2009 -- the year of November. <november-wiki.org> <github.com/viklund/november> <irclog.perlgeek.de/november-wiki> <nopaste.snit.ch>
Set by moderator on 10 February 2009.
00:45 ihrd1 joined 03:27 Tene_ joined 07:29 ihrd1 left 08:27 masak joined 12:17 ihrd joined
ihrd hi there 12:34
masak, I see you messages in the log 12:51
masak hello.
ihrd dispatcher fully in master, so we can remove this branch 12:52
asteriks was blocken, but now I fell I can make it works, I will do that 12:53
masak cool.
I'll remove the dispatcher branch. 12:54
ihrd++
ihrd ah, plz, use ++ when I done something :)
masak you answered my question :)
ihrd I still ill and that makes my sad :(
masak also, you implemented dispatch in November.
ihrd: I wish you good health. get lots of rest and liquid. 12:55
ihrd but I skid with new shine Web::Dispatcher
I cant first, but I do second :) 12:56
what you thinking about my tiny example, Forest? 12:57
masak can't parse 'skid with new shine'
sounds great, though :) 12:58
ihrd: I haven't had time to look closely at Forest, but I'll be glad to.
I really like the name.
I'll look at it this evening, when I'm not at $WORK.
ihrd I mean skid like car do 12:59
masak ok, I get it now.
ihrd or slip
masak 'slip' in projects means 'postpone'.
ihrd I just found thet words in dic
masak :) 13:00
ihrd in russian we use phrase like this
to sign situation when someone go forward slow and with trouble 13:01
masak so, just to make clear, 'skid with new shiny Web::Dispatcher' means 'make slow progress with the new shiny Web::Dispatcher'? 13:02
if so, I'm still looking forward to it when it arrives. :) 13:04
working code doesn't even have to involve a full stack... but you know that, so no need to say it. :) 13:05
ihrd Tene_, how are you? 13:07
masak: I hope fix my body, and make progress on this weekend 13:08
masak ihrd: I'm hoping for progress here too, tonight and in the weekend. 13:09
currently porting Rack's Util, Request, Response
ihrd masak++
masak discovering things missing in Rakudo that we need. 13:10
specifically, turning a string into its UTF-8 bytes.
Parrot should be able, but the Parrot devs seem slightly befuddled as to why I would want to do this. 13:11
ihrd my Template. blocked by rakudobug again
masak aye, saw that. 13:12
no workaround available?
ihrd Jonathan :)
masak sure, but short-term.
ihrd he said this one simple to fix, and he do thest
masak oki. 13:13
ihrd thet in next few days
can you put our propousal pod to the root of web repo?
masak sure, good idea. 13:14
masak does that
ihrd and I think we should make list of related repos in README
Maya, November, URI
masak definitely.
and also describe the contents of doc/ in the README 13:15
ihrd and links to blogs where we mark progress
masak yes, links to all blog entries about this.
ihrd I will do my first post on this weekend
masak goodie. 13:17
I don't think it's bad that we're slightly out of phase.
I might take a week's vacation later instead and let you catch up :)
ihrd yes I will wright to TPF about my ailment 13:20
masak sure. 13:21
ihrd I do not recived any emails from Adrian from the 5 March 13:44
masak same here. 13:46
2009-03-05 was the last time he wrote to me.
ruoso ihrd, please consider putting the Request::Method in the end of the arguments 13:55
since the verb is most relevant to the last part of the uri
not to the first
and I also suggest adding a marking for the start of the uri 13:56
additionally, I'd consider having all the multi installed in a single variable 13:58
my &a = Multi.new
&a.push($candidate); 13:59
&a.(args...)
ihrd, that already works in rakudo.... 14:00
ihrd hi Daniel :)
ruoso hi ihrd
moritz_ wow, ten non-bots in here. It's getting crowded :-) 14:01
ihrd what benefits of this?
masak enjoys #november-wiki being crowded
bots are welcome too.
ruoso ihrd, one thing that was making me unconfortable was how the candidates would be installed in the namespace 14:02
putting it into a variable makes it cleaner
so the actions can be registered in the &dispatcher multi 14:03
so I asked jhntn about it, and he implemented Multi.new in rakudo
to the second point...
having the request method in the end of the arguments, make the verb close to the last part of the uri 14:04
which means that the uri identifies which resource and the last argument identifies which verb 14:05
this is mostly important to later support calling reduce with that multi
ihrd if I have class with methods how I can put it in this variable? 14:06
ruoso ihrd, you can always declare the method as my or our
and use it as &foo
ihrd I mean I like to have normal class with methods 14:07
ruoso I see... but why do you want that?
action dispatch is completely non-oo
ihrd because this make app maintable
if you have standrad API for resources you can use roles 14:08
ruoso by resources you mean what catalyst calls "action", right?
ihrd with tou resouces
no, I mean controller 14:09
ruoso the thing is that a "controller" is not a "resource"
it is simply a "package" of "actions
ihrd yes, this is not fully the same 14:10
ruoso I mean, it's completely different
ihrd this is about app organization
ruoso right...
ruoso brb &
ihrd I think we can put all resource methods in one calss 14:11
ruoso ihrd, but that is not really sane in OO terms, is it?
ihrd in this case you can resource as object
ruoso right...
that's my point 14:12
each "resource" is an object
ihrd and resource has methods -- GET, POST, PUT, DELETE
other methods is... like private actions
ruoso maybe... although I would leave that as an argument... and just call "execute'
because you can then have also "begin" and "end 14:13
which are part of the processing
masak as long as you adhere to Rack's interface here, I'm happy.
and that will probably be easy.
ruoso rack seems to be way before that 14:14
masak right.
it's quite low-level.
but also very reasonable.
ihrd before and us agreematn to interact
s/us/as/ 14:15
masak aye.
ruoso you just need to build the "arguments" from the request... then call the dispatcher
ok, let me put my perspective
I'd expect the multi dispatcher to work as a "reducer"
which means that it would return the "resource"/"action" object that is going to handle this request 14:16
so you can call .begin, .execute, .end
so, the dispatcher would look like 14:17
my @parts = $*app.request_split($*request);
my $action = @parts.reduce: &dispatcher;
masak ruoso: is there somewhere I can read about this .reduce thingy? 14:18
I feel I don't know anything about it.
ruoso my $error is context<rw>; try { $action.begin; $action.execute; CATCH { $error = $_; $_.handled = 1 }}; $action.end;
masak, that's the standard list reduce
ihrd this is one of the basics of functionsl programming
moritz_ the standard .reduce doesn't work with multi methods, does it? 14:19
ruoso moritz_, it is supposed to work...
masak ruoso: oh! that's ingenious, really.
ruoso moritz_, we still didn't really figure out how it will do it, tho 14:20
masak I just hadn't made the connection. I know the op form well.
ruoso moritz_, I proposed a way of doing it, which was to define a named argument to be sent to .postcircumfix:<( )>
moritz_ ruoso: yes, I know 14:21
ruoso masak, so you see what I mean now?
masak ruoso: oh yes.
moritz_ ruoso: and it bothers me a lot do base the dispatch on something that's neither specced nor implemented yet
ruoso that's why it's important to have a marker for the start and for the end of the args
moritz_, it can work without the reduce, for start... 14:22
it's not as cool...
but it still works
masak let's start by building something that works.
then we can refine it.
we're not getting a grant for building air castles. :)
ruoso: I definitely like the idea, though. 14:23
ruoso masak, I'm just trying to point it to that direction
masak aye, of course.
moritz_ first build a very simple MMD based dispatcher, like I proposed
then it's a piece of cake to plug in other dispatchers
ruoso which is why i say to put two special args
one on the start 14:24
moritz_ and the default dispatcher can be changed later on
ruoso and other in the end
it wont hurt now
and enables reduce-based dispatch later
putting the req method in the end, for that same reason 14:26
masak, do you see what I mean? 14:27
masak sure.
but it's not me you have to convince, it's ihrd. 14:28
he's the one writing the dispatcher.
ruoso ihrd, do you see what I mean? ;)
ihrd not fully :(
Ok, use reduce looks like elegant way. 14:29
and you just want agreement
about chunks
but I still do not see how you see full app 14:30
I like idea to make resources useful as objects
I feel we can use OOP here
ruoso specially compose roles into a specific resource 14:31
ihrd we can make resource inheritance, too
If I understand right you see action like free pices of code 14:34
but I thinking about structure for app 14:35
mb this is not so perlish 14:36
but when I work with real life project I like to have strict structure for web apps 14:37
do you understand what I mean? 14:39
(I need dic to understand you, but you have no dic from russianenglish mess to understand my) 14:40
ruoso role Resource { multi method begin {...}; multi method execute {...}; multi method end {...} } 14:41
ihrd, for start, I see some code overhead in order to register the resources... 14:42
but that can later be fixed by a sub-language....
controller Fooo { resource bar { } }
ihrd, the dispatcher is then responsible for finding the appropriate resource to that request 14:44
in order to do that, it first generates the arglist from the request
my @parts = my_splitter($*request); 14:45
after having the parts, it can invoke the multi
my $action = &resource_finder(|@parts)
when we have reduce-with-multi working, we can replace that by 14:46
my $action = @parts.reduce: &resource_finder;
ihrd uhu
ruoso then you can perform the action... 14:47
my $error is context<rw>; try { $action.begin; $action.execute; CATCH { $error = $_; $_.handled = 1 }}; $action.end;
actually $action.*begin, $action.*execute, $action.*end 14:48
so composed roles can have its own execute called
ihrd yes 14:51
ruoso ihrd, basically it's the same as the sketch I wrote 14:52
but replacing the regex-based stuff
ihrd yes, I see
ruoso for a multi-based stuff
ihrd I read you what you wrote carefull, and execute it in my head 14:53
masak moritz_: what's underspecced re multis and .reduce?
ruoso masak, it isn't yet... 14:54
masak it isn't underspecced yet?
ruoso but the general expectation is that it should be able to reduce calling a multi with varying-arity
moritz_ masak: there's no mention of what it does with multis
ruoso the most logical way of doing it is begin trying with the higher arity... 14:56
and if that fail try with lower....
masak ruoso: ok. that sounds vaguely like LTM to me.
ruoso yes... 14:57
moritz_ ruoso: why it is the most logical way?
masak moritz_: because it gives the long-paramlist ones a change before the short-paramlist ones match.
ruoso I don't really know... maybe I'm just contamined by LTM in regexes
masak this sounds like a question for p6l. 14:58
moritz_ the most intuitive way for me would be to do a multi dispatch that allows left over items, and pick the narrowest match
masak moritz_: 'left over items'? 14:59
masak is so glad purl isn't here
moritz_ masak: do a dispatch with an argument list, but not all arguments have to be used for the dispatch
so some trailing items can be neglected
masak ah, as in ;; ?
moritz_ jnthn will kill me
masak :P 15:00
moritz_ masak: kinda like. But ;; is in the parameter list, not in the argument list
masak oh, ah.
moritz_ and I don't know before where to place the ;;
masak right.
ruoso ihrd, do you see why we need two marker arguments in the start and in the end of the @parts? while the request method is in the end...
masak but please let's try to design something that doesn't require changes in the Perl 6 Spec. :) 15:01
ruoso moritz_, but even with LTM, it needs to support "left overs"
moritz_ ruoso: aye
anyway, I think this is too much effort for a rather small problem set
ihrd ruoso: to see start and end of the chain?
ruoso ihrd, yes 15:03
ihrd but I still in confusing. You have array ofactions 15:04
actions is multies
ruoso I don't have an array of actions
I have an array of uri parts
that is reduced, by a multi, to an action object 15:05
ihrd ah, yes. this multies just return an action object 15:07
ruoso at first, you will be required to have all the uri parts as args of your multi... 15:09
but when .reduce works with reduce
with multi
you can have the action you're chaining as the first argument, and only the arguments of your own action
like
&multi.push sub (RootMarker, 'foo', $bar) { return Action::Chained('/foo/*') }; 15:10
&multi.push sub (Action::Chained('/foo/*'), 'baz', $bla) { return Action::Chained('/foo/*/baz/*') }; 15:11
&multi.push sub (Action::Chained('/foo/*/baz/*'), Request::Method::GET) { return Action::Public('GET /foo/*/baz/*') }; 15:12
my $action = (RootMarker, 'foo', 123, 'baz', 456, Request::Method::GET).reduce: &multi; 15:13
ihrd in my Ctalyst apps I use Chained to do security checks and get some data 15:19
for example /company/1/offers
ruoso ihrd, I *only* use Chained actions
ihrd chained get comapny, and check can I see offers 15:20
in me 'Forest' sketch I am trying to make
multies work with this
I do not need stash to put geted stuff in it 15:21
I can just use it in the call to the multies
ruoso the stash
?
ihrd $c->stash->{company} 15:22
ruoso $c is no longer needed
we have context variables
$*request in place of $c->req
ihrd I am speaking about how that done in Catalyst. I just do not see how you actions share data. 15:23
ruoso by using context variables
ihrd context variables just variables you can get by CALLER::$foo or you mean something different? 15:25
ruoso not CALLER::$foo, simply $*foo 15:26
it's not just the CALLER scope
it's the whole dynamic scope
ihrd ok 15:29
I do not see what changed 15:31
I mean you still put you stuff in var
ruoso right... that's what variables are for 15:32
ihrd and I am speaking about how I drafted chained 15:33
in my draft you can just return from Link args for the next call 15:36
ruoso ihrd, what you designed looks like forward, not chained
ihrd hm 15:37
you can use $rest to change way in that sketch, yes 15:38
but idea is to use multies again. you can have multi Link (...) {...} and same resource can work different as specific part of the chain 15:40
15:40 sri_kraih_ joined
ruoso ihrd, it still looks like forward... which is the opposite of chain 15:42
ihrd I do not see. why? 15:43
ruoso because in chained actions, this action assumes the other was already executed 15:44
you don't have to think about it 15:45
ihrd can you show real-life example? 15:46
ruoso the docs in Catalyst::DispatchType::Chained explains it 15:47
ruoso later & 15:48
ihrd re-read this pod and still do not understand what I missed. Yes, actions in the chain assumes prev Chained actions executed before. This is right for my model too. 15:55
deep night, I am going to sleep. See you! 16:08
16:10 ihrd left 17:02 ruoso left, ruoso joined
sri_kraih_ hmm, sounds like you are porting catalyst 18:50
19:00 Tene joined
ruoso sri_kraih_, at least I'm pushing a lot in that direction 19:03
sri_kraih_ will be interesting to see how far you can get with the rack port 19:36
i'd like to get Mojo running on rakudo as soon as possible 19:38
not such a big rack fan personally, too much constraints, but it's a good first candidate for porting 19:44
Tene okay, working on tags tonight. 21:54
going home in an hour or two, gonna start then.

Powered by Cro and the Raku® Programming Language.

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

Thank you!