Parrot 2.6.0 | parrot.org Log: irclog.perlgeek.de/parrot/today | Nopaste: nopaste.snit.ch:8001 | fix 'make html' (talk to Coke), merge gc_* branches, fix/replace/optimize hashing
Set by moderator on 3 August 2010.
00:04 particle joined 00:26 petdance joined
cotto_work chromatic: is the CPS you envision for Lorito something like the Lorito equivalent of this: nopaste.snit.ch/22586 00:39
chromatic Sort of. 00:44
00:51 cotto_work2 joined
kid51 Smolder broken: 4th night in a row 00:51
gsoc_nfg branch PASS make test
cotto_work2 There's too much hand-waving in the example you nopasted a couple days ago. I'm not clear how it'd translate into M0 ops. 00:53
00:53 dngor_ joined 00:56 workbench joined
dalek tracwiki: v1 | cotto++ | LoritoCPS 01:05
tracwiki: trac.parrot.org/parrot/wiki/LoritoC...ction=diff
01:10 plobsing joined
particle cotto: search for "typed assembly language continuation passing style" for some papers that may give you ideas 01:14
eg: www.scribd.com/doc/7554353/A-Multit...y-Language and www.cs.cornell.edu/talc/papers/stal-tic.pdf
cotto thanks
particle www.cs.cornell.edu/talc/ for more 01:15
cotto Mmmm. Science.
01:22 tcurtis joined
particle cotto: if you're an acm member... delivery.acm.org/10.1145/80000/7530...N=58554080 01:35
01:36 bacek_at_work joined
GeJ clock ? 01:50
purl GeJ: LAX: Thu 6:50pm PDT / CHI: Thu 8:50pm CDT / NYC: Thu 9:50pm EDT / LON: Fri 2:50am BST / BER: Fri 3:50am CEST / IND: Fri 7:20am IST / TOK: Fri 10:50am JST / SYD: Fri 11:50am EST /
GeJ Bonjour everyone.
01:53 dngor joined 01:54 Casan joined
kid51 Bonjour, Geraud 01:54
and bacek_at_work
purl bacek_at_work is bacek_without_scotch
kid51 Hey, purl is back! 01:55
purl, where were you?
purl kid51: wish i knew
kid51 purl, have you heard from nopaste lately?
purl kid51: i haven't a clue
GeJ Good evening James.
sorear purl, status 01:56
purl Since Thu Aug 5 20:55:17 2010, there have been 20 modifications and 29 questions. I have been awake for 5 hours, 1 minute, 5 seconds this session, and currently reference 852960 factoids. Addressing is in optional mode.
01:57 workbench joined
kid51 wanders off to read political philosophy 02:01
atrodo cotto> I got to thinking about the set op, and I don't think we need it. consider, coerce_i_i 02:14
02:28 workbench joined 02:37 theory joined 02:47 dngor joined 02:54 janus joined 02:57 ash_ joined
Coke cotto: I made it u. 03:29
*up.
atrodo: we dpm 03:30
atrodo: we don't need goto, either. :|
03:47 tcurtis_ joined 03:53 LoganLK joined 03:58 mikehh joined 05:18 petdance joined 05:23 theory joined
plobsing msg chromatic: I've implented/half-debugged lazy consttable loading, but it breaks cases where unpack has side effects (eg: loading subs, classes, and methods into the appropriate symbolic slots). 05:28
purl Message for chromatic stored.
05:35 theory joined
dalek kudo: 9dd4081 | pmichaud++ | (2 files):
Add somewhat cheating version of src/core/Order.pm .
 05:38
kudo: 6d32585 | pmichaud++ | src/core/Int.pm:
Add more Int numeric comparision operators.
05:49 brianwisti joined
cotto atrodo, coerce_i_i would be spelled set_i_i. 05:59
Giving them different names is just for clarity. 06:01
Coke++ for Rakudo Club 06:05
chromatic msg plobsing Can those turn into lazy thunks? 06:27
purl Message for plobsing stored.
07:03 robin-gvx joined
dalek kudo: 2c0ea31 | pmichaud++ | src/c (2 files):
Refactor Order::* constants--- as cheats we want them available before

back into the setting itself.		 07:05
kudo: 035efb0 | pmichaud++ | src/core/ (2 files):
Refactor some comparison operators to avoid unnecessary blocks and
kudo: 0e5edb5 | pmichaud++ | src/core/Int.pm:
Clean up &infix:<cmp>(Int, Int).
 07:52
07:52 jjore joined 07:53 fperrad joined 07:55 NOTevil left
TiMBuS pmichaud, yesterday you told me to wrap everything in a perl6scalar. doooes new ['perl6scalar'], some_other_p6scalar work? or do i need to descalarref it first 07:56
...
wrong channel 07:57
well. actually its pir so it counts!
pmichaud TiMBuS: (answering on #perl6)
07:57 eternaleye joined 09:19 clinton joined 09:34 AndyA joined 10:11 muixirt joined
muixirt ping fperrad 10:12
is lua supposed to run the test suite without errors? 10:30
10:39 lucian joined 11:34 nopaste joined 11:53 whiteknight joined 13:21 Paul_the_Greek joined
Paul_the_Greek Good morning. 13:22
purl For you maybe.
Paul_the_Greek Good morning.
purl For you maybe.
Paul_the_Greek Lovely.
Is this a good ticket to take: trac.parrot.org/parrot/ticket/1402
whiteknight good morning, Paul_the_Greek 13:25
Paul_the_Greek Hey, whiteknight.
purl whiteknight is, like, mailto:wknight8111@gmail.com or the grand master funk or wknight8111.blogspot.com/
whiteknight Paul_the_Greek: yes, that's a very good ticket for a new developer 13:26
Paul_the_Greek I'll take it.
So bear with me on more SVN questions.
whiteknight no problem
Paul_the_Greek I've submitted a patch file. It's waiting in the queue.
Now I want to make additional changes to one of those files.
That's a real pain, it seems.
If I'm sure I'll get the new patch in before the first one is applied, I can just replace that patch. 13:27
But otherwise I'm sort of stuck.
Should I wait until the pending patches are applied? 13:28
whiteknight post a note saying you don't want it to be applied and that you are working on a new one 13:30
Paul_the_Greek But the new one is really unrelated, it just happens to affect one of the files I've already touched. 13:31
The new patch would be related to a different ticket.
Are the pending patches applied periodically, or not until it's time to build a new version? 13:32
13:35 macroron joined, ruoso joined
Coke Paul_the_Greek: there is no schedule. 13:38
depends on committers tuits. (free time) 13:39
13:39 erikh left
Coke you are doing the right thing and being vocal-but-not-annoying. =-) 13:39
13:45 clinton joined 13:47 patspam joined
Paul_the_Greek Okay, so I'll just have to sort out any overlapping patches. 13:51
This will prompt me to work on different parts of the system so I don't have overlap. :D 13:52
Coke if I'm using svn and desperate, I'll just have multiple checkouts.
there was a question on one of your patches, btw.
not sure if you got the email.
Paul_the_Greek Ah, let me check ...
Would that have been sent to the list, or added to the ticket? 13:53
13:56 khai joined
Paul_the_Greek Ah, I see it added to the ticket. 13:56
Coke it was sent to ... parrot-tickets? 14:00
I am not a huge fan of our setup there, but it's... functional.
Paul_the_Greek Weird. I've been getting my ticket ownership emails, but I didn't get anything about Andy's post. 14:04
Does every change to a ticket I own result in an email to me?
Yes, it does, because I just got an email with my own post. 14:05
particle Paul_the_Greek: if your patches depend on each other, make sure the tickets have a dependent relationship 14:18
if they don't depend on each other (same files, but different parts) then no problem
Coke particle: trac doesn't have "dependent relationships" between tickets. 14:27
best you can do is say "See #12345" in a comment.
trac--
particle well, we haven't added the field
us--
we haven't yet replicated all the functionality we had in rt 14:28
Coke particle: so telling newbies to use it doesn't help. =-)
particle me--
Coke particle: yes, because <RANT ELIDED>
particle i didn't realize that.
14:29 brianwisti joined
particle Paul_the_Greek: i had a neighbor that had a tough time learning to spell his name in school, "daniel salvatoriello". he had nothing on you. 14:29
and, for now, yes, a comment in the ticket (or better both related tickets) will suffice. 14:30
Paul_the_Greek The tickets are unrelated; just happen to affect the same file. 14:56
Is there some problem spelling Paul Constantine Anagnostopoulos?
Just consider me a good database test. 14:57
15:05 sjn joined
particle :) 15:30
15:32 jsut joined 15:44 brianwisti joined 15:52 desertm4x joined 15:54 brianwisti left 15:58 Paul_the_Greek joined 16:01 brianwisti joined 16:23 theory joined 16:56 jan joined 17:10 Paul_the_Greek joined
Paul_the_Greek I propose a new C coding convention, although it's a bit too late now. 17:10
particle what is it? 17:16
purl it's it!
particle hey, purl-- is back!
Coke the c coding conventions are always fungible. we have teams of robots to fix them up when broken. 17:17
Paul_the_Greek In a C function declaration or definition, put a space between the function name and the parameter list. 17:21
Then you can search for all the calls without finding the declaration or vice versa. 17:22
Coke Paul_the_Greek: just use the tags. 17:25
do you use vi?
Paul_the_Greek No, I use Epsilon. I never bother to tag anything. Epsilon can search all of Parrot in about 1.5 seconds.
I suppose I should get with the program.
Coke I've never heard of epsilon - I know with vi, I can do "make tags-vi", then "vi -t foo" and have it go to the right place for foo. 17:26
that said, I see no real problem with your suggestion
Paul_the_Greek Epsilon is another Emacs-like editor. I've used it for about 20 years. 17:27
I like to do the search because then I can study the grep buffer to see where the identifier occurs.
But tags are good and I should use them more often. 17:28
cotto_work2 ~~ 17:29
Coke I think emacs-tags might also exist.
Paul_the_Greek Oof, getting rid of the mem_internal_* macros is going to have me sticking my fingers all over the place.
Oh yes, Epsilon has tags. I just rarely use them.
Coke Paul_the_Greek: didn't that suggest eliminating a /duplicate/ macro? (so just use the other one?)
Paul_the_Greek It's eliminating a set of wrapper macros in favor of a set of functions. 17:30
The macros generate calls to another set of functions that want to be eliminated.
For example, mem_internal_allocate appears in 9 files. 17:31
One possibility is to change the macros to call the desired set of functions, but leave the macro calls alone. 17:32
That would be a first step that would reduce the number of files touched.
particle Paul_the_Greek: that's a reasonable first step, especially for testing 17:33
Paul_the_Greek Yup, that's what I'll do. It still allows us to eliminate the other set of functions that the macros currently use. 17:34
17:48 [1]Casan joined 17:55 Coke joined
Coke we need to fix parrotbug and docs to mention that you can create new tickets by emailing to tickets@parrot.org 18:19
(so you don't have to login & register)
brianwisti ohhh. Yeah, that would be nice information to have. Then again, hiding that info is a good gateway path to sucker folks into helping out. 18:27
Coke (we also want to obfuscate the ticket address.) 18:29
cotto_work There's charm in that approach but we'll eventually need something to weed out spam. 18:36
all it takes is one person mentioning the address in a publicly logged forum (e.g. here) and the spambots will get ahold of it.
18:37 LoganLK joined
Austin Whoa. 18:38
I might have tricked parrot into building on windows...
18:40 jsut_ joined
Coke Austin: it does that all the time. 18:43
Austin This would be a first for me.
Coke (one of the rakudo devs mainly uses windows)
cotto_work: it's already out there and mentioned on the parrot-dev mailing list. 18:44
18:44 somebody_ joined
Austin I suspect it's the old spaces-in-directory-names problem. 18:44
Coke ... and I just mentioned it here, and it's already working, so here's hoping the spam prevention is already there.
Austin: ah. there is a simple fix for that. =-) 18:45
s/fix/workaround/
Austin Yeah, stop developing on parrot.
But I'm trying not to do that...
Oh, well. A bunch of test failures... 18:51
18:56 somebody__ joined
Coke Austin: I was going to say, just build in C:\\parrot. 19:03
19:12 LoganLK joined 19:18 davidfetter joined 19:39 theory joined 19:59 shawnhcorey joined 20:03 lucian_ joined
Coke adds that bug to his list of things to check out after $DAYJOBLIGATIONS 20:04
20:04 simcop238 joined 20:07 aloha joined 20:12 Paul_the_Greek joined 20:13 ash_ joined 20:21 tcurtis joined 20:28 aloha joined
cotto_work aloha, aloha 20:40
davidfetter mahalo 20:42
Paul_the_Greek Hey ho. 20:47
Coke ohana. 20:50
dalek rrot: r48333 | chromatic++ | trunk/src/pmc/callcontext.pmc:
[PMC] Simplified code slightly.
 21:07
rrot: r48334 | chromatic++ | trunk/include/parrot/hash.h:
[hash] Use all allocated hash buckets, not 75%.

reallocates hashes less frequently, which is good for speed. Here's a 2.489% performance improvement for Rakudo startup.
21:12 darbelo joined
cotto_work msg chromatic It looks like some unrelated changes leaked in with r48334. 21:13
purl Message for chromatic stored.
21:16 elmex joined
Coke ... we never used the last 1/4 of the slots? insanity. 21:19
THIS! IS! PARROT!
cotto_work The idea is that collisions are increasingly likely as the hash becomes fuller. Clearly the higher memory efficiency outweighed the cost of the collisions. 21:21
Coke I'm sure that idea was never benchmarked.
... certainly not with rakudo-size code.
cotto_work I'd be really surprised.
Coke chromatic++ 21:22
dalek rrot: r48335 | darbelo++ | branches/unshared_buffers/src/string/encoding (4 files):
Disable shared buffers in substring operations.
 21:24
rrot: r48336 | darbelo++ | branches/unshared_buffers/src/string/charset (2 files):
Disable shared buffers in no-op charset compose and decompose operations.
rrot: r48337 | darbelo++ | branches/unshared_buffers/src/string/api.c:
Remove shared buffer usage from the string API functions.
rrot: r48338 | darbelo++ | branches/unshared_buffers/src/string/api.c:
Redirect Parrot_str_copy to Parrot_str_clone
rrot: r48339 | chromatic++ | trunk/include/parrot/hash.h:
[hash] Reverted unrelated changes from r48334.
Coke ugh. do an svn update to get c's changes. run 'make install'. config doesn't know about the new svn revision, even though the code changed, so the installed version thinks it's older, so things that check the version die... 21:25
cotto_work A Configure.pl step isn't the ideal place to store that information. 21:27
Coke there's no good place, I fear. 21:28
cotto_work khairul++ 21:37
Paul_the_Greek Is this the sort of thing that is prompting the switch to git? 21:44
tcurtis Folks, I recommend finding a hiding place. dalek is about to be quite noisy.
Paul_the_Greek When you try to merge all these branches plus all the pending patches, do you need to consume lots of alcohol to remain sane? 21:46
tcurtis thought dalek was about to be noisy. 21:48
dalek ee-optimization: 826af57 | tcurtis++ | src/Tree/Optimizer (2 files):
Make passes dependency-aware.
 21:50
ee-optimization: e36f8c9 | tcurtis++ | t/tree-optimizer.t:
Add some tests for Tree::Optimizer.register with a Tree::Optimizer::Pass as
purl This is abuse.
ee-optimization: 231e920 | tcurtis++ | src/Tree/Optimizer.nqp:
Make Tree::Optimizer::Pass options to Tree::Optimizer.register work.
ee-optimization: eb54e94 | tcurtis++ | src/Tree/Optimizer/Pass.nqp:
Add .clone method to Tree::Optimizer::Pass and ensure @!dependencies is always
ee-optimization: 25f33c2 | tcurtis++ | t/tree-optimizer.t:
Add tests for adding dependencies when using a pass directly as argument to
ee-optimization: 4d67c12 | tcurtis++ | src/Tree/Optimizer.nqp:
Make the new :depends-on tests pass.
ee-optimization: ce84c71 | tcurtis++ | t/tree-optimizer.t:
Add test for changing the name of an existing pass when .registering it.
ee-optimization: f253b31 | tcurtis++ | src/Tree/Optimizer.nqp:
Implement renaming a pass when .registering it.
21:56 kj joined 22:00 aloha joined 22:08 arnsholt joined
cotto_work smolder? 22:25
purl hmmm... smolder is sourceforge.net/projects/smolder or web-based smoke test aggregator used by developers and testers to upload (automated or manually) and view smoke/regression tests using the Test Anything Protocol (TAP). or smolder.plusthree.com/app/public_pr..._reports/8
dalek TT #1733 created by fperrad++: calling convention broken 22:33
TT #1733: trac.parrot.org/parrot/ticket/1733
22:37 kj_ joined 22:53 ruoso joined 22:58 dafrito joined
dafrito Sorry to bother, but is Parrot's smoke test site down? I'm using smolder.plusthree.com/app/public_pr.../details/8 for the link 23:01
23:02 hercynium joined
cotto_work yup 23:02
dafrito okay, thanks
23:38 kid51 joined
kid51 smolder down 5th night in a row :-( 23:38
mikehh kid51: I contacted mpeters and he said: Sorry, I'm having some problems with that machine. I'm going to have to move smolder some place else. Please bear with me as I find the time to do that. 23:49
kid51: particle mentioned trying to move it to our server 23:50
or at least install it there 23:51
Michael Peters <mpeters@plusthree.com> 23:52
kid51 mikehh: Thanks for the update 23:56
And, yes, mpeters has provided yeoman (yo-person?) service to us for, IIRC, 3 years now -- but it's something we really ought to be running ourselves.
Not that he should no longer be running it -- but that we should be. 23:57
At a later time I would like to talk with you about some TTs, but right now ...
mikehh kid51: I couldn't agree more - but it also provides smoke testing for a lot of other parrot based derivitives 23:58
23:59 Psyche^ joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, test, fulltest) at r48339 - Ubuntu 10.04 amd64 (g++ with --optimize) 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!