Re: pg_ctl/pg_rewind tests vs. slow AIX buildfarm members

Michael Paquier wrote:

> > That would be WaitForSingleObject(handle, ms_timeout) ==
> > WAIT_OBJECT_0, no? The handle should be picked up from
> > CreateRestrictedProcess, and I think that CloseHandle should not be
> > called on pi.hProcess if we are going to wait for it afterwards.
> > Reference:
> > https://msdn.microsoft.com/en-us/library/windows/desktop/ms687032%28v=vs.85%29.aspx
> 
> I had a look at that, and attached is an updated patch showing the
> concept of using an HANDLE that pg_ctl waits for.

I wonder if you are interested in rewriting this whole thing to not use
cmd.exe at all, which as I understand is just about output redirection.
Earlier in the thread I pointed out that on Microsoft docs they claim
that you can redirect stdout/stderr when creating a new process without
going through cmd.exe; see
https://www.postgresql.org/message-id/20150903202114.GG2912%40alvherre.pgsql

-- 
Álvaro Herrera                http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services