API freeze timeline (was: [gst-devel] caps/caps2)

[David Schleef]

On Thu, Dec 04, 2003 at 02:20:43AM +0100, Benjamin Otte wrote:
> There is unfortunately quite some stuff that needs testing before we can
> agree on an API. The problem with lots of code is that noone has ever used
> it, so noone knows if the APIs make sense.
> URI handling and tagging stuff are offenders in the core there.
> The interfaces and superclasses are offenders in the plugins.
> So I'm not sure we can be done there until Rb and gst-player impelement
> that stuff.

API bugs may always be fixed.  We added functions to some API in 0.6.3.
It's just that bugs need to be fixed in a backward-compatible way.

> I think it is doable though it is probably timed bad. I've got a lot more
> time during the holidays that start exactly after the freeze point. I
> think there might be quite some other people with increased hacking
> activity during the holidays.

Freezing the API isn't something that requires a lot of work.  It's
about making hard decisions, merging existing code, and taking the
small amount of time to make minor adjustments that don't happen
otherwise.  Additional time invites one to procrastinate and/or start
new API projects, which is exactly the opposite of what we want.

Use your vacation time to rewrite spider instead.  It's significantly
more important than having a perfect API.



dave...