Re: Where is SSPI auth username determined for TAP tests?

On Wed, Jul 03, 2019 at 09:53:14AM -0400, Tom Lane wrote:
> I haven't checked that this actually works, but it looks plausible,
> and I agree it's simpler/easier.

Thanks, committed. While testing on Windows, I have been trapped by
the fact that IPC::Run mishandles double quotes, causing the tests to
fail for the environment variable part because of a mismatching
pg_hba.conf entry. The difference is that with we run pg_regress
--config-auth using IPC::Run::run on HEAD but the patch switches to
system(). So I have finished by removing the double-quote handling
from the restore user name which makes the whole test suite more
consistent. The patch has at the end the advantage of removing in
pg_ident.conf the entry related to the OS user running the scripts,
which makes the environment more restricted by default.
--
Michael