Re: cube and earthdistance diffs

On Mon, Sep 09, 2002 at 22:16:06 -0400,
  Bruce Momjian <pgman@candle.pha.pa.us> wrote:
> Bruno Wolff III wrote:
> > Reading a message in another thread makes it seem likely that I am going
> > to want to change things to add the grants into the creation scripts,
> > so I will probably need to change thinsg anyway and might as well fix
> > the cube install issues myself.
>
> Actually, I though that to make earthdistance, you were going to specify
> cube/* as a dependency, and have that compile before earthdistance
> compiled.  Maybe that would work, but earthdistance can't install cube.

No big deal. I sent in a new patch that doesn't install cube, but instead
says to install it first in the documentation.

It isn't like the psql using libpq case, because it needs the functions to
be loadable into the database to run the regression test. The real dependency
was that to usefully install earthdistance you needed to have cube installed.
I didn't need to have it to the install on all, to it would have to be
installed to run the regression test.

I also got a regression test failure reported to me this evening on an alpha
and it turns out the was a bug in cube_enlarge that didn't cause a
regression failure on my system by luck. I submitted a patch for it tonight.
It doesn't touch anything changed by my other patch and can be applied
either before or after the other one. I also changed a couple of ints to
int4s to match the sql declarations used for the functions.