Re: fixing pg_basebackup tests for modern Windows/msys2 - Mailing list pgsql-hackers

From Andrew Dunstan
Subject Re: fixing pg_basebackup tests for modern Windows/msys2
Date
Msg-id b2bb53ab-3f3d-5dba-b3b2-9b9ec0db8bd3@dunslane.net
Whole thread Raw
In response to fixing pg_basebackup tests for modern Windows/msys2  (Andrew Dunstan <andrew@dunslane.net>)
List pgsql-hackers
On 7/28/21 9:31 AM, Andrew Dunstan wrote:
> While looking into issues with fairywren and pg_basebackup tests, I
> created a similar environment but with more modern Windows / msys2.
> Before it even got to the test that failed on fairywren it failed the
> first TAP test for a variety of reasons, all connected to
> TestLib::perl2host.
>
> First, this function is in some cases returning paths for directories
> with trailing slashes and or embedded double slashes.  Both of these can
> cause problems, especially when written to a tablespace map file. Also,
> the cygpath invocation is returning a path with backslashes whereas "pwd
> -W' returns a path with forward slashes.
>
> So the first attached patch rectifies these problems. It fixes issues
> with doubles and trailing slashes and makes cygpath return a path with
> forward slashes just like the non-cygpath branch.
>
> However, there is another problem, which is that if called on a path
> that includes a symlink, on the test platform I set up it actually
> resolves that link rather than just following it. The end result is that
> the use of a shorter path via a symlink is effectively defeated. I
> haven't found any way to stop this behaviour.
>
> The second patch therefore adjusts the test to avoid calling perl2host
> on such a path. It just calls perl2host on the symlink's parent, and
> thereafter uses that result.
>
>


I've pushed these in master and REL_14_STABLE.


cheers


andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com




pgsql-hackers by date:

Previous
From: Dave Cramer
Date:
Subject: Re: pg_upgrade does not upgrade pg_stat_statements properly
Next
From: Jan Wieck
Date:
Subject: Re: pg_upgrade does not upgrade pg_stat_statements properly