Re: [ANNOUNCE] Git v2.24.0-rc1 - Test Summary for NonStop

From: SZEDER Gábor <szeder.dev@gmail.com>
Date: Oct 28, 2019, 11:48 UTC
Message-ID: <20191028114850.GR4348@szeder.dev>
In-Reply-To: <021601d58c32$4f632a10$ee297e30$@rogers.com>

On Sat, Oct 26, 2019 at 03:19:37PM -0400, randall.s.becker@rogers.com wrote:

Show 11 quoted lines

> This is the test summary for the HPE NonStop TNS/E platform for git
> 2.24.0-rc1. All tests passed except as follows:
> 
> [NEW] 1. t0500-progress-display - failed 7 out of 11. This appears to
> consistently fail when run under Jenkins but consistently succeeds when run
> from a non-disconnected real terminal. The difference from a pipe handling
> standpoint with Jenkins is that it provides a stdout opened using O_WRONLY
> rather than O_RDWR or other open() modes that might exist - we have seen
> problems resulting from differences between process initiation under Jenkins
> and connected terminals in other products. This failure was not present in
> 2.23.0.

> It would be nice to understand why t0500 is failing before we release the
> official build to our community.

Yeah, that would be nice, but I don't know how we could help, if you don't even tell us which 7 tests failed... The output of the failed tests could also be quite useful.

It may or may not be conincidental, but there happens to be 7 tests in t0500 that use 'test-tool progress --total=<N>', which was affected by an endianness bug that got fixed in 2b6f6ea1bd (test-progress: fix test failures on big-endian systems, 2019-10-20). It would be interesting to see whether t0500 failed before that fix as well.