On Jan 27, 2008, at 18:43, Tom Lane wrote:
> Oh, it finds it all right. It's just complaining (not incorrectly)
> that
> uuid.h fails when included after <unistd.h>. AFAICT this is just
> cosmetic, since we don't use it that way. If you go ahead and build
> then everything should be fine.
Hrm. It doesn't seem to have been installed:
david=# select uuid_generate_v1();
ERROR: function uuid_generate_v1() does not exist
LINE 1: select uuid_generate_v1();
^
HINT: No function matches the given name and argument types. You
might need to add explicit type casts.
david=# \q
trigger% ll /usr/local/pgsql/share/contrib/*.sql | grep uuid
trigger% ll /usr/local/pgsql/lib | grep uuid
> I don't know if there's any nice way to make configure avoid
> including <unistd.h> while testing uuid.h. Personally I think
> libuuid's attempt to usurp typenames that may be defined by the
> system headers is hopelessly broken, and that they'd be better off
> doing the #define's the other way around, ie, make the underlying
> real names of *their* typedefs different. It is just not acceptable
> that their header doesn't work if <unistd.h> is included first.
Thanks. See:
http://cvs.ossp.org/tktview?tn=164
Best,
David