On Jan 20, 2009, at 7:07 AM, Mark Struberg wrote:
Hi!
As I already tried to expain: it was never ment to use git for
productive changes - think about it more like a blog full with code ;)
Understood. However, there are other ways of accomplishing this. Which
would probably be more acceptable...
1) create an openwebbeans/sandbox in svn
2) generate a patch and post to the jira
3) discuss your intentions/proposed changes (which you've done), and
commit them in trunk. The community should be reviewing all changes.
It's not necessarily bad to discover disagreements, post-commit. These
disagreements can be resolved.
At Apache, all committers have earned the necessary karma to be fully
trusted. IIUC, this is un-Git like (at least it doesn't match my
understanding of the social norms in Git usage). We may be able to
work out acceptable usage of Git. I want to be sure we're avoiding
"private" communications about code and some usages of Git might lead
to the potential for private conversations. BTW, I'm certainly not
implying that this is your (or anybody else's) intent...
Back to the issue: how do we cope with the TCK code? Should I check
it in to SVN? It will compile, but cannot run due to API
incompatibilities between us and RI. But as long as we do not add
the <module> in the parent pom it will at least not break the build.
Guess we could add the JBoss snapshot repo to our builds... This would
cause problems during releases and as the TCK changes.
I will create a Jira and attach the TCK suite via patch in the
meantime.
Personally, I'd commit the changes
I saw Jukka will organise a GIT session on the ApacheCon EU. Maybe
we'll find some time there...
Sounds good. I'm hoping that I can attend, this year...
--kevan
