feat($timeout): pass arguments in callback

[PatrickJS]

PR has updated Docs and Tests
setTimeout allows you to pass arguments into the callback function in
the …3rd argument. Since that’s taken I added it to the …4th
http://mdn.io/setTimeout#Syntax

here is an example of why you would need to do this

import {$http} from 'angular.js/modules';
import {setTimeout, Promise} from 'facade/async';
var cache = {};

export class AwesomeModule {
  static getSwag(url:string):Promise {
    return new Promise(function(resolve, reject) {
      var complete = function(data) {
        cache[url] = data;
        resolve(data);
      };

      if (url in cache) {
        setTimeout(complete, 0, cache[url]);
      } else {
        $http.get(url).then(complete, reject);
      } // end url in cache
    });
  } // end getSwag
} // end AwesomeModule

similar pull request #10632

[gkalpak]

Note that according to MDN, IE9 does not support the additional arguments.

[PatrickJS]

due to the way angular is handling setTimeout intervally this code works in IE9 even if setTimeout doesn't support it natively

[gkalpak]

@gdi2290: Good point ;) BTW, if it gets merged, it would be good to have the docs updatd as well (imo).

[PatrickJS]

I added docs

* @param {...*=} Pass additional parameters to the executed function.

[gkalpak]

I wonder if making the same optimisation as in $interval here would make any tangible difference. E.g.:

var hasArgs = arguments.length > 3;
...
deferred.resolve(hasArgs ? fn.apply(null, args) : fn());

[shahata]

I'd go the other way and make this change a simple one liner:

deferred.resolve(fn.apply(null, sliceArgs(arguments, 3)));

there's no need for this hasArgs variable and no need to test the arguments.length on your own, sliceArgs will take care of it just fine.

[PatrickJS]

I don't have reference to arguments in that scope so the best I can do is sliceArgs regardless of argument length

//...
var args = sliceArgs(arguments, 3)
//...
  //...
  deferred.resolve(fn.apply(null, args));

[shahata]

👍

[lgalfaso]

this and #10632 look good, but I would like to see a few more tests that make sure that ngMock also works as expected

[petebacondarwin]

LGTM

[petebacondarwin]

Feel free to merge

[PatrickJS]

I need to update this PR due to the API change of $timeout 5a60302 but I should be good to go soon

[PatrickJS]

@petebacondarwin PR updated with rebase from master branch. Resolving merge conflicts are a pain I have to say.

Since the $timeout overload doesn't affect this feature too much I didn't account for it in the code (no callback so no point providing noop it arguments). I could check if they're using method overload and use $browser.defer(noop, delay) rather than having it run all the logic needed for the original api. The other PR (5a60302) didn't do this optimization so I can assume it doesn't matter

[petebacondarwin]

Welcome to our world (rebase hell)!

I think it is fine not to optimize the case where no handler function is passed for now.

Merging