On Wed, 30 Aug 2000, Tom Lane wrote:
> ghaverla@freenet.edmonton.ab.ca writes:
> > I ran across some interesting (at least to me) initial results
> > in regression testing 7.0.2 on my Ultra 2 running Solaris 2.5.1.
>
> > The makefile in regress/sql deletes the create_function_1.sql,
> > create_function_2.sql, copy.sql, contraints.sql and misc.sql
> > files when you run make all. Then when you run make runtest,
> > those tests fail (cause error messages) are the executable
> > scripts are no longer there.
>
> Huh? To prepare for a regression test you should do "make clean all" in
> src/test/regress, not in its subdirectories. The files you mention are
> supposed to get created from template files that are in regress/input.
> If you're doing the make per instructions and that's not happening, then
> there's something squirrely about your copy of make...
I was running make in src/test/regress. I was running
make clean, and then make all. I was trying to figure out
why those files were never there when the tests ran, so
I cd'd to sql and did make all. I observed that it was deleting
the files. After I commented out the rm, things seemed to work.
I know it was a hack, but I couldn't see why these files were
getting deleted. It is a fairly recent GNU make, but I got it
off a download site.
Gord
Matter Realisations http://www.materialisations.com/
Gordon Haverland, B.Sc. M.Eng. President
101 9504 182 St. NW Edmonton, AB, CA T5T 3A7
780/481-8019 ghaverla @ freenet.edmonton.ab.ca