Excerpts from Peter Geoghegan's message of vie oct 14 15:36:32 -0300 2011:
> This evening, David Fetter described a problem to me that he was
> having building the Twitter FDW. It transpired that it was down to its
> dependence on an #include that was recently judged to be
> redundant.This seems like another reason to avoid using pgrminclude -
> even if we can be certain that the #includes are redundant within
> Postgres, we cannot be sure that they are redundant in third party
> code.
I'm not sure about this. I mean, surely the problem was easily detected
and fixed because the compile fails outright, yes?
I do take time on ocassion to do some header reorganization and remove
inclusions that are redundant or unnecessary. I don't want that work
thrown aside because we have to support some hypothetical third party
module that would fail to compile. (In fact, we purposedly removed some
header from spi.h because it was unnecessary).
--
Álvaro Herrera <alvherre@commandprompt.com>
The PostgreSQL Company - Command Prompt, Inc.
PostgreSQL Replication, Consulting, Custom Development, 24x7 support