Here is the file: http://github.com/mattetti/macruby-httpwrapper/tree/master
I noticed I gave you some wrong info, currently the delegator doesn't receive all delegated messages. Instead it gets called with a custom method and the response gets passed to it. This currently only happens when the download is successful http://github.com/mattetti/macruby-httpwrapper/blob/93230af840d1cdc5124f36397c289a993d6dd164/macruby_http.rb#L222 That's something we probably should fix. I tried to make the delegation process easier than what obj-c does but maybe we need to support a more advanced mode. Anyways, I added Brian to the repo and I hope to a see a fully finished version when I come back from vacation ;) - Matt p.s: adding tests would be totally awesome, I'm just not sure how to address that yet. On Mon, Mar 30, 2009 at 4:36 PM, Brian Chapados <cha...@gmail.com> wrote: > 2009/3/30 Matt Aimonetti <mattaimone...@gmail.com>: > >> Does the delegate object also receive messages that would be sent to an > >> NSURLDownload? > > > > I'm actually using NSURLConnection and not NSURLDownload which seemed a > bit > > too simple when I looked at it. If the delegator is a proc, then it > > currently only gets called once the request finished properly. (I still > need > > to write the error handling etc...) > > If your delegator is an object or a class, all messages should be passed > > directly. > > Right... NSURLConnection is more flexible. > > >> Also, regarding multi-threading, couldn't you just use NSURLDownload > >> (which you're probably already using underneath) + > >> NSOperation/NSOperationQueue. > > > > I need to look into NSOperation/NSOperationQueue to see how it works, I'm > > still pretty new to Cocoa and I don't know all the tricks cool kids use > ;) > > The libraries are vast, and the NSOperation stuff is new in 10.5, so > there aren't as many examples floating around the net. > Just throwing it out there as a possibility. I'm not sure how hard it > would be to implement. I've only used NSOperation for really simple > stuff. > > > I'll do what Ernie and put what I have so far on GitHub and let people > hack > > on it until we have something nice that could be used in most use cases. > > Since I'm going on vacation, I'll add you (Brian) to my repo so you can > deal > > with pull requests and push your own stuff there ;) > > So, you are assigning me as the delegate? ;-) > > > Reharding using the class outside of HotCocoa, I agree, but you should be > > able to only require the file and include the module. What do you think? > > Yes, that makes sense. It sounds like you've got a good start. I > agree with your goal of just making the standard use cases easy. It is > nice to have a block right next to the download call, as it is easy to > understand the logic. If something requires more complicated > handling, then delegate option is always available. Anyway, I'll see > if I can make any progress on NSOperation and error-handling. > > Brian > > > > > On Mon, Mar 30, 2009 at 2:46 PM, Brian Chapados <cha...@gmail.com> > wrote: > >> > >> Seems like a good idea that would be useful even outside of HotCocoa. > >> Especially for simple cases, I think if Obj-C had blocks, this would > >> be a good place to use them. For more control, the delegate pattern > >> actually works really well. It's good that you have a way to do both. > >> Does the delegate object also receive messages that would be sent to > >> an NSURLDownload? That might be a nice way to allow easy integration > >> in cases where an object that responds to the delegate methods is > >> defined in Obj-C. > >> > >> Also, regarding multi-threading, couldn't you just use NSURLDownload > >> (which you're probably already using underneath) + > >> NSOperation/NSOperationQueue. Since your #download method already > >> wraps the underlying process, you could should able to generate > >> NSOperations to perform the downloads and add them to a queue. Those > >> classes handle creating threads (which have their own runloops) and > >> performing a task on the thread. That said, maybe this approach would > >> not work with the current IO implementation? > >> > >> Brian > >> > >> 2009/3/30 Matt Aimonetti <mattaimone...@gmail.com>: > >> > In my free time, I've been working on a http://lighthouseapp.comclient > >> > app > >> > which mixes API calls with webkit views (I'll release the source code > >> > later > >> > on). > >> > The app is 100% HotCocoa (meaning that I didn't use any nib/xibs). > >> > For people who are not used to work with delegation, having to deal > with > >> > file download, URL queries can be a bit challenging if you want to > start > >> > using the underlying obj-c methods. > >> > > >> > That's why I wrote a small HTTP wrapper to clean up my code and make > >> > download/queries easier. > >> > The library is divided in 2 major parts, a helper and the > Query/Response > >> > classes. > >> > > >> > Here is how the helper looks like so far: > >> > > >> > Download helper making file download trivial > >> > Pass an url, and a block or a delegator > >> > A block takes precedence over a delegator, meaning that you can't > >> > pass a delegator and use a block, you have to choose which approach > >> > to pick. > >> > > >> > ==== Parameters > >> > url<String>:: url of the resource to download > >> > options<Hash>:: optional options used for the query > >> > &block<block>:: block which is called when the file is downloaded > >> > > >> > ==== Options > >> > :method<String, Symbol>:: An optional value which represents the HTTP > >> > method to use > >> > :payload<String>:: - data to pass to a POST, PUT, DELETE query. > >> > :delegation<Object>:: - class or instance to call when the file is > >> > downloaded. > >> > the handle_query_response method will be > called > >> > on > >> > the > >> > passed object > >> > :save_to<String>:: - path to save the response to > >> > :credential<Hash>:: - should contains the :user key and :password > key > >> > By default the credential will be saved for > the > >> > entire session > >> > > >> > TODO > >> > :progress <Proc>:: Proc to call everytime the downloader makes > some > >> > progress > >> > :cache_storage > >> > :custom headers > >> > Error handling > >> > > >> > ==== Examples > >> > > >> > download("http://www.macruby.org/files/MacRuby%200.4.zip";, > {:save_to > >> > => > >> > '~/tmp/macruby.zip'.stringByStandardizingPath}) do |macruby| > >> > NSLog("file downloaded!") > >> > end > >> > > >> > download "http://macruby.org"; do |mr| > >> > # The response object has 3 accessors: status_code, headers and > >> > body > >> > NSLog("status: #{mr.status_code}, Headers: > #{mr.headers.inspect}") > >> > end > >> > > >> > path = File.expand_path('~/macruby_tmp.html') > >> > window :frame => [100, 100, 500, 500], :title => "HotCocoa" do > |win| > >> > download("http://macruby.org";, {:save_to => path}) do |homepage| > >> > win << label(:text => "status code: #{homepage.status_code}", > >> > :layout > >> > => {:start => false}) > >> > win << label(:text => "Headers: #{homepage.headers.inspect}", > >> > :layout > >> > => {:start => false}) > >> > end > >> > end > >> > > >> > download "http://macruby.org/users/matt";, :delegation => > >> > @downloaded_file, :method => 'PUT', :payload => {:name => 'matt > >> > aimonetti'} > >> > > >> > download "http://macruby.org/roadmap.xml";, :delegation => self > >> > > >> > download "http://localhost:5984/couchrest-test/";, :method => > 'POST', > >> > :payload => '{"user":"mattaimone...@gmail.com","zip":92129}', > >> > :delegation => > >> > self > >> > > >> > download("http://mysite.com/login";, {:credential => {:user => > 'me', > >> > :password => 's3krit'}}) do |test| > >> > NSLog("response received: #{test.headers} #{test.status_code}") > >> > end > >> > > >> > > >> > As you can see from the example list, the method name download isn't > >> > always > >> > appropriate and maybe I should give it the http_query alias. What do > you > >> > think? > >> > What's nice with this approach, is that on top of being simple, the > >> > query > >> > and the response handling are async so the rest of your code can be > run > >> > and > >> > you get notified only when the response comes back. > >> > > >> > The underlying API looks like that: MacRubyHTTP::Query.new( url, > >> > http_method='GET', options={} ) > >> > Queries don't run in a custom runloop yet but that's something I want > to > >> > look at once IO work on the experimental branch will be done. > >> > > >> > > >> > So, here is my question, do you guys think that this is something that > >> > should be part of HotCocoa or it's better if it stays its own separate > >> > lib? > >> > Also, what do you think about having a default vendor folder for > >> > hotcocoa > >> > apps and autoload the vendor files if available? > >> > > >> > - Matt > >> > > >> > > >> > _______________________________________________ > >> > MacRuby-devel mailing list > >> > MacRuby-devel@lists.macosforge.org > >> > http://lists.macosforge.org/mailman/listinfo.cgi/macruby-devel > >> > > >> > > >> _______________________________________________ > >> MacRuby-devel mailing list > >> MacRuby-devel@lists.macosforge.org > >> http://lists.macosforge.org/mailman/listinfo.cgi/macruby-devel > > > > > > _______________________________________________ > > MacRuby-devel mailing list > > MacRuby-devel@lists.macosforge.org > > http://lists.macosforge.org/mailman/listinfo.cgi/macruby-devel > > > > > _______________________________________________ > MacRuby-devel mailing list > MacRuby-devel@lists.macosforge.org > http://lists.macosforge.org/mailman/listinfo.cgi/macruby-devel >
_______________________________________________ MacRuby-devel mailing list MacRuby-devel@lists.macosforge.org http://lists.macosforge.org/mailman/listinfo.cgi/macruby-devel
