Re: 16-bit page checksums for 9.2

On Mon, Feb 20, 2012 at 12:53 PM, Josh Berkus <josh@agliodbs.com> wrote:
> On 2/20/12 5:57 AM, Robert Haas wrote:
>> 3. You're rearranging the page header in a way that I find ugly and baroque.
>
> Guys, are we talking about an on-disk format change?  If so, this may
> need to be kicked out of 9.2 ...

Yes, we are.  Simon's gone to some pains to make it backward
compatible, but IMHO it's a lot messier and less future-proof than it
could be with some more work, and if we commit this patch than we WILL
be stuck with this for a very long time.

The fact is that, thus far, so real effort has been made by anyone to
evict anything from the current CommitFest.  I did that for the last
two cycles, but as the minutes for last year's development meeting
succinctly observed "Haas ... doesn't like being the schedule jerk".
My resolve to be less of a schedule jerk is being sorely tested,
though: aside from this patch, which has its share of issues, there's
also Alvaro's foreign key stuff, which probably needs a lot more
review than it has so far gotten, and several large patches from
Dimitri, which do cool stuff but need a lot more work, and Peter
Geoghegan's pg_stat_statements patch, which is awesome functionality
but sounds like it's still a little green, and parallel pg_dump, which
is waiting on a rebase after some refactoring I did to simplify
things, and pgsql_fdw, and Heikki's xlog insert scaling patch, which
fortunately seems to be in the capable hands of Fujii Masao and Jeff
Janes, but certainly isn't ready to go today.  Any of these
individually could probably eat up a solid week of my time, and then
there are the other 40 as-yet-undealt-with patches.

I said five weeks ago that it probably wouldn't hurt anything to let
the CommitFest run six weeks, and Tom opined that it would probably
require two months.  But the sad reality is that, after five weeks,
we're not even half done with this CommitFest.  That's partly because
most people did not review as much code as they submitted, partly
because a bunch of people played fast and loose with the submission
deadline, and partly because we didn't return any patches that still
needed major rehashing after the first round of review, leading to a
situation where supposedly code-complete patches are showing up for
the first time four or five weeks after the submission deadline.  Now
none of that is the fault of this patch specifically: honestly, if I
had to pick just two more patches to get committed before beta, this
would probably be one of them.  But that doesn't make me happy with
where it's at today, not to mention the fact that there are people who
submitted their code a lot earlier who still haven't gotten the
attention this patch has, which is still less than the attention a
patch of this magnitude probably needs.

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