Adrian Knoth wrote:
> On Sun, Dec 06, 2009 at 08:45:00PM +1030, Jonathan Woithe wrote:
>> for us and confusion for potential users? Or maybe this is just a red
>> herring due to my somewhat rudimentary understanding of svn's capabilities.
> SVN doesn't have capabilities, SVN just has limitations. ;) SCNR
> (if compared to git or mercurial)
I'm personally not opposed to moving towards a more capable distributed
VCS as it would simplify things. I do have difficulties with GIT though
as I don't seem to be able to understand it. How about bazaar?
> For the whole branch discussion: the 2.0 should be a bugfix only branch,
> and development should take place in the trunk. Only very special
> development should have it's own side branch, think of a series of
> commits for new drivers. These branches need to be merged back at some
> time, because maintaining multiple branches consumes developer
> A consequence: when new stuff doesn't break the trunk, don't branch.
> That is, if scons ENABLE_SOMETHING_NEW=false still makes the trunk
> compile, it's ok.
Maybe we should indeed maintain the "development in trunk" model for a
while and see what happens
One thing we do have to achieve is more regular releases. The problem
there is that this requires that trunk is at a fairly high quality level
when branching. Otherwise the testing process takes much too long. The
consequence of this quality level is that there cannot be major commits
to trunk if a release is nearing. This basically means that a developer
then has to branch for her(;))/himself.
Maybe a distributed vcs is a good idea in any case.