Re: Documentation improvements for partitioning

On Tue, Feb 21, 2017 at 2:03 AM, Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> I have to admit my reaction was similar to Simon's, meaning that the
> lack of docs is a problem, and that the limitations are kind of a
> surprise, and I wonder what other surprises there are.

Did you read my message upthread pointing out that the initial commit
contained hundreds of lines of documentation? I agree that it would
be bad if table partitioning got committed with no documentation, but
that did not happen.

> I am thinking this is a result of small teams, often from the same
> company, working on a features in isolation and then making them public.
> It is often not clear what decisions were made and why.

That also did not happen, or at least certainly not with this patch.
All of the discussion was public and on the mailing list. I never
communicated with Amit Langote off-list about this work, except
shortly before I committed his patches I added him on Skype and gave
him a heads up that I was planning to do so real soon. At no time
have the two of us worked for the same company. Also, the patch had 7
other reviewers credited in the commit messages spread across, I
think, 4 different companies.

I think the issue here might be that with this feature, as with some
other features I've committed over the last few years, the email
discussion got very long. On the one hand, that does make it hard for
others to keep up, but not because anything is secret, only because
reading hundreds of email messages takes a lot of time. However, the
large number of emails on a public mailing list makes it absolutely
clear that this wasn't developed in isolation and presented as a done
deal. It was written and rewritten multiple times in response to
feedback, not only from me but from other people who did take the time
to keep up with the discussion. As Ashutosh Bapat and Amit Langote
already pointed out, I even posted (on a separate thread with a clear
subject line) some thoughts about the overall design of this feature,
in response to concerns articulated on an unrelated thread by Tom and
Alvaro. I did that in an attempt to give people a separate thread on
which to discuss those issues - without having to dive into the main
thread where things were being hashed out in detail - but it got no
responses, either because the logic was unarguable or because nobody
took the time to write back.

I think there's certainly room to criticize cases where a feature is
designed, developed, and committed almost entirely by people who work
at a single company, or where a significant amount of discussion
happens off-list, but it would be difficult to find a more clear-cut
case of where that DIDN'T happen than this patch.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company