On Tue, 12 Jun 2007 15:37:38 -0400
Bill Sommerfeld <sommerfeld at sun.com> wrote:
> On Tue, 2007-06-12 at 12:25 -0700, Michael Hunter wrote:
> > Which reduces it to an implementation technique not substantially better
> > then saying "fork the source".
>
> Strongly disagree. Using a documented plugin interface in the way
> intended by its creator is substantially different from "forking", even
> if that plugin interface is Volatile in our terms (often they're
> Volatile in ABI terms but closer to Committed at source level, which
> means that building the plugin as part of the same consolidation as the
> framework will not involve substantial maintainance overhead.
If you hadn't said "strongly" I would have thought we were just
disagreeing on terms. But since you did...
In the normal case I agree that it provides two things:
1) less likelyhood of interface change due to its documentation and use
2) less likelyhood of merge issues due to implied source organization
The case in question is where we can't get the community to accept a
patch. I think that is correlated with the case where this interface
changes. Thus my "not substantially" and not agreeing with your
"strongly".
mph
