Bug 20632
| Summary: | WebKitWebFrame should be replaced with GdomFrame | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Luke Kenneth Casson Leighton <lkcl> |
| Component: | WebKitGTK | Assignee: | Nobody <webkit-unassigned> |
| Status: | RESOLVED INVALID | ||
| Severity: | Normal | CC: | alp, mrobinson |
| Priority: | P2 | ||
| Version: | 528+ (Nightly build) | ||
| Hardware: | PC | ||
| OS: | OS X 10.5 | ||
Luke Kenneth Casson Leighton
... or the other way round.
GdomFrame contains auto-generated the complete set of bindings, whereas WebKitWebFrame is manually created glib code.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |
Alp Toker
(In reply to comment #0)
> ... or the other way round.
> GdomFrame contains auto-generated the complete set of bindings, whereas
> WebKitWebFrame is manually created glib code.
>
The core WebKit/GTK+ API is intentionally hand-written to provide a tailored interface for developers. There isn't anything in the DOM that would work as a good replacement for WebKitWebFrame without customisation so the convention in WebKit ports is to provide accessors to the DOM binding from the core API rather than trying to use the DOM directly for all interaction with the browser engine.
Luke Kenneth Casson Leighton
sorry, i'd forgotten about this bugreport - i've since worked out
that a much better suggestion would be for WebKitWebFrame to _use_
GdomFrame.
i.e. rather than having a Frame* have a GdomFrame*.
the reasoning is that there are a number of _functions_ which are
hand-created in WebKitWebFrame which could be replaced with
the gdom_frame_* equivalents.
plus, in some cases, especially the Location* ones - i remember
seeing someone talking on the irc about doing a URL class or
something, for WebKitWebFrame - of course, you now have gdom_location_*
ok - actually, you _don't_ have gdom_location_* yet because it's
a custom function in JSLocation which i haven't cut/paste over
to GDOM* yet.... )
... but you get the idea.
also, it would make it clearer to the qt porters that the possibility
exists to do exactly the same trick in there. yes, they might
squawk a bit at #include <gobject.h> their competitor, but that would
kick them into gear to do a Qt bindings generator.
if WebKitWebFrame itself continues to have a Frame* then it remains
unclear as to exactly how other ports should go about adding in the
gdom bindings.
Martin Robinson
I think that we want to follow the behavior of other ports here and keep them as seperate entities.