CC: enable server-driven gadget metadata caching.

Cache is busted according to what the server says (currently, fixed to shindig.cache.xml.refreshInterval), with some adjustments to differences between server/client absolute times. Before, this cache was infinite (ie: as long container instance lives). Cache will not be busted for navigated and preloaded gadgets. For simplicity, gadgets need to be closed (unnavigated) and/or unloaded (if preloaded) for its cache to be possibly evicted. Cache busting is particularly needed for long-live AJAX container pages (ie: ones that don't refresh pages).

[mhermanto, 2010-11-25 01:07:39]

Added unit tests

[zhoresh, 2010-12-07 19:16:59]

http://codereview.appspot.com/3260041/diff/4001/features/src/main/javascript/...
File features/src/main/javascript/features/container/service.js (right):

http://codereview.appspot.com/3260041/diff/4001/features/src/main/javascript/...
features/src/main/javascript/features/container/service.js:120: + currentTimeMs;
This calculation ignores the time it took to get the metadata, or the time it
took for this function to be called.
Probably not a bad thing, but maybe add a comment about it.

http://codereview.appspot.com/3260041/diff/4001/features/src/test/javascript/...
File features/src/test/javascript/features/container/service_test.js (right):

http://codereview.appspot.com/3260041/diff/4001/features/src/test/javascript/...
features/src/test/javascript/features/container/service_test.js:115:
this.assertEquals(99, response['resp1.xml'].localExpireTimeMs);
Im confused, should that be 101 (and the next 102, 104, 108)?

[mhermanto, 2010-12-07 20:49:41]

Update patch

[mhermanto, 2010-12-07 20:49:48]

On Tue, Dec 7, 2010 at 11:16 AM, <zhoresh@gmail.com> wrote:

>
>
>
http://codereview.appspot.com/3260041/diff/4001/features/src/main/javascript/...
> File features/src/main/javascript/features/container/service.js (right):
>
>
>
http://codereview.appspot.com/3260041/diff/4001/features/src/main/javascript/...
> features/src/main/javascript/features/container/service.js:120: +
> currentTimeMs;
> This calculation ignores the time it took to get the metadata, or the
> time it took for this function to be called.
> Probably not a bad thing, but maybe add a comment about it.
>
> True, noted. I figure it'd be small. Delta/tolerance time will be so small
compared to the gadget metadata duration.


>
>
http://codereview.appspot.com/3260041/diff/4001/features/src/test/javascript/...
> File features/src/test/javascript/features/container/service_test.js
> (right):
>
>
>
http://codereview.appspot.com/3260041/diff/4001/features/src/test/javascript/...
> features/src/test/javascript/features/container/service_test.js:115:
> this.assertEquals(99, response['resp1.xml'].localExpireTimeMs);
> Im confused, should that be 101 (and the next 102, 104, 108)?
>
> Good catch. Corrected. Turned out that I accidentally had the
responseTimeMs and expireTimeMs.

Snapshot reuploaded.


> http://codereview.appspot.com/3260041/
>

[zhoresh, 2010-12-07 20:54:40]

LGTM (keep in mind I am a js beginner)