On 9/16/19 11:46 AM, Alex Bennée wrote: > > Philippe Mathieu-Daudé <phi...@redhat.com> writes: > >> On 9/16/19 10:43 AM, Alex Bennée wrote: >>> Philippe Mathieu-Daudé <f4...@amsat.org> writes: >>> >>>> Signed-off-by: Philippe Mathieu-Daudé <f4...@amsat.org> >>>> --- >>>> .travis.yml | 18 ++++++++++++++++-- >>>> 1 file changed, 16 insertions(+), 2 deletions(-) >>>> >>>> diff --git a/.travis.yml b/.travis.yml >>>> index 69a37f7387..753276eb33 100644 >>>> --- a/.travis.yml >>>> +++ b/.travis.yml >>>> @@ -265,9 +265,23 @@ matrix: >>>> - "3.6" >>>> >>>> >>>> - # Acceptance (Functional) tests >>>> + # Acceptance (Functional) tests [enterprise] >>>> - env: >>>> - - CONFIG="--python=/usr/bin/python3 >>>> --target-list=x86_64-softmmu,mips-softmmu,mips64el-softmmu,aarch64-softmmu,arm-softmmu,s390x-softmmu,alpha-softmmu,ppc-softmmu,ppc64-softmmu,m68k-softmmu" >>>> + - CONFIG="--python=/usr/bin/python3 >>>> --target-list=x86_64-softmmu,mips64el-softmmu,aarch64-softmmu,s390x-softmmu,ppc64-softmmu" >>> >>> You could use: >>> >>> --target-list=${MAIN_SOFTMMU_TARGETS} >>> >>> and >>> >>> --disable-user --target-list-exclude=${MAIN_SOFTMMU_TARGETS} >> >> I like the idea, but this variable is slighly different: >> >> MAIN_SOFTMMU_TARGETS="aarch64-softmmu,arm-softmmu,i386-softmmu,mips-softmmu,mips64-softmmu,ppc64-softmmu,riscv64-softmmu,s390x-softmmu,x86_64-softmmu" >> >> Are you suggesting we use MAIN_SOFTMMU_TARGETS for 'Enterprise' >> targets? > > Broadly - I'm note sure how we want to make the distinction in QEMU. We > have targets which are actively maintained and represent current > architectures that people still ship software to run on. The other > targets are machines that are either esoteric designs that exist in > particular niches or of academic interest and the old architecture kept > alive out of a sense of nostalgia. > >> >> As long as mips64-softmmu is present, I think we can remove mips-softmmu >> from MAIN_TARGETS. Maybe i386-softmmu can be removed too. > > Well the MIPS targets are actively maintained although for 32 bit you > don't see many distros for them these days. There might be an argument > for demoting i386-softmmu as I'm guessing there isn't much 32 bit kvm in > the enterprise right now (although It would be unsurprising to be told > otherwise). > >> >> That would give the following hobbyist list: >> >> i386-softmmu,mips-softmmu,alpha-softmmu,ppc-softmmu,m68k-softmmu > > I'm not sure about the hobbyist moniker as pretty much any of these TCG > targets is more actively maintained than the x86 which is mostly all > about the HW virtualisation these days. Trouble is I can't think of > another name that doesn't sound condescending: > > HISTORICAL_SOFTMMU_TARGETS > LEGACY_SOFTMMU_TARGETS > FASTERTHANNATIVE_SOFTMMU_TARGETS > ARCHIVISTS_SOFTMMU_TARGETS > > ? As ever naming things is hard....
:) I'm not trying to classify tests, I simply want users/maintainers to not be reluctant/scared to add tests because "CI takes too long". Also, I find it practical to have small jobs, what I do when testing a series when I know it only affects a set of jobs, I push and directly click "cancel all jobs" on the Travis web, then I just click on "restart this job" on the set I'm interested in. This way I can quickly figure out the series is OK or rework it. >> (I plan to add hppa-softmmu there too). >> >>>> + - TEST_CMD="make check-acceptance" >>>> + after_failure: >>>> + - cat tests/results/latest/job.log >>>> + addons: >>>> + apt: >>>> + packages: >>>> + - python3-pil >>>> + - python3-pip >>>> + - python3.5-venv >>>> + >>>> + >>>> + # Acceptance (Functional) tests [hobbyist] >>>> + - env: >>>> + - CONFIG="--python=/usr/bin/python3 >>>> --target-list=mips-softmmu,arm-softmmu,alpha-softmmu,ppc-softmmu,m68k-softmmu" >>>> - TEST_CMD="make check-acceptance" >>>> after_failure: >>>> - cat tests/results/latest/job.log >>> >>> >>> -- >>> Alex Bennée >>> > > > -- > Alex Bennée >