On 21/04/2015 6:55 AM, Jacopo Cappellato wrote:
On Apr 21, 2015, at 11:23 AM, Jacques Le Roux <jacques.le.r...@les7arts.com>
wrote:
Switching the main repo to Git does not add anything to the project itself. As I said
before, Subversion handles our simple needs just fine. If Jacopo said something like,
"Managing our releases is impossible with Subversion, we really need to switch to
Git" - then we wouldn't be having this discussion, it would just happen because the
need for the switch is obvious.
But Jacopo is not saying that, and we don't have a problem using Subversion to
manage the project.
I don't remember Jacopo proposed to move to Git, it was Hans, then Taher, Adam
and Ean seconded. I guess Taher, AntWeb and Brainfood are using Git and they
would like to see OFBiz using it as well.
In fact I didn't express my opinion or preference.
At Hotwax we use Git for all our projects and we are happy with it. But this
doesn't mean that svn is bad or old or not a good tool for OFBiz.
In fact, as I mentioned to Hans at ApacheCon, before discussing "svn vs git" as
mere tools, it would be more interesting and useful to review/discuss the
contribution/commit workflows that these tools can enable and see if there is a room for
a better workflow for the project.
Jacopo
I think that this is a very good summary of the issue.
It appears to me that there is a fundamental shift of philosophy that
"Move to git" is discussing rather than a tool switch.
With git it is much easier to develop parallel "products".
I could chose to follow Jacopo's public repo or Ean's rather than the
"official" OFBiz repo if I felt that they were doing things that I liked
that were not being committed to the OFBiz repo.
I could take changes from more than one repo if I felt that I could
handle the integration workload and maintain a blended product.
I could make changes and request that any other repo that wanted to have
them, take them but I have no guarantee that Jacopo, Ean or OFBiz would
accept them.
I could use the Apache JIRA to document my patch or not.
Clearly a JIRA issue would get more discussion and might cause me to
revise my "contribution" or make it more likely to be accepted if the
concensus is that it is a "good thing".
I could create my own issue tracking system and invite others to look
there for my ideas.
Each one of the versions could have its own release schedule.
This is substantially different that the current situation and I think
that Jacopo's suggested course of action of looking at the proposal as a
question of workflow is a good idea.
Underlying this discussion, there is a dissatisfaction with the workflow
expressed by people who are not committers (in general).
There are a lot of private forks of OFBiz already but most of the ones
that show up here are based on the OFBiz trunk and are produced by
committers who contribute back some of their local changes.
In some ways, moving to a more distributed system, does help make OFBiz
stronger. The keepers of the OFBiz repo have control as they do now
about what changes get incorporated into the OFBiz "trunk". Others can
continue to use the contributions to the trunk but are still able to use
"improvements" from other repos that have been rejected by the community.
There are licensing issues that I would have to deal with as a
maintainer of a private repo if I start to include work from non-Apache
repos since I can not be quite so sure that the contributions to the
private repo have been released to me under an Apache license.
I am not advocating a shift to git, just trying to support Jacopo's
suggestion of looking at workflow first with a slight addition of
extending the discussion to project management philosophy since that
seems to be a key motivation underlying the suggestion to move.
Ron
--
Ron Wheeler
President
Artifact Software Inc
email: rwhee...@artifact-software.com
skype: ronaldmwheeler
phone: 866-970-2435, ext 102
