:
In conversation with dukeleto last week, and in looking over his blog posts about PLparrot, it's clear that he sees PLparrot and PLperl6 as *embedded* applications. 20:12
And I know that his interest in Parrot, if it had to be reduced to one word, could be said to be "embedding".
20:13 hudnix joined
kid51 Also re blog post: s/commens/comments/ in 1st paragraph 20:14
dalek rrot: r49646 | nwellnhof++ | trunk (4 files):
[io] Fix unbuffered read

for quite a while. Also includes a small fix to line-buffered reads after EOF.		 20:21
kid51 nwellnof++ for breaking the dry spell! 20:22
20:28 hudnix left 20:41 lucian joined
dukeleto is hacking with RTEMS devs at the GSoC mentor summit 20:43
kid51 Are you ... embedding? 20:44
dukeleto kid51: something like that 20:45
kid51 Apropos whiteknight's latest blog, what do they think of Parrot's API documentation, etc.?
post on parrot-users strongly recommended
I hope shockwave posts his 3-part critique on parrot-users
cotto is learning that he'll have to find a good balance between blogging and coding. 20:48
20:50 hudnix joined 20:52 PacoLinux left 20:56 jhelwig left, jhelwig joined
bacek Good morning, humans 21:06
kid51 Good Monday morning bacek.
bacek kid51, aloha 21:07
masak preved, bacek.
bacek masak, с Š“Š¾Š±Ń€Ń‹Š¼ утрŠ¾Š¼
dalek rrot: r49647 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Paint strings white as well after m&s.
 21:08
rrot: r49648 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Add more self-checking functions.
kid51 bacek: I tested the generational_gc branch and saw that we're down to 2 failing files.
dalek rrot: r49649 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Change pobj2gen signature to sematically correct version.
masak bacek: here it's still last night :)
bacek kid51, btw, I use "c++ comments" for code require additional attention.
masak, welcome to future! :)
kid51 Yes, cotto flagged that for me, so I think I reverted all my changes 21:09
bacek kid51, ok, thanks
kid51 Also, generational_gc at least built on my small-resource box.
bacek kid51, it's horribly broken atm... I' still trying to fix moving objects between generations. 21:10
dukeleto is cross compiling RTEMS on his netbook right now 21:11
kid51 particle: Attention! Cross-compiling in progress ;-) 21:12
bacek good thing - now I'm catching errors earlier.
21:22 zostay left 21:23 zostay joined
dalek rrot: r49650 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Ressurect usage of GC_generational_2_FLAG for bringing objects into same

Main reason - we still have to process objects from current generation because they can have fresh references to young objects.		 21:23
rrot: r49651 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Switch on valiadationf of generation after bringing them together.
rrot: r49652 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Set fred Linked_List_Header to 0 for simplify catching of errors.
rrot: r49653 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Reduce gc_threshold dramatically to enforce GC collecting for debugging reasons
whiteknight bacek++ 21:27
bacek: how is that branch looking now? What help do you need?
bacek whiteknight, a lot. More eyes basically.
whiteknight okay, what are my eyes looking for? broken tests? etc? 21:28
21:29 jsut_ joined
bacek overall logic 21:32
failed asserts 21:33
world domination
(last one is optional)
sorear whiteknight++ #More critical review, although I'd still like that PDD fix
whiteknight what PDD fix?
er, which 21:34
there are too many to choose from
sorear in the sense of a drug fix
whiteknight ah
sorear I saw that PDD03 blog post and wants more ;)
21:34 jsut left
kid51 sorear: But we need the fixing in the other sense! Which one troubles you the most? 21:35
sorear all the ones that have personally fought me havent' made it out of drafts/ yet 21:38
if you want to do 31 be my guest
(start with the lunacy of havign 2 versions of 31)
nopaste "kid51" at 192.168.1.3 pasted "Current 'svn diff' of generational_gc branch from its branch point." (3106 lines) at nopaste.snit.ch/24802
dalek rrot: r49654 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Don't process non-black objects during bringing to same generation.
 21:39
rrot: r49655 | bacek++ | branches/generational_gc/src/gc/gc_ms2.c:
Add bit of documentation.
kid51 sorear: Speaking of "drafts" ...
... can we draft you onto our fix-the-design-and-APIs team?
dukeleto just watched an RTEMS guy run Rakudo Perl 6 on RTEMS 21:42
sorear Yes, if my lack of Parrot contributor karma isn't going to be a problem 21:43
21:43 bacek left
sorear (relative lack( 21:44
plobsing purl, karma sorear
aloha, karma sorear 21:45
aloha plobsing: sorear has karma of 150.
plobsing sorear: 2 versions? 'find docs | grep 31' only turns up one 21:52
sorear I see some kind soul finally deleted docs/pdds/draft/pdd31_hll_interop.pod 21:55
plobsing \\o/ progress ... sorta 21:56
kid51 sorear: Against *my* recommendations, purl was banished from this channel. aloha as yet only grasps people's *recent* karma. 21:58
So your karma poses no obstacles to membership on that team :-)
21:58 kid51 is now known as kid51_at_dinner
plobsing nobody has karma! 21:58
aloha, karma plobsing
aloha plobsing: plobsing has karma of 19.
21:59 masak left 22:02 fperrad left
sorear aloha grasps karma both here and in #perl6 22:02
in #perl6 I've been frantically committing to something which does not (currently) add value to the Parrot ecosystem 22:03
plobsing I've only done 19 karma-worthy things in the last couple weeks. Tangential work is better than no work. 22:04
22:16 kid51_at_dinner left
dukeleto plobsing: i bumped into a Minix guy that said you have been asking them lots of detailed questions 22:25
plobsing: keep up the good work ;)
dukeleto is pair programming with registered Linux user #26 22:27
sorear ... Linux used to require registration? 22:30
dukeleto msg kid51 i have a few more config steps that are needed for Parrot+RTEMS, you game?
aloha OK. I'll deliver the message.
dukeleto sorear: it was optional
sorear: counter.li.org/ 22:31
aloha, karma aloha 22:32
aloha dukeleto: aloha has karma of -5.
dukeleto lollerskates 22:33
msg kid51 i just created TT#1840 and assigned it to you 22:41
aloha OK. I'll deliver the message.
22:44 bluescreen left 22:55 bluescreen joined

Powered by Cro and the RakuĀ® Programming Language.

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

Thank you!