Adding @oncreate and @x-oncreate support to server-side templates

the @oncreate attribute was discussed during the spec discussions but ended up not making the cut for the spec.

It is still a very useful attribute to support, so this patch adds support for @oncreate and @x-oncreate (and variants with capital "C") - the proposal being that we only document @x-oncreate until we convince other containers that @oncreate is useful.

[levik, 2009-05-12 22:11:46]

the @oncreate attribute was discussed during the spec discussions but ended up
not making the cut for the spec.

It is still a very useful attribute to support, so this patch adds support for
@oncreate and @x-oncreate (and variants with capital "C") - the proposal being
that we only document @x-oncreate until we convince other containers that
@oncreate is useful.

[Evan Gilbert (code review), 2009-05-15 17:47:00]

http://codereview.appspot.com/65052/diff/1/2
File
java/gadgets/src/main/java/org/apache/shindig/gadgets/templates/DefaultTemplateProcessor.java
(right):

http://codereview.appspot.com/65052/diff/1/2#newcode446
Line 446: buffer.append("}).apply(e);");
Might be easier to add an ID to the original inserted node (or use existing ID).

Pseudo-code
String id = el.getAttribute("id")
if (id == null) {
  id = generateUniqueId();
  el.setAttribute(id);
}
....
buffer.append("}.apply(document.getElementById('" + jsEscape(id) + "')");

[levik, 2009-05-20 13:40:32]

http://codereview.appspot.com/65052/diff/1/2
File
java/gadgets/src/main/java/org/apache/shindig/gadgets/templates/DefaultTemplateProcessor.java
(right):

http://codereview.appspot.com/65052/diff/1/2#newcode446
Line 446: buffer.append("}).apply(e);");
On 2009/05/15 17:47:00, Evan Gilbert (code review) wrote:
> Might be easier to add an ID to the original inserted node (or use existing
ID).
> 
> Pseudo-code
> String id = el.getAttribute("id")
> if (id == null) {
>   id = generateUniqueId();
>   el.setAttribute(id);
> }
> ....
> buffer.append("}.apply(document.getElementById('" + jsEscape(id) + "')");
> 
> 

This would indeed be simpler, but would misfire for the edge case where existing
IDs are non-unique (which is frowned upon, but possible in HTML).

[Evan Gilbert (code review), 2009-05-20 16:52:55]

http://codereview.appspot.com/65052/diff/1/2
File
java/gadgets/src/main/java/org/apache/shindig/gadgets/templates/DefaultTemplateProcessor.java
(right):

http://codereview.appspot.com/65052/diff/1/2#newcode446
Line 446: buffer.append("}).apply(e);");
On 2009/05/20 13:40:32, levik wrote:
> On 2009/05/15 17:47:00, Evan Gilbert (code review) wrote:
> > Might be easier to add an ID to the original inserted node (or use existing
> ID).
> > 
> > Pseudo-code
> > String id = el.getAttribute("id")
> > if (id == null) {
> >   id = generateUniqueId();
> >   el.setAttribute(id);
> > }
> > ....
> > buffer.append("}.apply(document.getElementById('" + jsEscape(id) + "')");
> > 
> > 
> 
> This would indeed be simpler, but would misfire for the edge case where
existing
> IDs are non-unique (which is frowned upon, but possible in HTML).

I'm a little worried about the complexity and performance of creating and
removing elements while processing scripts on the page. These can be in repeated
loops, so there can be a *lot* of these calls. Also, this construction might be
harder to sanitize.

I'm also fine if this fails in cases where users have manually assigned
duplicate HTML element IDs. If you're using JavaScript this seems like a
reasonable standard to enforce.

[levik, 2009-05-20 22:05:20]

Oh, fine, have it your way :)

I liked how clever I was being, but if we want to have performance over support
of crazy misuse of HTML, so be it :)

On 2009/05/20 16:52:55, Evan Gilbert (code review) wrote:
> http://codereview.appspot.com/65052/diff/1/2
> File
>
java/gadgets/src/main/java/org/apache/shindig/gadgets/templates/DefaultTemplateProcessor.java
> (right):
> 
> http://codereview.appspot.com/65052/diff/1/2#newcode446
> Line 446: buffer.append("}).apply(e);");
> On 2009/05/20 13:40:32, levik wrote:
> > On 2009/05/15 17:47:00, Evan Gilbert (code review) wrote:
> > > Might be easier to add an ID to the original inserted node (or use
existing
> > ID).
> > > 
> > > Pseudo-code
> > > String id = el.getAttribute("id")
> > > if (id == null) {
> > >   id = generateUniqueId();
> > >   el.setAttribute(id);
> > > }
> > > ....
> > > buffer.append("}.apply(document.getElementById('" + jsEscape(id) + "')");
> > > 
> > > 
> > 
> > This would indeed be simpler, but would misfire for the edge case where
> existing
> > IDs are non-unique (which is frowned upon, but possible in HTML).
> 
> I'm a little worried about the complexity and performance of creating and
> removing elements while processing scripts on the page. These can be in
repeated
> loops, so there can be a *lot* of these calls. Also, this construction might
be
> harder to sanitize.
> 
> I'm also fine if this fails in cases where users have manually assigned
> duplicate HTML element IDs. If you're using JavaScript this seems like a
> reasonable standard to enforce.