Peter Sorotokin wrote:
>
> > > Would it be feasible for Mozila to change IID of *any* interface every
> > > time its vtable layout is changed (this would not seem to require
> anything
> > > but IID string change in the same IDL file)?
> >
> > Are you also hoping that the old interfaces will still be available, or
>
> At the very least, I want to detect that interface is not available and tell
> user
> that my plugin will not work with this version of the browser. [...]
It seems reasonable to require anyone making an incompatible interface
change to use a new IID. Since the interface is incompatible, doing so
shouldn't make anything worse. This would also eliminate a lot of
potential crash bugs for a low cost. It would require looking at all
the interface clients in the tree and changing the IID there too, but I
see that as a win. Perhaps we could make it an approval requirement for
checking in incompatible interface changes? Someone spank me if I've
missed something.
> > sounds to me that the real win will be to put effort into reviewing and
> > freezing more interfaces.
>
> Yes, but who is supposed to do that? Do you suggest people like me (not
> very knowlegable in Mozilla internals) should take part in that?
I haven't been around in awhile, so I'm not sure what processes ended up
getting implemented (can someone help me out?) The embedding people
have a lot of interest in freezing interfaces and they have some docs up
that suggest they have a review procedure:
http://www.mozilla.org/projects/embedding/PublicAPIs.html
http://www.mozilla.org/projects/embedding/apiReviewNotes.html
I (speaking for myself, as always) see no reason why you shouldn't be
able to take some part in reviewing. As an end user of the interfaces,
I'm sure your feedback would be welcomed.
- Mike.
