Parrot 3.10.0 "Apple Pi" | parrot.org | Log: irclog.perlgeek.de/parrot | #parrotsketch meeting Tuesday 19:30 UTC
Set by moderator on 22 November 2011.
dalek sella/random_1: 3777857 | Whiteknight++ | / (13 files):
Add the random library back in, I don't know how it got deleted
 00:03
sella/random_1: 2b5f93e | Whiteknight++ | t/ (4 files):
fix weird conflicts
sella: 3777857 | Whiteknight++ | / (13 files):
Add the random library back in, I don't know how it got deleted
 00:04
sella: 2b5f93e | Whiteknight++ | t/ (4 files):
fix weird conflicts
sella: e246fdf | Whiteknight++ | / (7 files):
Merge branch 'random_1'
sella: 9835c47 | Whiteknight++ | src/random/randomnumber/MersenneTwister.winxed:
Fix MersenneTwister.get_float()
 00:19
sella: b40470d | Whiteknight++ | src/random/randomnumber/BoxMullerNormal.winxed:
last fix for BoxMuller, all tests pass
website: NotFound++ | Guitor Hero 00:44
website: www.parrot.org/content/guitor-hero
whiteknight has another project to watch on github now 01:00
thanks alot, NotFound
dukeleto msg NotFound don't forget to add a post-receive-url to Guitor to we get irc notifications :)
aloha OK. I'll deliver the message.
whiteknight dukeleto: I have a link you might find very interesting: www.scottaaronson.com/blog/?p=839 01:12
actually, I'm finding it very interesting
01:13 woosley joined
NotFound The problem with Xlib and NCI is the error callbacks. I'm thinking about adding a new callback mechanism into parrot, better than requiring bits of c for any library with the same issue. 01:17
whiteknight our callbacks have always been lacking 01:19
NotFound (without setting an error callback, any Xlib program can die for trivial issues)
cotto NotFound: please do. You're guaranteed not to be the last person who'll want a better callback mechanism. 01:22
01:39 davidfetter joined
dukeleto NotFound: why Xlib and not something more sane? 01:39
dalek sella: d2fa7e3 | Whiteknight++ | t/commandline/ (3 files):
A few commandline test updates
sella: 6345a20 | Whiteknight++ | / (9 files):
CommandLine is stable
dukeleto whiteknight: hilarious 01:42
whiteknight: poor mathematicians slaving over the third decimal in an exponent 01:43
whiteknight dukeleto: what would be far more important would be an algorithm specifically tailored to vectorized instruction sets and parallelized execution cores 01:44
cotto Is that an interesting result for people who care? 01:45
dukeleto cotto: for the people that care, yes. Basically, whoever can prove the lowest bound on any CS algorithm is the cool kid on the block 01:46
cotto: but in practice, none of it really matters. the matrices that would be effected by those results have more rows/cols than there are atoms in the universe
cotto fsvo "cool"
ok. I figured it was something on that order of interestingness. 01:47
I'm glad they have something to get excited about.
dukeleto cotto: academics need something to care about, because they surely don't care about reality 01:48
cotto: my advisor told me once: People fight tooth and nail in academia becaues the stakes are so low. 01:50
NotFound dukeleto: something more sane? Like what? 01:56
cotto dukeleto: I guess so.
NotFound We can start up a charity: "More atoms for our universe" 01:58
Our mathematicians have not enough atoms, please support them. 01:59
02:01 benabik joined
dalek Heuristic branch merge: pushed 17 commits to parrot/documentation_revisions by ayardley 02:07
cotto NotFound: sure, but where will you find a donor? 02:09
What if all the people in other universes are jerks?
NotFound cotto: doesn't matter, just let people send money and we'll manage to find something to do with it ;) 02:11
dukeleto NotFound: anything is more sane than Xlib. Cairo? 02:26
dalek sella: 5d198e4 | Whiteknight++ | src/unstable/dumper/ (4 files):
Several tweaks and fixes to the dumper library. Fix the way several things print. Filter out some excess whitespace. Try to call get_string on objects, and include that output
 02:37
cotto ~~ 02:57
dalek kudo/nom: 51838a5 | coke++ | t/spectest.data:
run fudged tests (yay!)
 03:26
alvis evening #parrot 03:31
has anyone tried parrot.github.com/ lately? 03:32
I get a "Page does not exist!" message.
dukeleto alvis: i get the same 03:33
something is borked
cotto alvis, I saw that earlier too
alvis dukeleto & cotto: Any ideas? I haven't changed anything. 03:34
dukeleto hmmm. let me kick the tires
alvis dukeleto: k. 03:35
dalek href="https://parrot.github.com:">parrot.github.com: 2021636 | dukeleto++ | README.pod:
Refactor the readme
03:35 Yuki`N joined
Yuki`N Leaderboards up -- results are clear: 03:35
dukeleto Huh. that didn't fix it. 03:36
Yuki`N I got the message
dukeleto, have you seen the leaderboards?
2 of the top 3 are being mentored by jumpyshoes
And this looks suspiciously similar to jumpyshoes last year: www.google-melange.com/gci/student_...lionaneesh
dukeleto Yuki`N: nope 03:37
alvis: just contacted Github support about parrot.github.com 03:38
Yuki`N: looks like task farming to me.
Yuki`N indeed.
all of the top 3 look to be farming tasks.
alvis dukeleto: Ok, that was my *first* thought. Let me know please.
dukeleto Yuki`N: i guess we care more about quality than quantity here.
Yuki`N I suppose so. 03:39
I took a look at creating parrot-nacl
seems like a bit more than one task, imo.
I mean, it's definitely doable, but I'd have to create the build system from scratch and write all the boilerplate documentation. 03:40
dukeleto Yuki`N: sure. That can be split into different tasks. That task is basically to get it started
Yuki`N: don't worry about the documentation and steal the build from parrot-gmp or parrot-libgit2
Yuki`N: the easy math-parrot-primality tasks are needed to accomplish the hard task you commented on 03:41
Yuki`N: i should have made that clearer
Yuki`N oh
dukeleto Yuki`N: basically, prime_count() has to call is_prime(), and the other 3 tasks that are easier are needed to implement is_prime() 03:42
Yuki`N alright!
I shall do all of them
dukeleto Yuki`N: i think you will have fun! 03:43
Yuki`N I shall first print out that math paper and read it during school tomorrow.
dukeleto Yuki`N: :)
Yuki`N I have to build parrot with gmp to get bigint support, right?
dukeleto Yuki`N: i read the paper a few days ago. Please let me know if anything doesn't make sense.
Yuki`N: yes
Yuki`N: do you want me to approve you for the hard task? Or can you still ask for the other tasks to be assigned?
Yuki`N I'll start with the easier tasks. 03:44
03:44 khisanth_ joined
dukeleto Yuki`N: the is_small_prime() task should only take you 5 minutes 03:44
Yuki`N: the others shouldn't be longer than a few hours unless you hit an odd bug in winxed or something
Yuki`N somehow, I remember looking at this perl library for some tasks last year. 03:45
dukeleto Yuki`N: math--primality was a gsoc project
Yuki`N I feel like I did something with it last year...perhaps implementing RSA in PIR?
dukeleto Yuki`N: it will be very interesting to benchmark these algorithms in Perl 5 vs. Winxed 03:46
Yuki`N: i think you worked on sha256 last year, in PIR, iirc
Yuki`N hmm
well sha doesn't use primality at all.
dukeleto but i could be misremembering
Yuki`N: no, it doesn't :)
Yuki`N I did work on SHA.
dukeleto Yuki`N: do you need me to approve you for the is_small_prime task?
Yuki`N I do remember I used these libraries for something.
Sure.
dukeleto Yuki`N: can you paste the url in here?
Yuki`N www.google-melange.com/gci/task/vie...11/7172291 03:47
dukeleto Yuki`N: done! 03:48
Yuki`N alright
are we namespacing this at all?
Perhaps Math::Primality?
dukeleto Yuki`N: yep
Yuki`N: _is_small_prime is an internal function in the perl 5 module, but i think it is reasonable to make this a public function 03:49
Yuki`N: it speeds things up greatly by filtering out small primes and not doing expensive stuff
Yuki`N: beware, setup.winxed does very little :) 03:50
Yuki`N: extra points if you make it better :)
Yuki`N: but just implementing is_small_prime with 2 trivial tests will be fine.
Yuki`N alright
did you check out the changes I made to the Makefile in libgit?
dalek rrot/documentation_revisions: 50068bb | (Alvis Yardley)++ | docs/TODO_revisions:
Minor edits.
 04:06
Yuki`N dukeleto, I can't figure out a good way to do small prime checking exactly the same way it does in Perl. 04:09
Since perl allows indexes to be specified when initializing an array.
whiteknight, ping 04:33
never mind. 04:35
dukeleto, task submitted for review www.google-melange.com/gci/task/vie...11/7172291 04:47
dalek rrot/documentation_revisions: 79887ff | (Alvis Yardley)++ | / (2 files):
Deleted README.
 04:55
rrot/documentation_revisions: 10fd17f | (Alvis Yardley)++ | README.pod:
Added new README.pod
rrot/documentation_revisions: 15bea4a | (Alvis Yardley)++ | docs/readmes/README (2 files):
Changed [year]-[year] to <creation_year>-<current_year>.
dukeleto Yuki`N: looking 04:56
Yuki`N: nice work! 04:58
Yuki`N: just marked the task as closed, will merge later. send me the url of the next task you want and I will assign
Yuki`N alright
www.google-melange.com/gci/task/vie...11/7174256
dukeleto Yuki`N: and thanks for adding the rosella submodule! 04:59
Yuki`N xD
I did a lot of work above and beyond
dukeleto Yuki`N: assigned! this one should be more fun.
Yuki`N a 1 point task. ;P
dukeleto Yuki`N: that extra work does not go unnoticed :) It might count for extra points in GCI, but that good karma will come back to you in another way. 05:00
s/might count/might not count/
Yuki`N oh dear
dukeleto Yuki`N: :)
Yuki`N I can't access a GMP function from Parrot
It's not in bigint.
dukeleto Yuki`N: you should use parrot-gmp
Yuki`N: github.com/bubaflub/parrot-gmp
Yuki`N: a gsoc project from last year :) 05:01
Yuki`N MORE SUBMODULES?!
dukeleto Yuki`N: you can just assume parrot-gmp is installed for now
Yuki`N: parrot-math-primality will require parrot-gmp to be installed
Yuki`N oh, really?
dukeleto Yuki`N: but a submodule is another choice
Yuki`N: going for a walk, I'll be back in a bit. 05:02
05:07 rfw joined
Yuki`N dukeleto, ping 05:43
dukeleto Yuki`N: pong
Yuki`N: good timing, just got back from my walk
Yuki`N So the pseudoprime test is finding composite numbers in the perl test data
can I just delete them and ignore? 05:44
dukeleto Yuki`N: please explain
Yuki`N Well, is_strong_pseudoprime.t in the perl lib has a long list of #s which should be strong pseudoprimes.
But some of them aren't.
dukeleto Yuki`N: oh really?
Yuki`N Because they're not prime, and the algorithm is saying that they're not prime.
dunno 05:45
dukeleto Yuki`N: there are different kinds of pseudoprimes, and sometimes a pseudoprime algorithm thinks a composite number is prime, because it can't tell the difference 05:46
Yuki`N well in this case the algorithm thinks the number is composite when it thinks it shouldn't.
dukeleto Yuki`N: hmmm. and you are talking about your winxed implementation or the perl 5 one?
Yuki`N: if you find bugs in the perl 5 math::primality you definitely get extra credit! 05:47
Yuki`N: do you have a specific example?
Yuki`N Hmm 05:48
I may have gotten perl's map{} wrong
does it go from [start, end) or [start, end]
dukeleto Yuki`N: [start, end]
Yuki`N alright, well I fixed that but it's still throwing errors 05:49
but with bigger numbers
dukeleto Yuki`N: it is inclusive
Yuki`N so that is a good sign
Alright! It passes.
I was going from [start, end)
dukeleto Yuki`N: you scared me for a minute. I really hope the Math::Primality test suite doesn't have bugs :)
Yuki`N xD
dukeleto Yuki`N: 'cause i remember triple checking stuff :) 05:50
Yuki`N alright, well I pushed that. 05:51
It's also relatively quick. 05:52
dukeleto Yuki`N: exciting!
Yuki`N: if I had to guess/hope, the winxed implementation should fly compared to the Perl 5 version 05:53
Yuki`N Really?
Also it looks like lucas-pseudoprime testing is going to be a pain without tuple support. 05:54
dukeleto Yuki`N: winxed should have a lot less memory overhead
Yuki`N: please explain. What do you mean by tuple support?
Yuki`N Tuples, e.g (a, b, c) 05:55
functions returning tuples, e.g function x() { return ('a', 'b', 'c'); }
dukeleto Yuki`N: winxed doesn't have tuples? 05:56
Yuki`N: what about an array?
Yuki`N Rather sure it doesn't.
I suppose I could do that, but those aren't as nicely typed.
dukeleto Yuki`N: what in the lucas function wants a tuple ?
Yuki`N _find_dpq_selfridge 05:57
dukeleto oh, yes. 05:58
Yuki`N: just use an array for now. Suboptimal, but it will get the job done.
Yuki`N Well I have to sleep now.
dukeleto Yuki`N: sleep is good!
Yuki`N: great work for today. If you ever need a reference or recommendation in the future, you know who to ask :) 05:59
Yuki`N :P
that jumpyshoes -sigh-
submitted this task, btw, www.google-melange.com/gci/task/vie...11/7174256
dukeleto Yuki`N: that has a link to the wrong pull request 06:00
Yuki`N: it links to the is_small_prime pull request 06:01
Yuki`N well, they're both in there.
Github doesn't allow you to have >1 pull request open, and all new commits go to that pull request.
sorear um, can't you have more than one pull request if you have more than one branch? 06:02
opbots names
dukeleto Yuki`N: blarg. no worries, i will figure it out
sorear -sigh- 06:03
slavorg can't do its job unless it's opped
opbots trust alvis
slavorg Ok
dukeleto Yuki`N: task closed, get some sleep. 06:04
dalek rrot/documentation_revisions: 2232a12 | (Alvis Yardley)++ | MANIFEST:
Updated MANIFEST to include documentation_revisions-related stuff.
rrot/documentation_revisions: 56d2a54 | (Alvis Yardley)++ | MANIFEST.SKIP:
Modified MANIFEST.SKIP to skip a working file in 'docs/revisions/'.
alvis sorear: Does the "opbots trust ..." quieten things a'bit? 06:10
sorear alvis: no, it tells them to auto-op you, 06:13
.
alvis sorear: Oh, ok. I was hoping. (It's a touch embarrassing to bleed-out all-over-the-place on just some branch work.) 06:15
dukeleto alvis: it is ok. We like to konw what people are up to :) 06:24
know, even. 06:25
alvis: you can also keep some commits local, then use something like "git rebase -i HEAD~3" to squash the last three commits into one, then push, for example. 06:26
alvis: but you can just doing what you are now, too. It is fine.
sorear alvis: opbots means slavorg and slavorgn 06:30
alvis: not dalek
alvis sorear: Ok, thanks.
dukeleto: Yeah, I was just looking at that a'bit ago, but can't say I really get it, yet. 06:31
dukeleto alvis: no worries. It will all come in due time :)
alvis dukeleto: I was looking at your leto labs and saw where you're a git/github guru, so ... I got a question or two. 06:32
dukeleto alvis: oh noes! Shoot.
alvis: i am the one to blame for converting parrot to github :)
alvis dukeleto: Ok, I've completed the README.pod and parrot.pod, a pod for the executable, and they're in my branch. 06:33
dalek kudo/nom: 4269d73 | moritz++ | t/spectest.data:
the Unicode properties regex tests need ICU
sorear alvis: dalek does what it does because it's trying to handle git stuff in a KISS way
Tene alvis: there are quite a few people comfortable helping with git problems in the channel
alvis Now, I want to merge it into master. So, my first question is, should I?
sorear: Ok. I like KISS (but maybe I need KISS-DA! :) 06:34
Tene: Thanks.
sorear alvis: much of dalek's current design is my fault, btw.
sorear is especially pleased with the splitting of responsibilities 06:35
dukeleto alvis: i hear that you want to merge a branch into the master branch, is that right?
alvis dukeleto: Yep. I think so, but I'm looking for alternatives as well 'cause I really only want a few of the files in at this point in time. 06:36
dukeleto: I should say, "Yeah, I think so ..., maybe so; I don't know." 06:37
dukeleto: I think more than just the 'README.pod' needs to go in for the GCI stuff. 06:38
dukeleto alvis: so you have a few files in branch X that you want to merge into master, but not the whole branch? 06:39
alvis dukeleto: Yes.
dukeleto alvis: how many files? 06:40
alvis dukeleto: Lemme see ... 06:41
dukeleto: Ok, it's three files and two directories in 'docs' with only a couple of files in each. 06:42
dukeleto: Most of my work, of course, is in 'docs', so ... I don't think it'll break the build, but .... 06:43
dukeleto alvis: why don't you want to merge everything in the branch? wondering 06:44
alvis: i can tell you how to do what you want, but not being in this situation is the easiest way to deal with it :)
alvis dukeleto: I can. It's not a problem. I ... just don't wanna break anything. 06:45
dukeleto: ha! you're right there! :)
dukeleto alvis: basically, you can use "git checkout branch -- file" to 'checkout' versions of files from a branch 06:49
alvis: so you would go on the master branch "git checkout master" then git checkout each file from your branch with "git checkout branchname -- filename" 06:50
alvis: then git add the files, then commit
alvis dukeleto: i'm sorry, i'm not being clear: i have a 'documentation_revisions' branch, and i have it where i want it, for now. 06:51
dukeleto: there is some stuff i want in the parrot master, e.g., 'README.pod', and stuff for the GCI, e.g., 'docs/binaries/' and 'docs/readmes/' 06:52
dukeleto alvis: i gotcha 06:53
alvis dukeleto: and i need to "push" it, but i'm a'bit concerned 'cause i haven't done that, and i don't wanna muck things up. 06:54
dukeleto alvis: but why are you mixing different kinds of things in one branch? When you do that, you can't do a simple merge, you have to do the complicated thing I described above :)
alvis: i think there is some confusion
alvis: you have your local copy of the branch "documentation_revisions" and you have also "pushed" it to github, yes?
alvis dukeleto: yes. 06:55
dukeleto alvis: you are asking about merging the branch into master, but you only want part of the branch, not the whole branch. That is causing all the complexity here.
alvis dukeleto: yes.
dukeleto alvis: what different things are in that branch? What is in there that you don't want merged?
alvis dukeleto: the stuff in 'docs/revisions' (but, i really am ok if it goes in; i was just trying minimize impact). 06:57
dukeleto: most of the stuff in 'docs/revisions' is not ready for prime-time; it's just my working stuff.
dukeleto: or ... some of it is. 06:58
dukeleto: here's my thinking (if you want to call it that): the GCI students need the stuff in 'docs/binaries/' and 'docs/readmes/', all of which is my branch. 07:00
dukeleto alvis: so you created a new directory called docs/revisions ? and you have stuff in there?
alvis dukeleto: within my 'documentation_revisions' branch, yes. 07:01
dukeleto: 'docs/readmes' is where i'm keeping all of the non-pod 'README's for the GCI tasks, 07:02
dukeleto: and 'docs/binaries' is where the PODs for the installable binaries go. 07:03
dukeleto: i've kept it all in the same branch 'cause it's all documentation related.
dukeleto alvis: github.com/parrot/parrot/compare/m..._revisions 07:05
alvis: have you run "make fulltest" on your branch? 07:06
alvis: i would email the above link to parrot-dev and tell people you would like to merge the branch and see if anybody has any issues with that
alvis dukeleto: No, not yet.
dukeleto alvis: documentation can cause tests to fail, since some of our tests check docs-related stuff
alvis dukeleto: Ahhh .... Ok, that helps! 07:07
dukeleto alvis: and since you are moving our main README to README.pod, something might rely on the old location
not likely, but possible
alvis: gonna get some sleep. Feel free to send me/parrot-dev an email with any other questions concerns. And thanks for working hard to improve our docs!
dukeleto .sleep() 07:08
alvis dukeleto: Ok, 'night (and thanks!)
dalek rrot/documentation_revisions: 4576ba2 | (Alvis Yardley)++ | docs/binaries/README.pod:
Minor edit to 'docs/binaries/README.pod'.
 07:44
08:14 he joined 08:29 lucian joined 08:51 mj41 joined 09:15 lucian joined 09:17 JimmyZ joined
dalek kudo/nom: c285a5e | moritz++ | src/core/Attribute.pm:
fix Attribute.gist
 09:26
11:01 bacek joined, woosley left 11:03 ambs joined 11:05 woosley1 joined 11:20 bacek joined
dalek rrot: 6445248 | mls++ | / (3 files):
define ids for the subprof runcores instead of using PARROT_SLOW_CORE
 11:34
rrot: 6081fbd | mls++ | / (7 files):
Add a set_runcore op to switch to a different runcore.

Also support CURRENT_RUNCORE in interpinfo_s() to get the name of the current runcore. We need an extra op for the runcore switching, because we need to restart the runops() loop. This can only be done in an op at the moment.
p: ef5f94f | mls++ | / (3 files):
use parrot op to switch to the profiling runcore
 11:46
ttbot Parrot 6081fbdc MSWin32-x86-multi-thread make error tt.taptinder.org/cmdinfo/60678 11:48
dalek rrot: 93a98e9 | mls++ | / (2 files):
move stdint include from runcore_subprof.h to subprof.c
 11:57
12:12 Psyche^ joined 12:56 benabik joined
dalek kudo/nom: 01ac0d8 | moritz++ | src/core/ (2 files):
bring Str.trans back

Mostly a copy&paste from 2011.07, plus some fixes from jnthn++ and a few small optimizations that avoid some box operations		 13:01
kudo/nom: d2e4b02 | moritz++ | t/spectest.data:
enable some tests for trans, as well as another test that depended on it
benabik o/ #parrot 13:03
13:17 whiteknight joined
whiteknight good morning, #parrot 13:17
moritz good am, wk 13:18
whiteknight hello moritz, how are you doing today? 13:20
dalek kudo/nom: 8c9caf7 | moritz++ | src/core/Str.pm:
[trans] use native str for collecting the result
 13:24
moritz whiteknight: I'm ok, hacking on rakudo a bit 13:26
whiteknight moritz: I've been watching lots of exciting progress on Rakudo in the last few days. 13:27
I'm really itching to try porting some of that stuff down into Parrot
moritz whiteknight: I haven't been involved with the most interesting part of it (the NCI stuff)
whiteknight yeah, it really looks like many of the NCI changes are things Parrot probably wants 13:28
especially if it's set up to work closely with 6model, which we still need to port
moritz though note that nqp bundles the 'dyncall' library
in the past, parrot bundled libs too
then they were thrown out 13:29
whiteknight yeah, we need to re-think that policy
moritz not sure what the current policy is
whiteknight we don't have a current policy, I don't think, but are certainly open to suggestion
I, for instance, don't see any particular reason why we wouldn't bundle ICU
moritz +1
whiteknight considering how important ICU is for Rakudo, and how ungracefully we fail if we don't have it 13:30
moritz the non-bundled version is said to be a PITA to get working on windows
same for FFI (which is why we don't use it in nqp/rakudo)
same for GMP (same here, we use libtommath instead)
whiteknight native call is such a huge pain, and Parrot's default fallback (pre-generated thunks) are so ugly and limited. It makes no sense that we wouldn't try to bundle something 13:31
we could bundle multiple options, and pick the ones that work best on each platform
We're almost certain to want to bundle at least one library to help with threading and concurrency. There's no sane platform-independent way for us to have things like atomic operations and locks without library help 13:38
14:04 JimmyZ_ joined 14:12 JimmyZ_ joined 14:23 contingencyplan joined
dalek kudo/nom: 8716958 | coke++ | t/spectest.data:
run fudged test
 14:41
14:45 bluescreen joined 14:55 PacoLinux joined
dukeleto ~~ 15:33
whiteknight hello dukeleto 15:34
dukeleto whiteknight: wazzup 15:36
whiteknight not much duke, you? 15:37
dalek rrot-math-primality: b277ed6 | Yuki`N++ | / (11 files):
Build system is about 20% cooler. Expanded basic winxed test harness. Added and tested is_small_prime. Added rosella as submodule for dependencies.
rrot-math-primality: 0695c1c | Yuki`N++ | / (6 files):
Remove and ignore intermediates. Add is_strong_pseudoprime.
rrot-math-primality: 030ff29 | dukeleto++ | / (9 files):
Merge pull request #2 from nol888/master

Add and test is_small_prime.
dukeleto it was merge:30 15:38
whiteknight wowsers 15:39
Yuki'N++
dukeleto yeah, he is a code machine
whiteknight we need to capture him and steal his essence
15:39 alvis joined
dukeleto whiteknight: bottle it and sell it 15:39
whiteknight the build system is indeed 20% cooler 15:42
dukeleto msg Yuki`N you have a commit bit to parrot-math-primality now, in case you need it. Please still use pull requests for gci tasks, but now you can commit a fix if something is blocking you
aloha OK. I'll deliver the message.
benabik Build system whatnow? 15:43
Oh for Math::Primality
dukeleto benabik: the parrot port of Math::Primality, yes. We got rosella as a submodule as a value add 15:44
dukeleto is excited to start benchmarking Perl 5 vs. Winxed numerics
dalek sella/gh-pages: 678ef98 | Whiteknight++ | / (2 files):
First draft of docs for Dumper
 15:48
sella: cd4b2c6 | Whiteknight++ | t/ (3 files):
Add in a test file for Dumper. Add in a test file for Path that I've had uncommitted for a while
sella: eefe7ac | Whiteknight++ | src/unstable/dumper/ (3 files):
Small cleanups and docs for part of the Dumper library
sella: 3194bcb | Whiteknight++ | src/unstable/dumper/ (6 files):
+cleanups and docs for dumphandlers. Remove the InspectAttrs dumphandler from the default Dumper configuration
whiteknight dukeleto: assuming the code is well written and Parrot isn't doing anything stupid, I would like to think the performance should be similar or better 15:55
benabik And if not, time to profile profile profile 15:56
whiteknight exactly 15:57
I think we are going to see things we already know to be an issue: start-up times especially 15:58
MMD, if the benchmarks use any of that
dukeleto whiteknight: i hope that winxed is faster than Perl 5. If it isn't, something is wrong.
whiteknight it's only a level playing field because P5 doesn't have JIT either 15:59
dukeleto parrot.github.com is still not working and i haven't heard back from github yet
but it is still early. I submitted a support ticket last night.
whiteknight I didn't know that page went down 16:00
..yep, dead as a door-nail 16:01
dukeleto possibly related to the index.html becoming a symlink recently, but I hope not. 16:02
whiteknight hmm.. why would it be a symlink? 16:03
it seems like that should be some kind of landing/welcoming page 16:11
dukeleto whiteknight: it is a symlink to a generated file a few directories down 16:14
whiteknight ok 16:16
dukeleto blarg. 16:17
dalek href="https://parrot.github.com:">parrot.github.com: 7b4df86 | dukeleto++ | index.html:
See if index.html being a symlink is making Github unhappy
 16:18
dukeleto yep, github doesn't like symlinks for index.html pages 16:19
whiteknight ...apparently
dukeleto comments on this are appreciated: github.com/parrot/parrot.github.com/issues/2 16:37
benabik +1 16:53
dalek p: 19f994b | jnthn++ | / (329 files):
Merge branch 'nci'
 17:07
dukeleto nqp has it's own nci now? 17:12
whiteknight yes, and I suspect we want to "borrow" some pieces from it 17:13
benabik Is there a particular advantage of dyncall over libffi? 17:15
whiteknight I don't know either library well enough to know 17:17
dukeleto dyncall.org/showcase.shtml 17:19
seems like the little sister of libffi. Not sure what feature set differences there are.
benabik Smaller version number? :-D
dukeleto looks like R people use dyncall 17:20
benabik Hm. "For the amxProcess module, the dyncall library now replaces the libffi library, because dyncall comes with ports for Unix/Linux, Macintosh and Windows (libffi only supported Unix-like systems)." ~~ code.google.com/p/pawnscript/updates/list 17:21
dukeleto dyncall.org/docs/26c3/dyncall_26c3.pdf is interesting 17:22
dyncall may support more platforms.
dalek p: e62d398 | jnthn++ | src/NQP/Actions.pm:
Ensure references to self in NQP are decontainerized.
 17:25
kudo/nom: b2a9370 | jnthn++ | src/ (7 files):
Merge branch 'nci' into nom
 17:26
kudo/nom: aa15f3a | jnthn++ | tools/build/NQP_REVISION:
Bump to latest NQP revision.
benabik Looks like big advantage for Parrot is that dyncall works in "native" Win32. FFI wants Cygwin/MSys 17:27
Util benabik: better link: code.google.com/p/pawnscript/source/detail?r=33
whiteknight if dyncall is the better library, maybe we move to that 17:30
and if nqp already has bindings, we can try to port those downward into parrot
17:30 alester joined
dukeleto thinking about all our NCI stuff changing is frightening. 17:31
whiteknight it was already going to change a lot for lorito
dukeleto whiteknight: that is kind of like saying "well, the heat death of the universe was already going to happen" 17:33
dalek rrot: 34fb4fd | petdance++ | / (4 files):
s/existant/existent/g per Larry
 17:34
p: c1c6435 | jnthn++ | src/ (3 files):
Spellos; TimToady++ for noticing.
 17:35
itor: 1cef5d4 | NotFound++ | / (4 files):
improve color handling and add constants for line style
 17:56
18:20 fperrad joined
dalek p/qbootstrap: 7508c6a | jnthn++ | tools/build/Makefile.in:
First steps towards getting NQP bootstrapping with QRegex. This just twiddles the makefile so we always build QRegex during stage 2, and then build NQPQ as the stage 2 NQP.
 18:39
p/qbootstrap: 6060370 | jnthn++ | src/QHLL/ (5 files):
Add a temporary QHLL copy of HLL to help facilitate the bootstrap.
p/qbootstrap: 70ac5d0 | jnthn++ | tools/build/Makefile.in:
Build QHLL in stage 2.
rrot: 0ea033c | (Paweł Nowak)++ | / (3 files):
Add a Parrot_api_pmc_box_float function to src/embed/pmc.c

Add a Parrot_api_pmc_box_float function to src/embed/pmc.c, change Parrot_api_pmc_box_integer to use Parrot_pmc_box_integer instead of duplicating it. Also add a test for these functions to t/src/embed/pmc.t.		 18:41
rrot: 44a3d69 | dukeleto++ | / (3 files):
Merge pull request #213 from Pawel834/box_float

Add a Parrot_api_pmc_box_float function to src/embed/pmc.c
cotto ~~ 18:43
whiteknight hello cotto 18:46
cotto 'morning, whiteknight
Our core competence isn't nci. I'm open to the idea of experimenting with dyncall. 18:48
whiteknight yay! 18:52
I'm not against libffi per se, but if dyncall is more portable and if Rakudo is using it, those are two pretty compelling arguments 18:55
I would really like to talk to plobsing about it, but I haven't seen him in a while
19:01 lucian joined
dalek itor: 71322bb | NotFound++ | src/Guitor (2 files):
improve window creation functions and add ResizeWindow
 19:09
itor: 0960d2c | NotFound++ | examples/pizarra.winxed:
add toolbars to example pizarra
p: 47e75d0 | jnthn++ | 3rdparty/dyncall/dynload/dynload_syms_elf.c:
Hopefully fix the build for tadzik++.
 19:10
cotto #ps in 16 19:14
19:18 particle1 joined
cotto #ps time 19:30
alvis msg whiteknight if there are any documentation gci requests, you can find everything -- if you need/want to -- in my 'documentation_revisions' branch in the 'docs/readmes/' and 'docs/binaries' directories. 20:11
aloha OK. I'll deliver the message.
whiteknight thanks, I'll look there
20:52 mj41 joined 21:26 lucian joined 21:39 Yuki`N joined
Yuki`N dukeleto, ping 21:41
dalek rrot-math-primality: 20b5037 | Yuki`N++ | Makefile:
Small fixes to Makefile.
 22:22
rrot-math-primality: 400cb6c | Yuki`N++ | /:
Merge branch 'master' of github.com:letolabs/parrot-math-primality
itor: 4ff4b3d | NotFound++ | / (3 files):
reorganize text drawing and start using Xft fonts
 22:52
22:58 benabik joined 23:37 whiteknight joined
Yuki`N whiteknight, how good are you at reading math papers consisting of 70% symbols and 30% text? 23:38
whiteknight Yuki`N: probably not as good as I was back in gradschool 23:39
Yuki`N Trying to decipher this method of counting primes.
whiteknight oh, that task from dukeleto?
Yuki`N yeah. 23:42
I give up. 23:44
whiteknight Okay, don't worry about the optimization 23:46
this is a coding competition, not a "who can read complicated math the best" 23:47
Yuki`N lol 23:48
Well, the entire task is pretty much the optimization. 23:49
However, if I could get dukeleto to write pseudocode for it. =3=
whiteknight this is a learning process for us too, we don't get any guidelines from google about what makes a "good" task 23:51
so we see people like you burning through tasks and we think "ok, they need to be harder"
then they get too hard, and we need to go back down
It's your own fault for being too good
Yuki`N lol 23:52
dalek itor: 59d69b1 | NotFound++ | / (3 files):
getTextWidth
 23:54
dukeleto Yuki`N: what now? 23:55
Yuki`N: trouble with the math paper?
Yuki`N only a little bit. 23:56
dukeleto Yuki`N: i am at your service
Yuki`N I don't understand all of it. 23:57
It seems that they're splitting the computation of pi(x) into sub-parts. 23:58
dukeleto Yuki`N: yes 23:59
Yuki`N however, I have no idea how any of that gets implemented in code.
dukeleto Yuki`N: what is the first sentence in the paper that doesn't make sense?

Powered by Cro and the Raku® Programming Language.

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

Thank you!