Tom Lane wrote:
> I wrote:
> > Okay, I'll do a trial patch and we can see exactly how much has to be
> > added (at least among core and contrib) before deciding for sure.
>
> This compiles and passes regression tests. It looks like the main
> things there might be an argument for adding back to spi.h would be
> pg_type.h and builtins.h, as a very large proportion of the files
> using spi.h had to have those added. Comments?
They are both very lean, so no objections. I guess that the pg_type.h
inclusion is needed due to the predefined type OIDs, and it makes me
wonder whether it would be useful to have them in a separate header.
Not enough concern for the idea to even make it to Bruce's open items
mailbox ...
--
Alvaro Herrera http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support