Re: More then 1600 columns?

Apologizes Tom I did not see that you had answered yes to my question about the hard limit.
You have all been very helpful, I will give up on the 1600+ columns and look into using hstore.

Cheers

- Mark

-----Original Message-----
From: pgsql-general-owner(at)postgresql(dot)org [mailto:pgsql-general-owner(at)postgresql(dot)org] On Behalf Of Tom Lane
Sent: Friday, November 12, 2010 11:09 AM
To: Mark Mitchell
Cc: pgsql-general(at)postgresql(dot)org
Subject: Re: [GENERAL] More then 1600 columns?

"Mark Mitchell" <mmitchell(at)riccagroup(dot)com> writes:
> I know storing in an array is possible but it makes it so much easier to query the data set when each element is in its own field. I had lots of comments on why I should not do this and the possible alternatives and I thank everyone for their input but no one answered the question about compiling with a higher block size to get more columns. Can anyone answer that?

Yes, I did answer it: there is no such compilation option.

If you were willing to run a very nonstandard version of Postgres, you
could try widening t_hoff (see src/include/access/htup.h) but there is
nobody who can tell you what the fallout from that might be. One big
concern that I would have is the likelihood of O(N^2) behavior on very
long query targetlists.

On the whole I think you'd be a lot better off looking into hstore,
especially the improved 9.0 version.

regards, tom lane