Parrot 3.3.0 released | parrot.org | Log: irclog.perlgeek.de/parrot/today” | Accepted GSoC Students announced! | GSoC student information emails coming out soon
Set by moderator on 26 April 2011.
cotto_work the first two are pretty unsurprising, given how nqp uses lexicals 00:01
still instructive
plobsing we could make fill_params a little simpler if the x_keyed vtable override magic handled filling keys with context data in stead of the general path 00:02
jnthn__ cotto_work: The sad thing with lexicals is how much we know but throw away. 00:03
I'm pretty sure we can do better there from NQP's side. 00:04
How does fill_params relate to lexical usage, ooc? 00:05
cotto_work no relation
ambiguous use of English on my part
jnthn__ ah, OK :)
I thought I'd missed something. :)
cotto_work fill_params is unsurprising for anything that uses subs 00:06
jnthn__ aye
It'll be a few months yet before I get to inlining though.
But I expect once I implement that we'll get a big win, given the cost of sub invocation. 00:07
Method inlining is a LOT more difficult though, and we do a lot of method calls.
And inlining only helps for really small things that are good candidates for it.
cotto_work I don't expect inlining to be a general-purpose solution. 00:08
jnthn__ No
Perl 6 will want to do it quite aggressively on operators
cotto_work optimizing pcc will be more productive 00:09
jnthn__ But I see that as independent of optimizing pcc really. Even if PCC was really fast, I'd still be looking to inline those.
The result of the inlining of some really simple things may, oddly, be smaller than the amount of bytecode we'd emit to make the call. 00:10
(I'm thinking arithmetic operations here...)
whiteknight we've known for a long time that the lexicals implementation is sub-optimal
jnthn__ whiteknight: Well, it's not just a "Parrot problem", it's an integration problem. (more) 00:11
whiteknight and I've suggested a number of changes to fill_params that could improve that situation. I'm sure there are others we've never considered
00:11 Andy_ left
jnthn__ I suspect the compiler knows more than Parrot can from the PIR about how lexical accesses could be done. 00:11
whiteknight jnthn__: if the fixed-size allocator is a hotpath, surely it performs better than malloc would? 00:12
jnthn__ whiteknight: It may well be doing, in which case it's awesome even though it's a hotpath. :)
whiteknight: It's just something that stood out in the profile.
whiteknight last numbers I saw showed the fixed-size allocator being better than malloc for most workloads
so we can try to improve it, but it's never before been a high priority 00:13
jnthn__ Right, that's why I mentioned it. :)
I've never seen it show up high in a profile before.
:)
(I'm not complaining, just providing some observations, btw...) 00:14
whiteknight jnthn__: maybe 6model has become so awesome that even the most awesome parts of parrot are less awesome than 6model
jnthn__ heh :)
It tends to show up a long way down in the profile. 00:15
whiteknight ok
jnthn__ Which is a good sign.
OTOH, the 6model core does very little really.
whiteknight with fill_params, I can think of ways to fix that and streamline it. With lexicals we're at the same roadblock we've been at: We have no reasonable plan forward
jnthn__ The most expensive bit of 6model at the moment in the profile is actually a path that we'd not take if nqp was actually taking advantage of it fully. 00:16
whiteknight: I may be able to prototype something for lexicals. 00:19
May be a few weeks before I get chance. 00:20
00:20 dmalcolm left
whiteknight jnthn__: no need to prototype. We're all code monkeys here. All we need is a design 00:21
jnthn__: Go grab a pack of crayons and an old napkin. Sketch something out and send it to us. 00:22
jnthn__: We'll do all the hard work
jnthn__ whiteknight: Well, the easiest idea is having a find_lex/store_lex variant that takes a number of frames down where the lexical is located (e.g. 0 = this callframe, 1 = one down, etc). Then we just follow the outer pointer that many times and do the hash lookup, rather than having to go and do an exists hash lookup in every callframe. 00:25
whiteknight: PAST::Compiler certainly has enough information available to be able to use that op.
Plus the exists checks can go away totally. 00:27
00:27 theory joined
jnthn__ "If you said it was in 1 frame down and there's no lexical called that one frame down, it's your own stupid fault." 00:28
:)
sleep for me... & 00:29
cotto_work 'night
00:29 kid51 joined
whiteknight that is an easy idea, but I wonder how we extend it to work with closures 00:32
er, nevermind. Stupid question
now I have to change around all my priorities to include lexicals 00:33
plobsing ping kid51 00:34
00:59 whiteknight left 01:11 woosley joined 01:15 bacek joined 01:33 bacek left
kid51 plobsing pong 01:40
plobsing kid51: are you still seeing those assertion failures? 01:44
kid51 I just pulled in that branch on darwin/ppc and am reconfiguring and building; 01:45
it will be several minutes
If you backscrolled to last night, you can see that I couldn't figure out how to get gdb past the Darwin DYLD problem. 01:46
plobsing looking at the backscroll, you got some mistaken advice. it isn't "start", it is "run" (or "r" for short). 01:49
it complained about dynlibs because gdb only loads the symbols for dynamically linked libraries once the library has been loaded (and you haven't run the program yet) 01:51
but you shouldn't even need to break. gdb should break at the assertion failure point automatically.
cotto ~~ 01:54
kid51 keeping fingers crossed
nopaste "kid51" at 192.168.1.3 pasted "same failed assertion" (27 lines) at nopaste.snit.ch/42468 01:57
kid51 We have to figure out a more robust way of debugging on Darwin. 01:59
plobsing does darwin have anything like LD_PRELOAD? 02:00
kid51 I don't know. I'm far from knowledgeable about Darwin 02:01
plobsing can you put the library where gdb wants it just for the duration of your debugging session? 02:03
kid51 Perhaps this: tinyurl.com/6bouhwz 02:08
02:09 mtk left, ShaneC joined
plobsing ok, I can work with that 02:10
use 'otool -L ./parrot-nqp' to show what libraries it expects. you should see libparrot somewhere. based on that output we can cheat with DYLD_LILBRARY_PATH 02:11
kid51 I just created this symbolic link: sudo ln -s /Users/jimk/work/pseudoinstall/lib/libparrot.dylib /usr/local/lib/libparrot.dylib 02:14
nopaste "kid51" at 192.168.1.3 pasted "after creating symbolic link to a libparrot.dylib" (15 lines) at nopaste.snit.ch/42469 02:15
02:16 mtk joined
nopaste "kid51" at 192.168.1.3 pasted "otool -L ./parrot-nqp" (9 lines) at nopaste.snit.ch/42470 02:16
plobsing and when you run this program normally (no gdb)? does it exit normally without even a hiccup? 02:17
kid51 Yes. 02:18
./parrot-nqp --target=pir --output=compilers/opsc/gen/Ops/File.pir compilers/opsc/src/Ops/File.pm
compilers/opsc/gen/Ops/File.pir
is created
plobsing without the symlink even? 02:19
are we debugging the wrong program here? 02:20
kid51 No, that was with the symlink in /usr/local/lib
This is starting to remind me of a situation we were in a couple of months back.
Probably when bacek was doing GC work.
plobsing so if you remove the symlink, or change the name temporarily, does the assertion get violated? 02:21
kid51 I tried 'make' to see what happens when I continued.
It failed on the *next* step:
make
./parrot-nqp --target=pir --output=compilers/opsc/gen/Ops/Compiler/Grammar.pir compilers/opsc/src/Ops/Compiler/Grammar.pm
src/gc/gc_gms.c:1231: failed assertion '!PObj_on_free_list_TEST(pmc) || !"Resurrecting of dead objects is not supported"'
make: *** [compilers/opsc/gen/Ops/Compiler/Grammar.pir] Abort trap
I then ran that program manually. 02:22
./parrot-nqp --target=pir --output=compilers/opsc/gen/Ops/Compiler/Grammar.pir compilers/opsc/src/Ops/Compiler/Grammar.pm
And then called 'make' again. 'make' is now continuing as it should.
plobsing that is, well, odd
kid51 Yes, but as I say, I have the feeling we've been here before. 02:23
'make' fails on a certain step, but if you execute the failed command as its own shell command, it completes that step, which permits you to resume 'make'. Repeat as needed. 02:24
It failed again, but it got farther along this time.
plobsing are you running a parallel make? maybe that would put memory pressures that would affect GC.
nopaste "kid51" at 192.168.1.3 pasted "3rd fail point" (10 lines) at nopaste.snit.ch/42471 02:25
kid51 plobsing: On this old machine? *No* possibility of parallel anything.
plobsing well this failure is something I can take a stab at. 02:26
kid51 I am going to have to sleep soon, so feel free to paste suggestions or send email.
plobsing nm /usr/local/lib/libparrot.dylib | grep Parrot_dt_get_datatype_name 02:27
kid51 I will now try to build with my "regular" configuration on this box, i.e., cc=gcc and g++ for link and ld.
$ nm /usr/local/lib/libparrot.dylib | grep Parrot_dt_get_datatype_name
0005fe30 T _Parrot_dt_get_datatype_name
U _Parrot_dt_get_datatype_name
02:30 kid51 left 03:02 luben left
plobsing msg kid51 Can you run the parrot_nci_thunk_gen failure through gdb and break on Parrot_dlsym? Please try to confirm that the call with symbol="Parrot_dt_get_datatype_name" (ignore other calls eg: for Parrot_lib_bit_ops_load) has handle=NULL, that your system's dlsym is being called, and that that function is returning null. 03:03
aloha OK. I'll deliver the message.
dalek TT #2043 closed by plobsing++: .loadlib works when compiling, but not when running PBC 03:39
TT #2043: trac.parrot.org/parrot/ticket/2043
TT #1374 closed by plobsing++: :immediate semantics have changed 03:54
TT #1374: trac.parrot.org/parrot/ticket/1374
TT #2041 closed by plobsing++: :immediate ignored when compiling to PBC
TT #2041: trac.parrot.org/parrot/ticket/2041
rrot: 0ed916d | petdance++ | / (2 files):
updating function arg modifiers
 03:59
rrot/tt1931-nci-parameters-deprecation: cfc1a11 | petdance++ | / (32 files):
Merge branch 'tt1931-nci-parameters-deprecation' of github.com:parrot/parrot into tt1931-nci-parameters-deprecation
04:08 bubaflub joined, bubaflub left
dalek rrot: d93b697 | petdance++ | lib/Parrot/Pmc2c/PMC (2 files):
adding consts and proper function annotations for generated code
 04:12
04:16 preflex left 04:18 jrtayloriv joined 04:19 preflex joined
tadzik theory: ping 04:30
good morning #parrot
PerlJam good evening tadzik :) 04:31
theory tadzik: I get the impression that maybe our days don't quite line up. 04:37
tadzik theory: that may be. I just woke up, it's 6:39 AM 04:39
theory tadzik: Oh, good, so by the time I wake up tomorrow, you might still be around. 04:40
Anyway, off to bed shortly, will try pinging you in the morning. 04:41
But just wanted to introduce myself, see if maybe we could swap distribution war stories.
I just launched PGXN, a CPAN for PostgreSQL.
04:41 jrt4 joined
theory pgxn.org/ 04:43
Anyway, gnite.
tadzik oh, nice. Gnite 04:45
theory API docs: github.com/pgxn/pgxn-api/wiki
theory & # Zzzzzzz
04:45 jrtayloriv left, theory left 04:49 jrt4__ joined 04:53 jrt4 left 05:16 plobsing left 06:16 fperrad joined 06:22 kuku joined, kuku left, kuku joined 06:43 jrt4 joined 06:46 jrt4__ left 06:53 jrt4__ joined 06:57 jrt4 left 07:06 cosimo left 07:07 AzureSto_ left 07:11 contingencyplan left, AzureStone joined 07:15 mj41 joined 07:19 kuku left 07:31 jrt4 joined 07:35 jrt4__ left 07:40 jrt4__ joined 07:41 elmex joined 07:44 jrt4 left 08:02 birdwindupbird joined 08:20 jrt4__ left 08:37 GeJ_ is now known as GeJ
GeJ Good morning everyone. 08:38
08:59 dafrito left 09:57 mtk left, varta left 09:59 ambs joined 10:01 woosley left 10:03 mtk joined 10:18 bacek joined
bacek ~~ 10:23
msg PerlJam, yes, thanks for it.
aloha OK. I'll deliver the message.
10:58 mikehh left 11:21 bacek left 11:49 woosley joined 11:57 Patterner left, lucian joined, Psyche^ joined, Psyche^ is now known as Patterner 12:04 birdwindupbird left 12:06 mtk left 12:11 mtk joined 12:13 whiteknight joined 12:14 whiteknight left 12:20 whiteknight joined
whiteknight good morning, #parrot 12:25
coke_ really needs to sign back in from his normal location. 12:34
whiteknight not in a normal location? 12:38
coke_ I'm coming in through feather instead of my home box. 13:00
13:08 bubaflub joined 13:22 mikehh joined 13:39 mikehh_ joined 13:43 mikehh left
dalek tracwiki: v19 | fperrad++ | ListOfPlatforms 13:49
tracwiki: trac.parrot.org/parrot/wiki/ListOfP...ction=diff
14:16 woosley left 14:22 jjore left 14:37 jjore joined 14:39 plobsing joined 14:41 bluescreen joined 15:05 mj41 left 15:11 autark left 15:16 autark joined 15:18 dmalcolm joined
dalek rrot: 4981ede | fperrad++ | t/ (4 files):
[t] fix shebang (when parrot is not yet installed)
 15:20
cotto ~~
dukeleto cotto: mornin' 15:23
15:25 davidfetter joined, birdwindupbird joined
whiteknight good morning, dukeleto and cotto 15:32
15:35 theory joined 15:38 davidfetter left 15:43 bluescreen left 15:48 davidfetter joined 15:56 hercynium joined 16:00 benabik left 16:06 Andy joined 16:28 birdwindupbird left 16:31 benabik joined 16:32 contingencyplan joined
benabik ~~ 16:42
whiteknight hello benabik 16:48
benabik 'lo whiteknight!
Anything exciting happening?
whiteknight not a thing, no. Deadlines looming at work, so I'm stressing about that
benabik Ditto for me with s/work/school 16:49
lucian AAAAAHHH!
lucian really hates his uni work
whiteknight lucian: why? What's to hate? 16:51
lucian whiteknight: stupid deadlines, stupid requirements 16:52
i likely complain too much too
dalek tracwiki: v150 | fperrad++ | Languages 17:01
tracwiki: update with 3.3.0
tracwiki: trac.parrot.org/parrot/wiki/Languag...ction=diff
17:05 darbelo joined
dukeleto likes to complain as well 17:07
Why is it that wifi cards choose the day before travel to freak out? 17:08
whiteknight and what's the deal with airplane food?
dukeleto whiteknight: very small profit margins. Airlines lose money on food and coach class, barely break a profit with first class 17:09
whiteknight dukeleto: so that sounds to me like an argument to stop having coach class and food 17:10
dukeleto whiteknight: they require the law of averages, though. just first class doesn't work 17:14
whiteknight they make decent money on business class I've heard tell 17:15
I don't see any reason why not. Fewer flights, fewer seats, each seat worth more money 17:19
somebody could make it work
17:24 [hercynium] joined 17:26 hercynium left
arnsholt IIRC airlines are required to feed you on flights longer than some amount of time in Europe 17:28
17:29 [hercynium] is now known as hercynium
PerlJam Southwest airlines continues to be profitable. 17:31
dalek tracwiki: v29 | allison++ | ChrootSetup 17:32
tracwiki: trac.parrot.org/parrot/wiki/ChrootS...ction=diff
benabik I think airlines have to feed you on long flights in the US too... But they're not required to feed you _good_ food. 17:33
arnsholt Oh, certainly not over here either. The food's horrid, obviously 17:34
dukeleto found some failures on master: smolder.parrot.org/app/projects/rep...ails/15803
plobsing: turns out some of those errors I told you about on your branch are also on master: smolder.parrot.org/app/projects/rep...ails/15803
that is a "plain parrot", with everything turned off 17:35
lucian i'm not convinced it's horrid because they don't make money on it
it's more hit&miss
i've had excellent food on planes a few times
and on the flight back, same food & caterer & airline, it was crap
i think it has something to do with hydration/heating 17:36
dukeleto how are the gsoc students doing? 17:38
anybody ripping into code yet?
or are finals still looming?
lucian personally, i have *no* time for anything but uni 17:39
likely until 13th may
then there's exams, but those aren't a proble
benabik dukeleto: School until the 20th for me. 17:42
bubaflub school for me till the 13th; i walk on the 17th 17:46
17:52 rohit_nsit08 joined
rohit_nsit08 hello #parrot 17:54
rohitnsit08.blogspot.com/2011/04/ca...ments.html 17:55
17:55 ShaneC left 17:57 bluescreen joined
Andy Why do we only have one builder on taptinder.org 17:57
Do we need help there? 17:58
17:59 bluescreen left 18:01 whiteknight left
plobsing dukeleto: is that failure still happening? 18:02
dalek tracwiki: v30 | allison++ | ChrootSetup 18:04
tracwiki: trac.parrot.org/parrot/wiki/ChrootS...ction=diff
plobsing dukeleto: ah, I see. If you configure with both --without-libffi and --without-extra-nci-thunks, there is no framebuilder. I'll file this one under NotABug 18:07
18:10 ShaneC joined
rohit_nsit08 Coke:ping 18:17
darbelo tcurtis: ping 18:21
18:27 ShaneC left 18:44 whiteknight joined 18:48 baest joined 18:52 benabik left 18:55 benabik joined 18:56 benabik left 19:06 benabik joined 19:08 hercynium left 19:25 rohit_nsit08 left
dukeleto plobsing: shoudln't the tests skip themselves properly? 19:26
19:27 benabik left
allison where are deprecated items tracked now? 19:31
google isn't helping much
cotto api.yaml
allison cotto: thanks
cotto allison, what's getting deprecated? 19:32
np
allison cotto: not adding anything, just trying to explain it to the newbie debian packagers
cotto ah
dukeleto allison: please let us know if more docs about api.yaml are needed. all the deprecation docs should mention api.yaml instead of DEPRECATED.pod
allison dukeleto: well, there's a lot of entries in NEWS that still refer to it, but no way around that 19:33
dukeleto: also, the cygwin ports files are still using it
dukeleto: not sure where it should be documented, maybe in some general introduction for new parrot developers? 19:34
dukeleto allison: yes, our NewParrotDeveloperGuide should mention it 19:36
allison dukeleto: where is that in the repository/tarball?
dukeleto allison: there are some wiki pages that describe how to deprectate 19:37
allison: trac.parrot.org/parrot/wiki/NewParr...loperGuide
allison dukeleto: should be something in the actual source code
dukeleto: (additional, or pointers to the wiki docs, or something)
Andy allison! Wow, I remember an "allison". :-)
allison hi Andy!
dukeleto allison: trac.parrot.org/parrot/wiki/HowToDeprecate 19:38
allison Andy: rumors of my death have been greatly exaggerated :)
Andy: though, I was in a nasty car accident recently
dukeleto allison: we have a general problem of some useful docs being only in the repo, or only on the wiki
Andy allison: :-( Forgot what side of the road to drive on? 19:39
19:39 rohit_nsit08 joined
dukeleto allison: sorry to hear that 19:39
Andy "If it's tuesday, it must be right-hand drive"
allison Andy: heh. :) Steering in the old car I was borrowing gave out while I was driving at 50mph on the I-5 bridge into Portland
Andy ouch
allison Andy: surprisingly, I escaped with a busted-up nose, and a few fractured ribs. Really not bad at all, considering. 19:40
Andy You were seat-belted I trust?
dukeleto allison: wow.
cotto allison, glad you're still hacking
allison Andy: Aye, seatbelt responsible for the fractured ribs. (Much better than the alternative.) 19:41
cotto: thanks
cotto: was a bit of a zombie for a couple of weeks, but getting back into the swing of things now 19:42
cotto: and with the Ubuntu release yesterday, I have time for a little outside hacking :)
cotto indeed
allison cotto: hoping to get the Parrot 3.3 Debian packages up this weekend 19:43
dukeleto allison++
cotto allison++
dukeleto allison: cotto++ and I will be at LinuxFestNW this weekend
Andy HOLY CRAP
ALLISON
I FORGOT YOU WERE AN UBUNTUER 19:44
I need to pick your branes!
dukeleto allison: and hopefully an M0 minihackathon will occur
19:44 benabik joined
allison dukeleto: excellent! I'd come up but I'll be on my way to Budapest for the Ubuntu Developer Summit 19:45
dukeleto: happy hackathoning!
Andy: sure, Ubuntu brain available for query at any time :) 19:46
dukeleto allison: awesome. indeed we will.
Andy chromatic and I arestarting a little project for open source projects
cotto Andy, do tell 19:47
Andy and we could probably use existing Ubuntu giant shoulders.
Guide for sustainable projects.
cotto If it weren't for his blog, I'd think chromatic had fallen off the edge of the world.
That sounds like something he'd do.
Andy Human-centered.
cotto looking forward to it
Andy Code is only 2/3rds of your project. The other 2/3rds is the community around it. 19:48
etc etc etc
Trying to make a canonical aggregation of these sorts of ideas 19:49
(not Canonical)
Deliverables that people can adopt.
like a license. 19:50
whiteknight Andy: sounds awesome 19:53
Andy Thank you.
Got some free time? :-)
whiteknight yeah, I can pull some of that out of thin air
Andy I'm serious. 19:54
None of this is tough.
Right now I'm working on basic underlying principles.
ala the agile manifesto
@people > @code
if your project is to be sustainable 19:55
dukeleto This article talks about Mozilla's new JIT compiler 'IonMonkey' blog.mozilla.com/dmandelin/2011/04/...ript-2011/
cotto They love their jits. 19:56
whiteknight we're going to love ours too 19:57
Andy: Sure, I would love to be involved in that kind of thing, even if just as a well-wisher 19:58
my pile of parrot-related TODO items never gets any smaller
Andy Well, I'm sure you follow perlbuz, right?
whiteknight nosir
Andy I'm certianly going to make noise there.
OK, @petdance or @perlbuzz then?
tadzik did you mean: make buzz?
whiteknight Andy: I'll start following now 19:59
brb
19:59 whiteknight left
tcurtis darbelo: pong 19:59
20:01 jrt4__ joined
Tene Andy: that sounds fascinating. 20:04
20:08 darbelo_ joined 20:13 rohit_nsit08 left 20:14 darbelo left
coke_ goes for "straightman" on the mailing list. 20:17
taptinder - I probably forgot to restart my client after a reboot. will get on that. 20:18
coke_ tries to upgrade to ubuntu 11.04 and cringes. 20:19
(mainly because of the netbook+ubuntu combination of going to sleep and not waking up, not ubuntu perse.) 20:20
20:24 rohit_nsit08 joined 20:26 particle1 joined 20:28 particle left
cotto coke++ 20:32
20:33 particle joined 20:36 particle1 left
tadzik theory: ping 20:39
theory tadzik: D'oh! Forgot to ping you this morning. Apologies. 20:40
20:41 hudnix left
theory tadzik: Just finishing up a blog post in the PGXN API. 20:41
tadzik theory: I almost thought you were still sleeping, I started feeling jealous :>
theory sorry
no, it's 1:41 pm here.
tadzik theory: oh, I definitely want to read that. The META.info design is rather accidental 20:42
theory: no worries
theory tadzik: I've written about that in a few places. Mostly just stole from the PGXN Meta Spec 2.0. 20:43
tadzik: Just posted: blog.pgxn.org/post/5049235040/about-pgxn-api 20:44
Anyway, the META spec is here: pgxn.org/spec
And this is about PGXN::Manager, which is basically like PAUSE. blog.pgxn.org/post/4854707157/pgxn-manager 20:45
I wrote it all so that it needn't be specific to PostgreSQL, so in theory you could use stuff if you want. Although it's all Perl 5. :-)
Gonna go release a bunch of stuff to CPAN now. 20:46
cotto shiny new laptop just shipped 20:53
cotto looks forward to sub-hour rakudo spectest runs 20:54
dukeleto, privmsg pong 20:55
20:55 hudnix joined 20:57 hercynium joined 21:13 hercynium left 21:15 Coke left, Coke joined 21:22 rohit_nsit08 left
darbelo_ tcurtis: reping 21:25
21:31 ambs left 21:35 mikehh_ is now known as mikehh
mikehh opbots, names 21:35
tcurtis darbelo_: repong 21:37
21:42 rohit_nsit08 joined
darbelo_ tcurtis: Hey. 21:45
Missed you earlier.
Got distracted. 21:46
I have exams looming over me for next week. So I'll be somewhat hard to reach in the weekend. 21:47
tcurtis: Do you have time for a quick meeting now?
tcurtis darbelo_: sure. 21:48
darbelo_ Excelent. 21:49
So, what's your status?
Are you free to start the bonding process or do you need time to attend to non-GSoC things? 21:51
I recall some of the students were busy with schoolwork, but I can´t recall which :)
tcurtis I haven't done any work yet, but I'm starting to read some of the relevant papers.
darbelo_ Excellent. Do you have pointers so I can follow along at home? 21:52
tcurtis I have schoolwork, but it's not blocking GSoC. 21:53
The dissertation which introduced LALR(1) parsing can be found at computer-refuge.org/bitsavers/pdf/m...TR-065.pdf 21:55
I feel like I found a digital copy of "Efficient Computation of LALR(1) Look-Ahead Sets", but I can't seem to find it now. 21:58
I plan to read "Practical LR(k) Translators" (the linked dissertation) in the coming week. 21:59
I'll also probably write a little introductory blog-post this weekend. 22:00
benabik tcurtis: Efficient Computation - www.win.tue.nl/~wsinswan/softwareto...nnello.pdf
tcurtis benabik++ 22:01
benabik has many parser links around from a compiler course last quarter. 22:02
davidfetter tcurtis, generally, i've had pretty good luck with plugging the name of the paper (all in quotes) and pdf into search engines 22:04
well, good luck *finding* the paper. understanding it is a whole 'nother kettle of fish ;)
benabik I think davidfetter's method is how I found them the first time around. :-D 22:05
davidfetter :)
tcurtis ah, yes. Adding pdf is a good idea. Thanks, davidfetter++ 22:06
davidfetter happy to help :)
dukeleto ~~
davidfetter hai dukeleto 22:07
dukeleto waves hello from the train to LinuxFestNW
cotto: ping
tadzik o/
dukeleto davidfetter: werd up 22:08
darbelo_ tcurtis: Ok. I'll look into that. 22:09
cotto dukeleto, pong
davidfetter ooh, lfnw. i keep meaning to apply to that one
cotto davidfetter, might be a bit late
davidfetter for this year, yes 22:10
dukeleto cotto: you planning on being in Bellingham tonight?
cotto they're a little last-minute, but not *that* last-minute
dukeleto, possibly. I woke up possibly sick. If I feel fine for the rest of the day, yes.
22:11 whiteknight joined
dukeleto cotto: oh noes! Vitamin C, my friend. And lots of B and D too 22:11
cotto At worst it'll be an annoyance. 22:12
davidfetter was in terrible shape for his talk at jdcon east 22:13
...which nevertheless went ok :)
whiteknight good afternoon, #parrot 22:15
darbelo_ tcurtis: I almost forgot. Do you have any plans for the 'project infrastructure'?
Repo, blog, issue tacker, build tools, etc. 22:16
tcurtis darbelo_: github for the repo and issue tracker. probably distutils for the build tracker. Probably rosella/test for testing. 22:18
22:18 lucian left
tcurtis For blog, during GSoC, I'll use Parrot.org, but I'll probably either mirror it on the github pages or a blog of my own (which I'll need to get around to setting up). 22:18
dukeleto tcurtis: sounds good. the parrot.org solution is meant to minimize the needs to configure+setup other blogs 22:21
tcurtis: mirroring it anywhere else is encouraged
cotto dukeleto, are we calling that a hard requirement, or is it fine if there's a regularly-updated blog somewhere?
22:24 Andy left 22:25 sjn joined
darbelo_ tcurtis: Nice. And, last thing. What's your typical work environment going to be like. 22:26
22:26 dmalcolm left
darbelo_ OS, editors, etc 22:26
dukeleto cotto: i would like to have all the students blog posts appear on parrot.org, because then people can go to one place to see GSoC stuf 22:27
cotto: if that is a problem, we can work something out, but I am making it a hard requirement unless somebody has a good reason
22:27 GodFather joined
whiteknight cotto: hard requirement 22:27
cotto dukeleto and whiteknight: ok
darbelo_ Ideally I'd like to set up one as close to it as possible. To make sure I can track your work without worrying about environmental mismatches.
whiteknight seen cgaertnr 22:28
aloha Sorry, I haven't seen cgaertnr.
whiteknight seen cgaertner
aloha cgaertner was last seen in #parrot 27 days 8 hours ago saying "btw, who would be willing to mentor the project? the proposal template ask for that information, I I don't remeber anyone actually mentioning that...".
darbelo_ And one as different from it as possible. So I can provide you with a canary in the metaphoric coal mine.
tcurtis darbelo_: Mostly GNU/Linux(Debian squeeze, specifically) with either vim or emacs. I'll probably also work on it on OS X occasionally.
darbelo_ tcurtis: Ok. I have Linux and OpenBSD systems at home, and a MacBook at work. I can track your work on all of them, provide as wide OS coverage as you need. 22:32
I'll try to read up on those papers, so we can discuss some more technical details next time. 22:35
tcurtis: Ok. Last question. Whats your time zone / regular working hourts ?
tcurtis darbelo_: my time zone is US central time. I'm not sure when I'll usually be working on GSoC. Probably mostly after 3pm on weekdays. But that will change once the quarter end (June 10th or so). 22:38
darbelo_ US Central is GMT-? 22:40
I can never keep US timezones straight.
tcurtis UTC-5 currently and for the duration of the summer.
darbelo_ The seasonal rearrangement doesn't help
tcurtis darbelo_, dukeleto, whiteknight: I should inform you that I've committed to working 20-hours a week for a part-time job for this summer. I would have mentioned it on my proposal had I known at the time. (more) 22:42
I don't think it will interfere with GSoC, particularly since it involves a lot of downtime in which I will probably end up working on GSoC.
whiteknight ZOMG UNACEPTABLE 22:43
where are you working?
dukeleto tcurtis: as long as you repeat your past performance or better, it should be fine :)
tcurtis whiteknight: at the computer lab at my university.
whiteknight preferrably better
oh, that's not bad
open up a window in the background and hack 22:44
plus if you need help with any of the theory, you can ask a professor
davidfetter theory's a pretty helpful sort
theory I tr 22:45
try
GodFather newbe, need pir example using postgres 22:47
whiteknight GodFather: welcome! 22:49
GodFather: dukeleto is your man, I think
dukeleto GodFather: howdy
GodFather: what are you in need of?
GodFather howdy, I'm so new smell like a baby, I'd like a little example doing a select so I can build upon that code 22:50
select from a Postgres database
darbelo_ tcurtis: Gotta go now. msg me if you need anything. 22:52
22:55 darbelo_ left
GodFather I'm trying to get the code on www.sidhe.org/~dan/blog/archives/000253.html to run 22:55
dukeleto GodFather: give me a little more context. You want to use the postgres PIR libraries to run a select on a table in a database from PIR?
whiteknight GodFather: that's an extremely old post
GodFather OK, is there some other post that would be better for me? 22:56
dukeleto GodFather: which version of parrot are you using?
GodFather one min
dukeleto GodFather: parrot --version should tell you 22:57
GodFather did a git on version 3.3.0
3.3.0-devel
dukeleto GodFather: ok 22:58
GodFather: github.com/parrot/parrot/blob/mast...brary/pg.t
GodFather: if I understand you, that should send you in the correct direction
GodFather: look at line 103 22:59
GodFather ok, thanks will look at it
dukeleto goes back to writing his conf talk
23:09 bacek joined 23:37 rohit_nsit08 left 23:43 fperrad left

Powered by Cro and the Raku® Programming Language.

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

Thank you!