On Sep 11, 2006, at 11:51 PM, Rafael Schloming wrote:
Steve Vinoski wrote:
Hi all, I've made some progress with maven and qpid. The only
mildly difficult parts so far have been the code generation parts,
mainly because I've been experimenting with various ways to
implement them in maven.
Does anybody know the mechanisms by which we'll be dropping the
code to Apache? Specifically, will we transfer the svn repository
in its entirety, or will we just be copying files off the trunk
and creating a new repository?
We can't transfer the repository in its entirety because it
includes BDB based code. I would also prefer to preserve the
revision history where possible, so I'd like to avoid just copying
the trunk. I'm pretty sure it would be straightforward to simply
exclude the bdb message store implementation from the dump with
svndumpfilter, but I haven't tried it yet and I don't know what
other code if any needs to be excluded.
The reason I ask is that I'd like to create a maven branch to
allow others to see the changes and try them out for themselves,
sooner rather than later. This approach would allow the group to
get comfortable with the changes and choose to switch over to them
if and when we collectively decide to do so.
Are there any structural changes, or could you just post the maven
files?
There are minor structural changes required to conform to the
directory structure preferred by maven, such as moving src/org
directories to src/main/java/org, moving non-Java files to src/main/
resources, etc. Following the default directory structure, while not
strictly required, makes for much smaller pom files and conforms to
the Principle of Least Surprise.
If we're planning on an "svnadmin dump" of the current svn
repository to move to Apache, then creating a branch now won't be
that big of a deal, as it'll just go into the dump along with
everything else. If however we're doing some other kind of copy,
then perhaps holding off on creating the branch until after we hit
the incubator would be better.
If we try to preserve history then another branch would add a
little work since every extra branch produces an additional set of
paths that need to be excluded in order to produce a clean dump to
import into apache, but it's not a huge deal either way.
Well, if the branch has to be excluded to get the code over to
Apache, then it doesn't make sense to create it now, as we would just
have to create it again once the code gets over there. It makes sense
to do it now only if it can go with the dump.
thanks,
--steve
