.gitlab-ci.yml: tag fast on fast runners; tag workingrunner; fix issue #64
This MR has evolved a bit. It improves .gitlab-ci.yml
in general:
- uses tag:
fast
on fast runners so that build will use the fastest servers that we have - Tag
workingrunner
was removed as not necessary:uses tag:workingrunner
which we use for marking whether a runner is working or is abrokenrunner
- It makes the tag:
lightweight
to be obsolete. Not used anymore. - it fixes #64 (closed). After adding
-DENABLE_POTENTIAL_BLOCKS=ON -DENABLE_POTENTIAL_PARTICLES=ON -DENABLE_MPI=ON
to.gitlab-ci.yml
we have documentation for openfoam, PotentialBlock, PotentialParticle- it changes docker-yade: docker-yade!4 (merged) so please merge that one as well.
- adds a note in index.html about Gdańsk :)
Previous description of this MR, a bit outdated.: Revert beb219ce : lightweight everywhere → fast only on fast runners
How about requiring that only the build stage has to use a fast
gitlab-runner? I added this tag to 4pak and 6pak runners [1] but I cannot edit the tags for the group runners (right column in [1]). Maybe I am overdoing this, so feel free to discard this MR. It's just to remove the lightweight
tag from all runners, and add a fast
tag to fast runners.
By this way if yade-runner-01 someday will start working it will never try to compile yade. It will just do some tests and documentation.
After some testing in cosurgi/trunk!4 (closed) I discovered that using -DCHUNKSIZE=4
can speed up compilation (on 4pak and 6pak) by 3 minutes, down to 9 minutes. I would like to test this -DCHUNKSIZE=4
on yade-runner, but to get this test to work you would have to add fast
tag to yade-runner. (Or we would have to remove the tag fast
from .gitlab-ci.yml, which misses the point of this MR ;)
OTOH if it turns out that yade-runner cannot handle that, we could experiment with case
and use a different -DCHUNKSIZE
for it. It still could make it faster by few minutes.
[1] https://gitlab.com/yade-dev/trunk/settings/ci_cd#js-runners-settings