Chris Anderson wrote:
On Mon, May 4, 2009 at 8:36 PM, Paul Davis <[email protected]> wrote:
Noah Slater wrote:
On Mon, May 04, 2009 at 06:32:04PM -0700, Chris Anderson wrote:
Are we ready for 0.9.1? My pet patch is in and backported, how about
yours?
Mine are done.
Prefacing with: Version names are called $(MAJOR).$(MINOR).$(REVISION)
I am so confused!
Damien convinced me pretty well that $(REVISION) numbers should only get
bumped when we're backporting a fix for a bug that shouldn't be in that
$(MINOR) version number. Ie, end user code wouldn't have to change. I'm
definitely guilty of backporting code that would break client code based on
error reporting strictness etc, but adding new features definitely seems
greater than $(REVISION) changing importance.
So, I feel that I'm entirely over-thinking this entire issue, but at the
moment I'd probably lean towards releasing 0.10.0 from trunk as opposed to a
new $(REVISION) release. Either way we should probably try and codify the
rules for backporting and put up a wiki page with some guidance on when we
do what. My first litmus test is "If it's visible from a client library
perspective, it's at least a $(MINOR) revision change without overwhelming
support."
I think we're pretty clearly not ready for 0.10.0
If there's breaking changes between 0.9.0 and the 0.9.x branch, they
should be reverted, as some users may decide to track the 0.9.x branch
to avoid just those as they appear in trunk.
This is definitely one of those "No matter what you decide, someone will
be upset" type of conversations. Seeing as I don't generally run
anything less than trunk I'm not overly invested, but I don't want to
irritate anyone that's following with any specific version.
That said, what keeps us from 0.10? I did contemplate the the jump from
0.9 isn't as big as the jump from 0.8. Though, I would probably say that
the jump from 0.8 to 0.9 was fairly delayed. At the moment there are at
least three new features: config.d updates, bulk=ok, and the reduce
warnings; none of which seem like only a $(REVISON) change. To me that
doesn't seem like something to ignore but I would be perfectly happy
referring to version numbers via subversion revision so I'm a bit not
normal on that front.
Just to restate, I don't really care except to minimize work for Noah on
the release aspect and to have some publicly visible logic in what goes
in a $(MINOR) vs. $(REVISION) branch so people know what to expect.
HTH,
Paul Davis
Anyway, that's just my two hundredths of a greenback.
Paul Davis
