[Scummvm-devel] ScummVM 0.13.0 has been released. 0.13.1 plans
max at quendi.de
Sun Mar 1 19:24:51 CET 2009
Am 01.03.2009 um 12:44 schrieb Johannes Schickel:
> Max Horn schrieb:
>> * I think trying the RC system out is worth a try. Many projects use
>> that, including e.g. the Linux kernel. Why not us. If you dislike the
>> RC term, then let's call it prerelease build or something. But still,
>> a coordinate effort to make binary built from more or less the same
>> source revision available to people would help. Esp. the fact that
>> prerelease builds would be made from the same revision, because that
>> way it would be clearer which bug fixes are already in each build and
>> which not. If we can't coordinate building from the same rev, then at
>> least each build should contain the SVn rev it was made from.
> IMHO we should really try to use the same rev for those. At least I
> would find it strange when we would call a build say "RC1" and all
> used a different revision to built from and of course it would be the
> point of having an RC to release a fixed revision for most (every?)
> out there so people could test it.
Of course if we use *RC-builds*, then all those must be built from the
same source, in fact, there should be a SVN tag for each RC release.
My point was that we might not end up using strict RC builds, but
rather less-formal "release builds", for the simple reason that it
might turn out impossible to organize coherent RC builds for ports not
using the SDL backend. Typical example from history: We tag an RC (a
release), but port X in it is not actually compilable. Result: That
port can either (a) not participate in that RC round, or (b) needs to
build from other sources than the rest. Case (b) is highly
undesirable, we seem to agree on that. And case (a) renders the whole
point of RCs moot if too many ports are affected.
Anyway, we have to try it to see how it works out. And I might see
things to negatively based on my past experiences. *shrug*
More information about the Scummvm-devel