Re: pgsql: Fix EXEC BACKEND + Windows builds for group privs

Greetings,

* Tom Lane (tgl(at)sss(dot)pgh(dot)pa(dot)us) wrote:
> Stephen Frost <sfrost(at)snowman(dot)net> writes:
> > Fix EXEC BACKEND + Windows builds for group privs
>
> jacana seems to think there's still an issue here:
>
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=jacana&dt=2018-04-09%2000%3A00%3A41
>
> It's pretty much the same symptom:
>
> Apr 08 20:58:22 Running this script will delete the old cluster's data files:
> Apr 08 20:58:22 delete_old_cluster.bat
> Apr 08 20:58:33 ++ find /home/pgrunner/bf/root/HEAD/pgsql.build/src/bin/pg_upgrade/tmp_check/data -type f '!' -perm 640
> Apr 08 20:58:33 ++ wc -l
> Apr 08 20:58:42 + '[' 3361 -ne 0 ']'
> Apr 08 20:58:42 + echo 'files in PGDATA with permission != 640'
> Apr 08 20:58:42 files in PGDATA with permission != 640
> Apr 08 20:58:42 + exit 1
> Apr 08 20:58:42 make: *** [check] Error 1

Yes, I was just looking at that and working up what I believe will be a
fix. This is exactly the kind of test I was anticipating getting run
unintentionally on Windows, sorry about that, but I don't have a Windows
test environment currently.

I've pushed what I believe will fix this, which is essentially "don't
try to run this test on this platform." Tested it locally (under Linux)
and it didn't blow up, so hopefully this doesn't make things worse, in
any case.

If there's more that's needed to fix jacana, I'll see about dealing with
it first thing in the morning.

Thanks!

Stephen