Re: appendChild

Sheppy wrote:
> This comes from bug 425153 and 418356.  Nothing I see in those bugs
> says that the limitation is restricted to chrome documents, that I can
> find.  If this is in error, I'd appreciate it if someone would let me
> know for certain so I can revise the documentation appropriately.

Ah, I see.

The change in bug 425153 makes sure that any <script> loaded from a 
document that gets XPCNativeWrapper automation also gets 
XPCNativeWrapper automation.  It does affect whether script are loaded. 
  It does affect XBL loading: XBL that does not get XPCNativeWrapper 
automation cannot be applied to a node in a document that does.

The change in bug 418356 doesn't affect <script> loading behavior at 
all; it merely affects whether scripts loaded programmatically ia the 
subscript loader get XPCNativeWrapper automation.  They're forced to if 
the loading code does.

Neither one affects appendChild behavior in terms of whether the script 
executes; all that's affected is whether the executing script gets 
XPCNativeWrapper automation.  This is relevant only to chrome documents, 
of course.

The one recent change I know of that affected _whether_ scripts are 
loaded (or rather whether they're executed) is bug 424426.  This affects 
only documents in a signed jar, and only allows such documents to 
execute scripts whose principal subsumes the principal of the document 
(in practice, either chrome scripts or scripts from a signed jar on the 
same server and signed with the same signing certificate).  That change 
was made for 1.9.0, though.

-Boris