[webkit-reviews] review requested: [Bug 7012] Coerce javascript objects to appleScript compatible types prior to returning them to applescript : [Attachment 6237] proposed patch

[bugzilla-request-daemon at opendarwin.org]

Alexey Proskuryakov <ap at nypop.com> has asked  for review:
Bug 7012: Coerce javascript objects to appleScript compatible types prior to
returning them to applescript
http://bugzilla.opendarwin.org/show_bug.cgi?id=7012

Attachment 6237: proposed patch
http://bugzilla.opendarwin.org/attachment.cgi?id=6237&action=edit

------- Additional Comments from Alexey Proskuryakov <ap at nypop.com>
(In reply to comment #5)
> I would fall back to always returning a string in the "default" case instead
of
> the "Unknown JavaScript type: %d", jsValue->type()" error, unless this is
never
> reached.

  Yes, it should be never reached - the switch has cases for all primitive
JavaScript types (I think).

> The WebView.h header can't be changed until we get the API approved. You will

> need to use WebViewPrivate.h for this interface.

  Done.

> Coercing to list descriptors is easy to do using
> [NSAppleEventDescriptor listDescriptor] and insertDescriptor:atIndex:.

  Yes, that's easy - I just don't see how the end result should look like for
non-trivial JavaScript arrays. And adding a conversion that only works "most of
the time" is IMHO not a good idea...

  Still, one option to consider: only convert arrays that are not sparse, for
which element indices start with zero, and don't descend into them recursively
(to avoid circular references). What do you think?