OpenLayers OpenLayers

Ticket #994 (closed bug: fixed)

Opened 1 year ago

Last modified 1 year ago

Google Layer is broken.

Reported by: crschmidt Assigned to: tschaub
Priority: blocker Milestone: 2.5 Release
Component: Layer.Google Version: 2.5 RC1
Keywords: Cc:
State:

Description

Looks like Google changed the internals, so dragging a google map breaks.

Attachments

google.patch (0.7 kB) - added by tschaub on 09/18/07 14:50:00.
use documented getContainer method
goog.patch (0.6 kB) - added by euzuro on 09/18/07 18:31:36.
vladimir

Change History

09/18/07 14:20:41 changed by tschaub

  • keywords set to review.
  • owner changed from euzuro to tschaub.
  • status changed from new to assigned.

Will commit w/ review.

09/18/07 14:28:44 changed by euzuro

good fix but is there a reason why we are changing where the function is tagged on?

09/18/07 14:34:15 changed by tschaub

yes - extending the mapObject prototype doesn't give us access to the layer - which is what you need to traverse down the dom

all tests pass and examples work in IE/FF

09/18/07 14:34:45 changed by tschaub

note that it is different to extend the prototype and to extend an object - thus the binding

09/18/07 14:37:03 changed by tschaub

and if you're trying the same with the prototype extension - all will work fine with one GMap instance per page - but will fail with more as the layer that gets bound to the prototype method is the last layer initiatiated

09/18/07 14:50:00 changed by tschaub

  • attachment google.patch added.

use documented getContainer method

09/18/07 14:50:32 changed by tschaub

New patch, easier on the eyes, and uses a documented method!!! Thanks Uz!

09/18/07 15:01:16 changed by euzuro

  • keywords changed from review to commit.

sweeeeet! new patch is excellent, cuts right to the heart of the problem. tests pass in ie6, ff. please commit, mark as pullup.

note that we're going to have to produce some retroactive patch for old versions of OL so that they can still use gmaps.

also note that since we're still doing firstChild.firstChild on the the result from getContainer(), we're still technically doing a little mucking about.. and maybe it's best to put a special disclaimer about this on the google layer. or something like that. just so people aren't freaked when google pulls out the rug like this.

also note that open source development *rules*! how quickly was this bug tracked down and stomped out? wicked.

09/18/07 15:04:56 changed by tschaub

(In [4375]) Use documented getContainer method instead of chasing obfuscated properties. Note that we are still tip-toeing around the dom here assuming they won't pull a fast one on us - Google users beware (see #994). Big thanks to Uz for help sleuthing this one out.

09/18/07 15:05:36 changed by tschaub

  • keywords changed from commit to pullup.

09/18/07 18:31:36 changed by euzuro

  • attachment goog.patch added.

vladimir

09/18/07 18:33:15 changed by euzuro

  • keywords changed from pullup to review.

latest patch fixes prototype problem (previous check would always fail) pleae review

09/18/07 19:05:52 changed by tschaub

  • keywords changed from review to commit.

Done reviewing. All green from FF/IE. Please make it happen.

09/18/07 19:47:41 changed by euzuro

(In [4380]) test to see if 'fromLatLngToContainerPixel' is defined on the prototype, not on the class. (See #994).

09/18/07 19:47:58 changed by euzuro

  • keywords changed from commit to pullup.

09/19/07 07:40:36 changed by crschmidt

  • keywords deleted.
  • status changed from assigned to closed.
  • resolution set to fixed.

r4390 closes this.