On Thu, Sep 14, 2017 at 02:48:06PM +0100, Daniel P. Berrange wrote:
On Thu, Sep 14, 2017 at 03:09:04PM +0200, Andrea Bolognani wrote:
> The build will fail if any of the commands fail, but this way we
> might catch more errors in a single run.
>
> Signed-off-by: Andrea Bolognani <abologna(a)redhat.com>
> ---
> .travis.yml | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/.travis.yml b/.travis.yml
> index ba8ff49a1..c2526bc6d 100644
> --- a/.travis.yml
> +++ b/.travis.yml
> @@ -76,7 +76,9 @@ before_install:
> before_script:
> - ./autogen.sh
> script:
> - - make -j3 && make -j3 syntax-check && make -j3 check
> + - make -j3
> + - make -j3 syntax-check
> + - make -j3 check
>
> # Environments here are run in addition to the main environment defined above
> matrix:
The downside of this is that if syntax-check fails, but check succeeeds,
you now have to search through the middle of the logfile to find the
failure, instead of just jumping straight to the end.
So I'm on the fence about this patch - I'd have a slight preference for
existing behaviour of failing fast to keep errors at the end of the log
I, personally, would go with:
script:
- make -j3 check
- make -j3 syntax-check
The former executes the whole build and fails on first compilation or
test error. The second one has a format that's easily skippable
(visually) and that's not that much to scroll through.
I'm leaving the choice up to you, though.