On Feb 4, 2009, at 11:36 PM, Chris Anderson wrote:
On Wed, Feb 4, 2009 at 11:24 PM, Ted Leung <[email protected]> wrote:
Yes, real time communication is useful, and is permitted, but the
actual act
of making/finalizing the decision is to be done on the mailing
lists and
recorded in the public mailing list archives. The expected answer
to Geir's
query is a mailing list archive url. In this case, a message
saying "we
discussed this in IRC, the major points pro/con were <pro>/<con>
and we've
agreed to do <decision>. You can see the full discussion at <url to
relevant IRC log>.", would suffice.
It's hard to say when the full discussion would have taken place. The
decision to make the CouchDB API invariant no matter the number of
nodes supporting a database is basically what drew me to the project
in the first place.
I can't speak for the other committers, but I agree with the vision of
CouchDB as database that can scale to any number of host nodes while
presenting the same API to its clients.
I'm not questioning the actual decision itself. It's the process
that I am concerned about. If this was decided during the time that
CouchDB was at Apache, then we (speaking as an ASF member) expect that
project to make decisions in the same way that every other Apache
project makes its decisions. I thought that this was clear when I
recommended CouchDB for graduation. I'm a little unhappy that we're
needing a review here.
Ted
