- 19 Mar, 2019 40 commits
-
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Modify the Yaml to generate much more images (using the newly created Docker images for third party corrected from the Openblas issue).
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Change in third party path; I am aware Ubuntu build will fail but I want here to test Docker one.
-
GILLES Sebastien authored
#1360 Test new Docker image with one different setting for Petsc (namely don't use /proc informations).
-
GILLES Sebastien authored
#1360 Deactivate momentarily a job which could only be taken by a Docker runner (due to lack of shell runner). To reactivate it, I would need to change in images the location of installation from /opt to /opt/MoReFEM.
-
GILLES Sebastien authored
#1360 Yaml: fix for shell runner (will break docker ones at the moment - images will be modified to fix that later on... at least if I manage to make them suitable!
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Attempt with shell runner: I'm not sure I will be able to figure out why the docker images runs fail due to Petsc whereas it is completely ok in a full-fledged environment. The only idea left (which will require to be at my office so in March...) is to see if creating the Docker image on Linux system is more robust).
-
GILLES Sebastien authored
#1360 Yaml: I try a different Docker image: the images generated on my MacBook repeatedly fail the MoReFEM tests due to a Petsc issue on virtually all other machines, including MacPro Nostromo. So I try with one generated on Nostromo: the point is seeing if the QLF runner is happy with it or not.
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Yaml: merge build and test stage (not happy with this, but artifact are too limited in size and cache isn't made to propagate information from a stage to another).
-
GILLES Sebastien authored
#1360 Yaml: experiment with artifact/dependencies (apparently cache doesn't support yet multiple runners).
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Small change in key to see if the env variable is correctly interpreted; should not solve the test issue.
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Gitlab CI test: my own image failed; I therefore try a lighter one in which I do not command entrypoint at all.
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Gitlab - CI: try a very simple test (akin to a hell world) to see if I am correctly doing the Docker runner. The error I get currently is related to zsh, so maybe I can skip using this shell in my Docker images, but I will check step by step what is working or not.
-
GILLES Sebastien authored
#1360 Gitlab CI: try with Ubuntu image rather than Fedora: I have with the latter an error message about an issue with sh, and behaviour for entry point was not the same for Ubuntu.
-
GILLES Sebastien authored
-
GILLES Sebastien authored
#1360 Gitlab-CI: add a new test build which would build in a Docker image starting from a Docker image put in the gitlab registry.
-
GILLES Sebastien authored
-