We know that, by default, gitlab ci runners uses set -o pipefail, as explained in coderwall.com this particular option sets the exit code of a pipeline to that of the rightmost command to exit with a non-zero status, or zero if all commands of the pipeline exit successfully.
We all use the "|| true" statement to prevent a gitlab ci job from failing on a real exit 1 (to allow post process failure), for example if my grep makes an exit 1 but I consider this as normal and therefore I don't want my job to fail i write:
job:
script:
- grep "a" myfile.txt || true
But when i use functions instead of commands or scripts, it did not work anymore :(
working example with script (giving me exit 0):
job:
script:
- echo "exit 1" > test
- chmox u+x test
- test || true
working example with command (giving me exit 0):
job:
script:
- exit 1 || true
non working example with function (giving me exit 1):
job:
script:
- function test { exit 1; }
- test || true
non working example with function (giving me exit 1):
job:
script:
- function test { exit 1; }
- (test || true)
I don't understand what the difference in processing an exit code is between a script, a command or a function.
Does anyone have a solution ?
finally it is not a problem of behavior of the functions but an error on the management of the return code of a function, indeed it is necessary to use return instead of exit in order to guarantee the same functioning as a command.
The solution was therefore very simple, in the case of a function i have to write:
job:
script:
- function test { return 1; }
- test || true
Now exit code is 0 :)