On Aug 15, 2012, at 9:21 AM, Guthmann, Scott wrote: >> 4. Classic Model (see Description 2) on SVN > > +1 non-binding. > > I do not see the strong evidence for moving to Git now. > > I wish we would delay this vote until the research can be done to establish > the Git options as *actual* choices that are supported through Apache. Lack > of support for Git makes the Git option a logical non-starter. Further, we > don't know if Apache Infra is just lacking volunteers before they support Git > as an option or if there is more to the story. Three volunteers in the Apache > Flex project: Om, Jeff Conrad, and Carlos Rovira stated they are willing to > do this research to help us remove risk from the project and also provide > support to Apache Infra should volunteer time be the only factor preventing > the use of Git as a standard supported option. I believe it makes sense to > delay this vote until these three report back on this list about what Apache > Infra needs to fully support Git before we consider Git to be a valid choice > that we should feel free to choose.
This is a good point: > Note 1: The switch to Git will occur when Apache Infra approves Git as an > official SCM option (and after any release that might be in-progress at that > time. > > Note 2: Requires finding volunteers to work with Apache Infra and Apache > Infra and Apache Incubator PMC approval. If Apache does not approve, votes > will be automatically re-cast for "SVN now and then on Git". The reality here is: Note 1: Git is an option as described on the various urls, but with limitations which make it not yet "official". Note 2: Requires volunteer work with Infra for the many tasks needed to make a Writable Git archive work in a way that follows Foundation rules to preserve the IP. There is still work to do, hooks to create, and runbooks to write. This VOTE is really a compound vote around. (A) Git vs. SVN - the Git and SVN hybrid that represents the ASF's current Git support was not an option. (B) Branching Strategy Along with a poll regarding Git volunteers. > > We have heard claims that future work planned will be too hard to do unless > we use Git - but this is work we haven't started on and nobody has actually > tried doing this work in SVN and failed, yet. (Therefore this is a logical > fallacy) > We have also heard that there are multitudes of interested developers who > will only work on our project if Git is used - who are they? How are you able > to accurately predict the future? (again, a logical fallacy) > > I will say that the supporters of option #9 have been persuasive with me > because they are such strong advocates of this clearly better tool (Git) and > model for this project. In my view, their reasoning makes me wish to delay > this decision and vote until after Git is a supported option because we may > have the ability to make Git a supported option ourselves. And, the decision > that Apache Infra has made to support SVN and not support Git may not be just > due to a lack of willing volunteers. Making the decision to move to a tool & > model that will *never be supported* by Apache Infra is a mistake that is > preventable. I think we need to find out more about what Apache Infra needs > and not consider Git to be a valid option before we know. Git support is provisional, find out what needs to be done from Infra, but be very, very patient. They are busy and the right person may not be present for weeks at a time. Regards, Dave > > Regards, > > Scott Guthmann
