On Wed, 11 Jun 2008, Kevin Ballard wrote:
On Jun 11, 2008, at 12:57 PM, Todd Lipcon wrote:
My assumption is that the long-running-patch-series type projects (e.g. C
bindings, new Erlang bindings, Ruby rewrite, etc) are primarily undertaken
by a group of core developers who have no problem learning to use and using
git. The majority of contributions from non-core contributors are more
likely to be the one-diff JIRA submissions.
Speaking as the author of the ruby rewrite, I'm not actually a core
developer. I am a third-party contributor whose company uses thrift and
wanted better ruby libs, so I simply forked Kevin Clark's git repo and
started working. So even as just a contributor I have been able to work on a
long-running patch series, get feedback and code review from Kevin Clark (on
IRC), and post some information to this list about what I've been up to. If
we were using svn, what I've been doing would have been extremely difficult.
Sorry -- should have been a bit clearer on terminology. By "core
developer", I mean the smaller group of developers who write code for
thrift on a fairly regular basis. I'm another third-party contributor
whose company uses thrift, but I'd consider myself a "core developer" as
someone who is active on the mailing list, IRC, channel, etc.
This is in contrast, for example, to my use of other Apache projects like
Hadoop and Hbase. I submitted a patch to Hadoop last week that fixed a
particular bug, and using JIRA was perfect for that - it was a 40-50 line
change that was atomic and self-contained. Although I've now *contributed*
to that project, I would hardly consider myself a Hadoop developer.
-Todd
