preliminary es53 taming in shindig

with this patch, osapi doesn't work, flash doesn't work, but most other things should.

I'm constructing the es53 frames with document.write because I thought it would be easier to debug than eval'ed script, but this causes somewhat odd behavior in chrome, so I'm rethinking that.

[Jasvir, 2011-04-06 17:58:10]

http://codereview.appspot.com/4373043/diff/1/features/src/main/javascript/fea...
File features/src/main/javascript/features/caja-guest-frame/feature.xml (right):

http://codereview.appspot.com/4373043/diff/1/features/src/main/javascript/fea...
features/src/main/javascript/features/caja-guest-frame/feature.xml:24: <script
src="res://com/google/caja/plugin/es53-guest-frame.js"/> -->
Extraneous -->

http://codereview.appspot.com/4373043/diff/1/features/src/main/javascript/fea...
File features/src/main/javascript/features/caja/taming.js (right):

http://codereview.appspot.com/4373043/diff/1/features/src/main/javascript/fea...
features/src/main/javascript/features/caja/taming.js:147: '-g___', uriCallback,
imports, gadgetBody);
We could pass a pseudo window location to attachDocumentStub here too.

http://codereview.appspot.com/4373043/diff/1/features/src/main/javascript/fea...
features/src/main/javascript/features/caja/taming.js:153: console.log([msg,
source, line]);
Is console guaranteed to be available?  I think it is undefined on FF at least
when the firebug window is closed.

http://codereview.appspot.com/4373043/diff/1/features/src/main/javascript/fea...
features/src/main/javascript/features/caja/taming.js:159: imports.opensocial =
tamingWin.___.tame(window.opensocial);
I think we should put the imports.opensocial and imports.osapi lines in
opensocial/taming.js and osapi/taming.js instead so they only get run if the
gadget requests opensocial or osapi.  Does tame throw on undefined args?

http://codereview.appspot.com/4373043/diff/1/features/src/main/javascript/fea...
features/src/main/javascript/features/caja/taming.js:160: imports.opensocial =
tamingWin.___.tame(window.osapi);
s/imports.opensocial/imports.osapi/

[felix8a, 2011-04-14 22:31:20]

updated snapshot

[felix8a, 2011-04-15 17:38:35]

new snapshot; pass caja=1 query parameter to JS servlet for cajoled gadgets
(otherwise taming files get stripped to reduce js size)

[felix8a, 2011-04-16 00:01:49]

gadgets.json now works

[Jasvir, 2011-04-18 15:57:18]

Looking good.

http://codereview.appspot.com/4373043/diff/13001/features/src/main/javascript...
File features/src/main/javascript/features/opensocial-reference/taming.js
(right):

http://codereview.appspot.com/4373043/diff/13001/features/src/main/javascript...
features/src/main/javascript/features/opensocial-reference/taming.js:25: //
TODO(felix8a)
Please include a description of the TODO

http://codereview.appspot.com/4373043/diff/13001/java/gadgets/src/main/java/o...
File
java/gadgets/src/main/java/org/apache/shindig/gadgets/render/RenderingGadgetRewriter.java
(right):

http://codereview.appspot.com/4373043/diff/13001/java/gadgets/src/main/java/o...
java/gadgets/src/main/java/org/apache/shindig/gadgets/render/RenderingGadgetRewriter.java:231:
if (!gadget.requiresCaja()) {
Is this a temporary workaround?  Doesn't this mean onload handlers won't fire if
the gadget is cajoled?

[felix8a, 2011-04-18 16:30:10]

http://codereview.appspot.com/4373043/diff/13001/features/src/main/javascript...
File features/src/main/javascript/features/opensocial-reference/taming.js
(right):

http://codereview.appspot.com/4373043/diff/13001/features/src/main/javascript...
features/src/main/javascript/features/opensocial-reference/taming.js:25: //
TODO(felix8a)
On 2011/04/18 15:57:18, jasvir wrote:
> Please include a description of the TODO

ok

http://codereview.appspot.com/4373043/diff/13001/java/gadgets/src/main/java/o...
File
java/gadgets/src/main/java/org/apache/shindig/gadgets/render/RenderingGadgetRewriter.java
(right):

http://codereview.appspot.com/4373043/diff/13001/java/gadgets/src/main/java/o...
java/gadgets/src/main/java/org/apache/shindig/gadgets/render/RenderingGadgetRewriter.java:231:
if (!gadget.requiresCaja()) {
On 2011/04/18 15:57:18, jasvir wrote:
> Is this a temporary workaround?  Doesn't this mean onload handlers won't fire
if
> the gadget is cajoled?

no, I'll add a comment.  in the non-cajoled case, code to fire the onload
handlers gets injected into the gadget iframe at the bottom.  however, in the
cajoled case, this might fire before the cajoled iframes have finished loading. 
in caja/taming.js I'm firing the onload handlers after everything has loaded.

[felix8a, 2011-04-18 16:37:09]

updated snapshot

[felix8a, 2011-04-19 00:50:52]

found a race condition: in some cases tamingFrame's done function will fail
because it gets triggered before the guestFrame is constructed.  new snapshot
ensures guestFrame is constructed first.

[felix8a, 2011-04-19 01:16:06]

in some situations, shindig will inline caja/taming.js in html.  this is a
problem because taming.js includes instances of the string "</script>", which
prematurely closes the script.

updated snapshot ensures "</script>" does not appear literally in taming.js

[felix8a, 2011-04-19 19:11:34]

oops, forgot to run mvn without tests disabled, some of the easymock checks
fail.  fixing...

[felix8a, 2011-04-19 21:52:50]

"mvn test" passes cleanly now

[Jasvir, 2011-04-19 23:12:29]

LGTM

http://codereview.appspot.com/4373043/diff/11003/java/gadgets/src/main/java/o...
File java/gadgets/src/main/java/org/apache/shindig/gadgets/Gadget.java (right):

http://codereview.appspot.com/4373043/diff/11003/java/gadgets/src/main/java/o...
java/gadgets/src/main/java/org/apache/shindig/gadgets/Gadget.java:228: return
getAllFeatures().contains("caja");
Nice.