Camelia, the Perl 6 bug

IRC log for #november-wiki, 2009-06-05

| Channels | #november-wiki index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:39 ihrd joined #november-wiki
01:40 ihrd left #november-wiki
02:52 ihrd1 joined #november-wiki
03:40 ihrd1 left #november-wiki
04:00 ihrd joined #november-wiki
05:26 ihrd left #november-wiki
06:16 viklund_ joined #november-wiki
07:46 masak joined #november-wiki
07:57 lichtkind_ joined #november-wiki
07:59 zarah joined #november-wiki
08:02 Tene masak: ping
08:03 masak Tene: pong.
08:04 Tene masak: I have vague memories of you talking with someone about working on dealing with the auth string on module loading in rakudo?
08:04 Tene Or did I misunderstand or make that up?
08:05 masak Tene: I created a Rakudo branch for such work. haven't actually committed anything yet, though.
08:05 Tene masak: I'm currently using that in rakudo trunk for foreign hll loading... auth colonpairs are being parsed and passed to use.
08:05 Tene So... what are you trying to do?
08:06 masak I'm trying to build something concrete that actually handles several same-named modules of different versions and authors.
08:07 masak I know the colonpairs already parse in Rakudo.
08:07 masak but it's another thing to actually handle the ver/auth information.
08:07 Tene So you're working on actually responding to that information.
08:07 Tene Okay.
08:08 Tene Just wanted to avoid conflicting or duplicate effort.
08:08 masak aye. good.
08:10 masak I'm currently trying to orient myself in the Genshi codebase, to get some hints on Hitomi design.
08:11 masak my most recent find is that Genshi has tests! originally I thought it didn't.
08:12 Tene I'm still winding down after being very busy for the past weeks...
08:12 Tene working on some HLL stuff and some blog posts about it.
08:12 masak nod.
08:12 Tene I should be getting back to Web stuff soon.
08:13 Tene I'm starting to get a little disappointed at the lack of work on most other languages on Parrot.
08:13 Tene I guess it makes sense that there won't be much interest from most people until Parrot is proven a bit better.
08:13 Tene Would be nice, though.
08:13 Tene Found out yesterday that Pynie doesn't do classes at all.
08:14 masak Pynie is the Python implementation?
08:14 Tene yes
08:14 Tene I'm thinking of trying to improve it enough to get basic classes and namespaces working, and loading foreign libraries.
08:15 masak would be nice.
08:15 Tene I got a very basic app running on cardinal (ruby) using rakudo's HTTP::Daemon
08:16 Tene But... cardinal's parser is really really bad.
08:16 Tene I never did figure out exactly why it was so absurdly slow... I should return to it someday.
08:17 Tene (like... in some cases it would take several *minutes* to parse a several-line program.  Mostly depends on how deeply nested anything is.)
08:17 masak maybe it does backtracking wrong.
08:17 Tene Probably.
08:17 masak it's usually something like that.
08:18 Tene I mean, I get the general idea, but I don't know the specific cases that were failing the worst.
08:19 Tene anyway, that's work for another day.
08:20 Tene It was pretty cool.
08:20 Tene Hmm... I bet scheme would be friendlier...
08:20 Tene oh, no, blocks in scheme are a pain.  Unless I introduce some friendlier lambda syntax, Tags.pm would suck in scheme.
08:23 Tene Eh, http;;daemon could still be nice.
08:25 masak :)
08:25 masak I'm pursuing two tracks with Web.pm right now.
08:26 masak one is building a simple Druid web app. the other is porting Genshi->Hitomi.
08:26 masak both are going a bit slower than I'd like. I should take an evening and do just those two.
08:26 Tene oh man, no way to call methods from steme... ><
08:27 masak that sounds problematic.
08:27 Tene r6rs doesn't talk about classes...
08:29 Tene 'night
08:37 masak 'night.
09:09 viklund masak: November crashes when unicode characters are submitted ;(
09:10 masak viklund: aye.
09:10 masak viklund: do we have a ticket about that?
09:10 viklund no
09:10 viklund or maybe I did it
09:10 masak feel free to create one.
09:10 viklund no, no ticket
09:11 masak no tickee?
09:11 viklund I can fix one though
09:11 masak please.
09:11 viklund it seems that it's not sent as utf-8, which is strange
09:11 masak yes.
09:11 masak we've discussed this before. it's a mystery.
09:12 viklund I looked at the post using live http headers, and there it was 3 characters instead of —
09:12 masak aye.
09:12 viklund maybe ask the catalyst folk, they should know smth.
09:12 masak I'm on catalyst right now. I could ask them.
09:12 viklund please do
09:12 masak what exactly should I ask?
09:15 viklund zarah: welcome back
09:15 zarah viklund: you've just exceeded my capabilities :/
09:17 viklund it's filed now at least
09:17 masak viklund++
09:17 viklund maybe we can detect utf8 somehow and stop it
09:18 masak yes, but why doesn't it work?
09:18 viklund_ joined #november-wiki
09:18 viklund I think it has something to do with how it is sent..
09:18 masak from tonight's backlog, I got the impression that it's a Rakudo problem...
09:19 viklund_ joined #november-wiki
09:19 viklund masak: no, I think that that was a problem with interactive mode in Rakudo...
09:19 viklund but I'm not sure
09:19 masak ok.
09:20 viklund since it worked if we had it in a perlscript
09:20 masak ah, right.
09:20 masak so, what should I ask the #catalyst people?
09:21 viklund ehm, how to recieve utf8 from a post?
09:21 viklund I'm going to test it with p5w I think and see what happens...
09:22 masak goodie.
09:23 viklund this is interesting for Web.pm as well I guess
09:24 masak very much so.
09:24 viklund lichtkind thought it was a stopper for putting stuff up, oh well
09:24 viklund it
09:24 viklund it's frustrating anyhow
09:25 masak it's not a blocker, but it's certainly embarassing.
09:25 masak encodings are hard! :/
09:25 viklund if we only could stop november from choking on it we would be far better of
09:25 viklund off?
09:25 viklund no
09:26 viklund I guess that when a character is several bytes the first byte tells how long it is somehow...
09:26 viklund maybe we can check that
09:28 masak it's possible, but guessing encodings is somewhat of a last resort.
09:28 masak it will lead to other, possibly worse, problems.
09:28 viklund my point isn
09:28 viklund isn't guessing encodings, it's saying, "oops, we can't handle this, remove those squiggles"
09:29 viklund until we have a real fix
09:29 masak yes, sure.
09:29 masak we can do that.
09:30 masak http://en.wikipedia.org/wiki/UTF-8#Description describes how the bytes of UTF-8 encoding look.
09:30 zarah masak's link is also http://tinyurl.com/5ha63
09:32 viklund ty
09:33 viklund p5w is using a lot of modules I don't have... thank god for the cpan shell
09:33 viklund would be nice with a program that scanned perl files and installed all deps...
09:34 masak doesn't sound too hard to do.
09:34 viklund no
09:34 viklund and yes
09:35 viklund since i p5 you can do 'use "Stuff"'
09:39 masak sure, but if the program finds and installs 99 out of 100 modules for you, it's still of some use.
09:39 viklund hmm, in p5w the � gets removed
09:39 viklund hmm
09:40 masak that character you just posted comes up as '\227' here.
09:40 viklund hmm
09:41 viklund I'm on a different computer with a different irc client now
09:41 viklund can you paste the dash from november-wiki.org?
09:42 masak sure, hold on.
09:42 masak
09:43 viklund that worked at least
09:43 viklund
09:44 masak I got \227 again.
09:45 p6eval joined #november-wiki
09:45 viklund ;(
09:45 viklund p5w doesn
09:46 viklund doesn't work either...
09:50 masak viklund: that's sad. :/
09:50 masak viklund: ok, step 1: disallow the utf-8 in November.
09:50 viklund masak: but it doesn't die...
09:51 masak step 2: isolate the error as a test case and show it to the Rakudo devs.
09:51 viklund dang
10:06 viklund masak: ugly hack plan: if ord($chr) > 127 { die somehow }
10:06 masak viklund: aye.
12:20 ihrd joined #november-wiki
12:20 ihrd left #november-wiki
13:58 ruoso joined #november-wiki
18:13 lichtkind does november work now :) ?
18:53 ruoso_ joined #november-wiki
18:57 masak joined #november-wiki
19:16 viklund heh
19:17 masak viklund++
19:17 viklund lichtkind: we don't know how to fix that utf thingy yet, I'm working on implementing a fix, I've just come home from beeing out all day
19:17 viklund so instead of an ERROR you will get a "sorry try again, and please make sure it's only ascii this time"
19:18 masak ascii-- # :(
19:18 masak where's our promised land of UTF-8?
19:19 lichtkind viklund: so i have time to finish the tutorial in the tpf wiki :)
19:20 viklund "Welcome home, to my broken promised land, where the skys are grey, there's no stars to lead my way"
19:21 masak viklund: well, aren't you the cheery one.
19:21 viklund mmm, cherries!
19:35 viklund I'm working on the utf8 fix now
19:35 viklund in the mean time, check this out:
19:36 viklund http://github.com/viklund/november/commit​/113fe78a221b88b242a58492f58933cb84afe413
19:36 viklund closing issues via commit messages! Nice
19:36 viklund github++
19:43 masak :)
19:53 viklund masak: How do I slurp all arguments into an array?
19:54 masak viklund: either | or * depending on whether we're talking argument or parameter.
19:54 viklund method call?
19:55 viklund method stuff( *@all ) { ... }
19:55 viklund ??
19:55 masak aye.
19:55 masak but that's not a call, that's a declaration...
19:55 masak hence, parameters.
19:55 viklund ah
19:55 viklund ty
19:56 masak np
20:26 viklund now I need to test, *sigh*, time to reinstall apache...
20:28 masak viklund: we should really make a prepackaged november/apache pseudoproject on proto.
20:28 masak viklund: maybe a good project for Tuesday.
20:29 viklund that
20:29 viklund maybe, might be good...
21:02 zarah joined #november-wiki
21:08 viklund oops, I curl'd a png...
21:08 masak :)
21:19 viklund I've got it running, but for some reason it doesn't get styled...
21:20 masak that's odd.
21:20 masak can you get at the style file directly?
21:20 viklund yes
21:21 viklund then I curl'd the png...
21:22 masak ok.
21:23 viklund luckily, I run screen, so I just closed the vt
21:24 viklund masak: grammar checkup, this is my error message:
21:24 viklund Sorry, november can't currently handle UTF8
21:24 viklund "currently can't"?
21:27 viklund no, we should say that we only handle ascii...
21:28 masak aye.
21:29 masak fwiw, I'd say "currently can't"
21:30 viklund "We're sorry, but for the moment November can only eat ASCII"
21:30 masak s/eat/consume/
21:58 viklund masak: now, keep your fingers crossed ;)
21:58 * masak crosses 'em
22:02 viklund wait for it...
22:02 masak ...
22:03 masak ...
22:03 viklund_ yes...
22:03 viklund_ it's some other error too.
22:03 masak :/
22:03 viklund_ ahh
22:04 viklund_ I haven't cleaned the server ;)
22:05 viklund_ huzzah!
22:05 viklund_ it works, that is, we get an error message ;)
22:06 masak viklund_++
22:06 viklund but what about me?
22:08 masak viklund: are you kidding? viklund_ did all the work!
22:08 masak the insolence...
22:08 viklund_ viklund: there you go
22:10 viklund masak: there is one more alley to check for this issue
22:10 viklund apache configuration
22:10 masak oh. indeed.
22:10 viklund other than that I don't know
22:10 viklund but, now, I want to sleep
22:10 masak viklund: thanks for the good work tonight.
22:10 masak we should do this more often. :)
22:10 viklund_ masak: Yes
22:11 viklund_ and look at my comment on the bug:
22:11 viklund_ http://github.com/viklund/november/issues#issue/6
22:11 zarah viklund_'s link is also http://tinyurl.com/q77dpu
22:11 * masak looks
22:12 masak viklund_: .split takes '' as a default argument nowadays.
22:12 viklund_ github++ again
22:13 masak I know, since I blitz-changed that after TimToady changed his mind.
22:13 viklund_ masak: oh, good, I was wondering that when I wrote that code...
22:13 masak oh, wait. might even be more appropriate to use .comb nowadays.
22:13 viklund_ comb?
22:13 masak it also takes '' as default. :)
22:13 masak viklund_: yes. :) TimToady usually mumbles about 'figure vs ground' with these two.
22:14 masak that is, split removes and comb keeps.
22:16 viklund_ my comb doesn't work:
22:16 viklund_ ./perl6 -e 'my $a="bacadae"; for $a.comb("a") -> $c { say $c }'
22:16 masak rakudo: my $a="bacadae"; for $a.comb("a") -> $c { say $c }
22:16 p6eval rakudo fb2fd4: OUTPUT«No applicable candidates found to dispatch to for 'comb'␤in Main (/tmp/pfAiDBPxNb:2)␤»
22:16 masak hm.
22:17 viklund_ what should the answer for that be. Or rather, where does the a's go?
22:17 viklund_ do*
22:17 masak .comb should return a List.
22:18 masak very strange. I'm looking at the source code right now.
22:18 masak let's run this by #perl6.
22:18 masak in #perl6, it works :)
22:19 masak viklund_: oh!
22:19 viklund_ rakudo: say 'abcdefghi'.comb(/<[aei]>/).perl
22:19 p6eval rakudo fb2fd4: OUTPUT«["a", "e", "i"]␤»
22:19 viklund_ ah!
22:19 masak you use a string, but only a Code variant is defined.
22:19 masak checking spec...
22:20 viklund_ rakudo: my $a="bacadae"; for $a.comb(/a/) -> $c { say $c }
22:20 p6eval rakudo fb2fd4: OUTPUT«a␤a»
22:20 masak nope, it's only defined for regexes.
22:20 viklund_ rakudo: my $a="bacadae"; for $a.comb(//) -> $c { say $c }
22:20 p6eval rakudo fb2fd4: OUTPUT«Statement not terminated properly at line 2, near "(//) -> $c"␤in Main (src/gen_setting.pm:0)␤»
22:20 viklund_ rakudo: my $a="bacadae"; for $a.comb(//m) -> $c { say $c }
22:20 p6eval rakudo fb2fd4: OUTPUT«Statement not terminated properly at line 2, near "(//m) -> $"␤in Main (src/gen_setting.pm:0)␤»
22:21 masak viklund_: I guess taking Str as a first parameter is less useful for .comb than it is for .split :)
22:21 masak viklund_: the // pattern is forbidden in Perl 6.
22:21 viklund_ ;)
22:22 masak viklund_: and there are no longer any trailing flags.
22:23 viklund_ anyway, good night
22:24 masak viklund_: 'night.

| Channels | #november-wiki index | Today | | Search | Google Search | Plain-Text | summary