Re: [PATCH-for-5.1 7/8] .travis.yml: Run fetch-acceptance-assets before check-acceptance

[Philippe Mathieu-Daudé]

On 4/6/20 5:31 PM, Alex Bennée wrote:
> Philippe Mathieu-Daudé <address@hidden> writes:
>
> > Signed-off-by: Philippe Mathieu-Daudé <address@hidden>
> > ---
> >   .travis.yml | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/.travis.yml b/.travis.yml
> > index 2fd63eceaa..c6b32da447 100644
> > --- a/.travis.yml
> > +++ b/.travis.yml
> > @@ -317,7 +317,7 @@ jobs:
> >         dist: bionic
> >         env:
> >           - CONFIG="--enable-tools 
> > --target-list=aarch64-softmmu,alpha-softmmu,arm-softmmu,m68k-softmmu,microblaze-softmmu,mips-softmmu,mips64el-softmmu,nios2-softmmu,or1k-softmmu,ppc-softmmu,ppc64-softmmu,s390x-softmmu,sparc-softmmu,x86_64-softmmu,xtensa-softmmu"
> > -        - TEST_CMD="make check-acceptance"
> > +        - TEST_CMD="travis_retry make -j1 fetch-acceptance-assets
> > check-acceptance DEBUG=1"
>
> You could use TEST_BUILD_CMD for the fetching of acceptance tests - can
> that be done in parallel?

I'd say calling fetch-acceptance-assets parallelized is what seems to 
break this Travis job (which is why I enforced -j1), but I'll refer to 
Willian here.

> Also no point in -j1 here - it's implied.

You are right, I wanted to be sure this still work even if the globlal 
script running the tests is updated. But the correct fix is probably to 
use .NOTPARALLEL in the fetch-acceptance-assets rule in the previous patch.

> >           - CACHE_NAME="${TRAVIS_BRANCH}-linux-gcc-acceptance"
> >         after_script:
> >           - python3 -c 'import json; r = json.load(open("tests/results/latest/results.json")); [print(t["logfile"]) for 
> > t in r["tests"] if t["status"] not in ("PASS", "SKIP")]' | xargs cat