Re: knngist patch support

On Fri, 12 Feb 2010, Robert Haas wrote:

> 2010/2/11 Oleg Bartunov <oleg(at)sai(dot)msu(dot)su>:
> > On Thu, 11 Feb 2010, Robert Haas wrote:
> >
> >> On Thu, Feb 11, 2010 at 3:00 AM, Oleg Bartunov <oleg(at)sai(dot)msu(dot)su> wrote:
> >>>>
> >>>> version I saw hadn't any documentation whatever. =A0It's not committab=
> le
> >>>> on documentation grounds alone, even if everybody was satisfied about
> >>>> the code.
> >>>
> >>> well, there is enough documentation to review patch.
> >>
> >> Where is there any documentation at all? =A0There are no changes to doc/
> >> at all; no README; and not even a lengthy comment block anywhere that
> >> I saw. =A0Nor did the email in which the patch was submitted clearly lay
> >> out the design of the feature.
> >
> > Well, initial knngist announce
> > http://archives.postgresql.org/pgsql-hackers/2009-11/msg01547.php
> > isn't enough to review ? We made test data available to reproduce results=
> ,
> > see http://www.sai.msu.su/~megera/wiki/2009-11-25
> > We are here and open to any reviewer's question.
>
> Well, just for example, that doesn't document the changes you made to
> the opclass infrastructure, and the reasons for those decisions.
> Actually, I've been working on an email on that topic but haven't
> gotten around to finishing it. There's some description of the
> overall goal of the feature but not much about how you got there. Is
> it enough to review? Perhaps, but it would certainly be nice to have
> some more discussion of the overall design. IMHO, anyway.

This is not fair,Robert. Everything was discussed in -hackers.I assume reviewer
should follow discussion at least, he is a member of our community. Mailing
list archive was/is/will our the best knowledge base. For example, regarding
changes in the opclass infrastructure you complain, you can see your reply
to Teodor's message http://archives.postgresql.org/message-id/603c8f070912292255u30e1983bi22ed5778bd2ce890@mail.gmail.com
which contains description of amcanorderbyop flag.

Frankly, I think we see here limit of our resources. Let me explain this.
We splitted patch by several parts - 2 parts are about contrib modules
(rather trivial), 1 - is a gist changes, 1 - planner changes and 1 part -
some proc changes. The most serious are gist and planner patches. We develop
GiST for many years and know almost everything there and could say that we're
responsible for GiST. I don't know if anybody from -hackers could review our
patch for planner better than Tom, but he is busy and will be busy.
So, any serious feature, which touch planner doomed to be rejected because of
lack of reviewer.

We tried to find compromise for 9.0 (Tom suggests contrib module), but all
variants are ugly and bring incompatibility in future. If there are no hackers
willing/capable to review our patches, then, please, help us how to save
neighbourhood search without incompatibility in future.

Regards,
Oleg
_____________________________________________________________
Oleg Bartunov, Research Scientist, Head of AstroNet (www.astronet.ru)
Sternberg Astronomical Institute, Moscow University, Russia
Internet: oleg(at)sai(dot)msu(dot)su, http://www.sai.msu.su/~megera/
phone: +007(495)939-16-83, +007(495)939-23-83