Re: GNOME ABI review

[Jeff Waugh <jdub perkypants org>]

<quote who="Luis Villa">

> <devil's advocate>
> Is it time to admit that GNOME 2.8 should be GNOME 3 and break binary
> compat for a slew of new/improved developer APIs? Continued indefinite
> support of APIs we know to be badly broken is not really a great idea,
> assuming that when we do break compat we do it with very good new APIs
> (which is currently far from guaranteed, of course.)
> </>

*NO*! We don't have time.

The sane way to go about this is the iterate/deprecate cycle: Each release
gets a refurbished piece of the developer API puzzle, with the old one
stubbed over the new (if possible, or worthwhile), and deprecated. Hackers
can look forward to cool new (usable!) features in each release, and we can
concentrate on getting each piece of the puzzle right, instead of doing
another speedy, short-sighted hack around it.

Perhaps we could even state a goal of one - even two - major developer
platform problem areas fixed in each six month iteration. Once we've solved
the current problems, we could switch to integration of new features, etc.

It's hard to argue that we have really strong developer buy-in on the last
big, sudden, messy change. Let's not do it again soon.

- Jeff

-- 
linux.conf.au 2004: Adelaide, Australia         http://lca2004.linux.org.au/
 
     "I think a lot of the basis of the open source movement comes from
                procrastinating students." - Andrew Tridgell