On Thu, 8 Feb 2001, Tom Lane wrote:
> Alex Pilosov <alex@acecape.com> writes:
> > when I include files needed for SPI, it drags also a lot of other
> > garbage in, which conflicts with other things (namely, trying to get a
> > file to simultaneously include SPI and perl headers is impossible).
> > I realise it might be a lot of pain to clean up, but, you may consider
> > having a separate top-level include for SPI, which would not define (by
> > default) things like DEBUG, USE_LOCALE, union semun, etc.
>
> Unless you want to write SPI code that never calls elog(), it's gonna be
> tough to avoid the conflict on DEBUG. I suppose sooner or later we'll
> have to rename the elog severity symbols ... but it's not a change that
> I'm looking forward to making.
Yes, how about ELOG_DEBUG, etc? Global search and replace should get it
correctly, shouldn't be TOO painful.
It would indeed break other people's internally-written modules, I don't
know a good solution. Maybe do it like perl, have a #define ELOG_POLLUTE
which would pull in defined by their old names?
> In any case there's too much time pressure to consider wide-ranging code
> changes for 7.1. Right now I don't want to do more than rearrange the
> contents of a small number of include files.
Oh no, I'd plan SPI cleanup for 7.2 timeframe, if possible.
--
--
Alex Pilosov | http://www.acecape.com/dsl
CTO - Acecape, Inc. | AceDSL:The best ADSL in Bell Atlantic area
325 W 38 St. Suite 1005 | (Stealth Marketing Works! :)
New York, NY 10018 |