Camelia, the Perl 6 bug

IRC log for #november-wiki, 2009-04-12

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

All times shown according to UTC.

Time Nick Message
01:21 ihrd joined #november-wiki
01:21 ihrd left #november-wiki
02:16 Matt-W_ joined #november-wiki
02:24 Matt-W joined #november-wiki
02:35 zarah joined #november-wiki
02:35 wayland76 joined #november-wiki
02:35 sri_kraih joined #november-wiki
02:35 mj41 joined #november-wiki
02:35 PerlJam joined #november-wiki
02:43 wayland76 joined #november-wiki
03:29 Tene_ joined #november-wiki
03:55 ihrd joined #november-wiki
04:08 ihrd left #november-wiki
05:27 Tene_ joined #november-wiki
05:38 zarah joined #november-wiki
05:49 zarah joined #november-wiki
07:01 ihrd joined #november-wiki
07:58 ihrd left #november-wiki
10:27 ihrd joined #november-wiki
11:59 ihrd left #november-wiki
12:07 masak joined #november-wiki
12:13 ihrd joined #november-wiki
12:13 ihrd masak, hi!
12:14 ihrd how are you?
12:14 masak hi ihrd!
12:14 masak I'm good -- I'm ready to blog my fourth week of Web.pm.
12:14 masak how are you?
12:14 ihrd I`m happy to see you progress :)
12:15 ihrd I still slow, but mb you see my 2w post
12:15 masak aye.
12:15 masak I read it this morning.
12:15 ihrd do you like Routes?
12:15 masak to the extent that I understand them, yes.
12:15 ihrd I have no feedback from you
12:15 masak I think I need to see them in a working app to really grok them.
12:17 masak did you run into a Rakudo bug that prevented you from progressing?
12:17 masak or did I misunderstand?
12:17 ihrd it`s close to usefull version, but I have block -- | %args NIY
12:18 masak NYI.
12:18 masak ok.
12:18 ihrd yes, one of them
12:18 masak so it's not a bug so much as a missing feature.
12:18 ihrd simplest one
12:18 masak we should assault jnthn and pmichaud about this.
12:19 masak I have a feeling that "we need this for Web.pm" carries some weight.
12:19 ihrd yes, you are right, btw it is block, and I do not like to work around it
12:19 masak I see.
12:19 ihrd :)
12:20 masak I know the feeling -- I've had things blocking that I didn't want to work around either.
12:20 ihrd mb Tene can help with this stuff too
12:20 masak possibly.
12:22 ihrd I tring to find out how implemented | @args
12:22 ihrd with hope I can understand how to implement | %args
12:22 ihrd but have no progress in this direction
12:23 ihrd just do not understand internals enought
12:23 masak internals are hairy.
12:24 masak maybe jnthn and pmichaud can give guidance in some places.
12:24 masak ihrd: I have the lobster running. this morning I made Druid show up in the browser too. so, progress.
12:24 ihrd yes, I will speak with Jonathan about it
12:25 ihrd yes, this is good
12:25 masak Adrian, our friendly neighborhood grant manager, seems very absent.
12:26 ihrd aye :(
12:26 masak good thing you emailed him. I will do the same today when I've blogged.
12:26 ihrd it feels like we do not have grant manager at all
12:26 masak indeed.
12:33 masak it will be nice to get him back.
13:11 masak joined #november-wiki
13:52 Tene joined #november-wiki
14:29 ihrd I was push my last Routes
14:30 ihrd masak: I like you to make review, as yo do that in the past for my code
14:48 ihrd masak: a u here?
14:50 ihrd left #november-wiki
15:14 sri_kraih_ joined #november-wiki
15:21 masak ihrd: will do, sometime during the day.
15:28 Tene masak: Tags exports map?!?!
15:29 Tene will look during flight.
15:29 Tene and will look into refactoring it a bit.
15:29 masak Tene: I pushed a commit making it no longer do that.
15:29 Tene ah
15:29 Tene I didn't get a chance to pull updates before leaving for this flight.
15:30 Tene the past few days have been insane.
15:30 Tene got a parrot branch that reports HLL line number for exceptions, though. :)
15:31 Tene but will work on web stuff all this week.
15:31 masak Tene++
15:31 Tene travel week, so no problems at home to deal with
15:31 Tene I really need to catch up. I've fallen way behind
15:31 Tene I want tagsets for tags
15:32 Tene html5, xhtml, etc.
15:32 Tene Also, if I can get context vars, I can remove the singleton
15:32 Tene and PM added an opcode specifically for context vars.
15:33 Tene so, promising.
15:33 Tene any requests for Web work this week?
15:33 Tene I'd like direction.
15:33 Tene (yes, again...)
15:33 Tene oops, phone off now.
15:33 Tene email?
15:33 Tene bye.
15:34 masak bye.
15:35 masak I'll send an email.
15:36 Tene :) thanks
15:36 Tene and sorry again.
15:36 masak np.
15:36 masak I'll write thoughts here, and then paste them into an email. :)
15:39 masak since we now can haz browser apps, things get a lot more interesting and palpable.
15:40 masak I think we should go for the pastebin clone pretty soon.
15:45 masak what do we need for that? we need to flesh out Web::Request and Web::Response more, and clear up the relationship between that and HTTP::Daemon.
15:49 masak there's also the question of a storage backend. files is the current default, I'm afraid.
15:50 masak apart from that, now is the time to start serious porting of either Genshi or Hpricot, I think.
16:03 masak the former will come of use in November (if we can get it fast enough), and the latter could be used for wrapping use.perl.org to make it usable.
16:03 masak both are pretty nice use cases, and building them will be fun.
19:17 masak http://use.perl.org/~masak/journal/38800
19:17 zarah masak's link is also http://tinyurl.com/dc7r46
19:51 * sri_kraih_ doesn't understand why someone would want to port rack, it's CGI.pm all over again
19:51 masak sri_kraih_: it is?
19:52 masak and here I've been trying to avoid exactly that mistake. :)
19:52 * sri_kraih_ nods
19:52 masak sri_kraih_: what, in your opinion, would _not_ be CGI.pm all over again?
19:53 sri_kraih_ something based on the idea of abstracting http
19:53 masak ok. I thought Rack does that.
19:55 sri_kraih_ it expects a cgi like environment
19:56 masak you're right, it does.
19:56 masak it seems a very common model for the webapp flow.
19:57 masak are you advocating something more like Seaside?
19:57 sri_kraih_ that doesn't make it right :)
19:57 masak sri_kraih_: I hear you. I'm now at the point where I'd like to know what you're proposing instead.
19:57 sri_kraih_ i'm very happy with how Mojo is turning out
19:57 masak ok.
19:58 sri_kraih_ it's all about http
19:58 masak haven't looked at that.
19:58 masak how come you're happy about it if it's all about http?
19:58 masak didn't you want something that abstracted http?
19:59 sri_kraih_ like?
19:59 masak oh, you meant 'built upon'. I see.
19:59 masak I read 'abstracted away'.
19:59 sri_kraih_ ah
19:59 masak now it's much clearer to me what you prefer.
20:00 masak I'll take a look at Mojo and see if I can learn something from it.
20:00 sri_kraih_ the core idea is to have message modules that can represent everything in the http spec
20:00 masak what does that give you that CGI doesn't?
20:01 sri_kraih_ it's a full http 1.1 stack
20:01 sri_kraih_ parser and generator
20:01 masak sounds nice.
20:01 sri_kraih_ you can use subsets of it's features depending on the environment
20:01 masak I don't really see why that should be incompatible with Rack or Web.pm.
20:02 sri_kraih_ different philosophy imo
20:02 masak very possibly.
20:02 sri_kraih_ rack wants to stay away from http as far as possible :)
20:03 masak sri_kraih_: its core abstractions seem to be Request and Response.
20:03 masak sri_kraih_: isn't that what HTTP is about?
20:03 sri_kraih_ and Transaction
20:03 sri_kraih_ oh you mean rack
20:03 masak aye.
20:04 masak I'm looking at the Mojo synopsis on CPAN right now.
20:04 masak I see many similarities to Rack.
20:04 sri_kraih_ in Mojo we abstract a bit further with Transaction and Pipeline
20:04 masak ok, sounds interesting.
20:04 masak will investigate.
20:05 masak ah, so the $tx is a Transaction, holding both the request and response?
20:05 sri_kraih_ yes
20:06 sri_kraih_ and in the next version we'll have Pipeline to hold multiple transactions
20:06 masak ok. when will that be useful?
20:06 sri_kraih_ for pipelining
20:07 masak well, duh.
20:07 masak what's that?
20:07 sri_kraih_ normal http connection is Request -> Response... with pipelining in http 1.1 you can have Request -> Request -> Response -> Response
20:08 masak this sounds both exciting and frightening.
20:08 sri_kraih_ not exactly async, but you don't have to wait for a response to send the next request
20:08 masak aha, so the pipeline simply consists of requests and responses treated in a queue-like manner?
20:08 sri_kraih_ without pipelining a http 1.1 implementation is usually much slower than http 1.0
20:09 * sri_kraih_ nods
20:09 masak I see.
20:12 sri_kraih_ in the next release Mojo will also contain a reusable state machine for http 1.1 (client and server)
20:12 sri_kraih_ all you'll have to do is take care of IO and read/write when you are told to
20:13 sri_kraih_ 100 continue and pipelining will just work
20:13 masak sri_kraih_: what you are saying sounds quite important, and I hope it will some day be incorporated into Web.pm too -- hopefully without too much retrofitting.
20:13 sri_kraih_ we also have parsers and generators for chunked encoding and multipart messages
20:13 masak it's not within the bounds of our current grant, unfortunately.
20:13 sri_kraih_ i know
20:14 sri_kraih_ actually i'm following your grant work to see how far i could get with a Mojo port :)
20:14 masak :)
20:14 masak you mean a port of Mojo to Perl 6?
20:14 sri_kraih_ Mojo was designed to be ported
20:15 masak oh, was it? in what way?
20:16 sri_kraih_ mostly being careful with prereqs
20:16 sri_kraih_ Mojo has none outside of core perl5 atm
20:16 masak cool!
20:16 masak I imagine that took some effort.
20:17 sri_kraih_ was a lot of fun and a great learning experience :)
20:17 * sri_kraih_ learned to hate all the strange edge cases in rfc2616
20:18 masak heh.
20:20 masak sri_kraih_: in summary: I hear what you're saying. I'm on a learning curve too here, so all I can promise is to incorporate Mojo into the things I will look at and hopefully be inspired by.
20:20 sri_kraih_ i'm looking forward to whatever you can come up with that makes porting for me easier :)
20:21 sri_kraih_ the netcat stuff in http::daemon is scary though
20:21 masak indeed.
20:21 masak mberends++ is taking care of that as we speak.
20:22 masak sri_kraih_: I think we all are in need of the Buf type in Perl 6.
20:22 masak with a bit of luck, I'll be able to start implementing that shortly.
20:29 Tene_ joined #november-wiki
20:39 sri_kraih_ oh yea, another thing about rack that i hate is that it allows direct writing on the socket
20:39 sri_kraih_ in Mojo we use a pull model, that works way better inside event loops
20:40 masak I haven't come across that direct-writing in Rack yet.
20:40 masak will watch out for it.
23:15 wayland76 joined #november-wiki

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