Re: pg_regress: stat correct paths

Tom Lane wrote:
> Ah.  The reason this works in VPATH mode is that setup of the build tree
> duplicated all subdirectories of the source tree, so ./input/ should
> exist iff $srcdir/input/ does.  I agree it's a bit ugly though; it'd
> be better to stat what we really plan to read.
>
> Maybe push the stat operation inside convert_sourcefiles_in ?

Yes, that was what I tried to do in the patch, but unfortunately I was
too eager :) I'll send a new proposal tomorrow. Thanks for your help!

-J
--
Jørgen Austvik, Software Engineering
Database Technology Group