Re: [Qemu-devel] [PATCH 05/11] travis: build tests objects in parallel, then run tests sequentially

[Peter Maydell]

On 19 June 2017 at 11:58, Alex Bennée <address@hidden> wrote:
>
> Philippe Mathieu-Daudé <address@hidden> writes:
>
>> tests are run sequentially to avoid mixed results output.
>
> I'm not so sure we are worried about mixed results output. I don't think
> anyone can compare the entire log and they tend to scroll to the end
> where the error will be.

Indeed, and make's --output-sync flag should be enough to keep
individual test output distinct.

What is perhaps a problem is that I'm not sure our tests all support
being run in parallel with each other without tripping each other up by
using the same temporary file / TCP port / etc at once. To the
extent that this doesn't work it's a bug, but we should start
by doing a solid soak test of the test suite to make sure -j<bignum>
works reliably.

thanks
-- PMM