nightly linux coverage job failing due to malformed arguments about ci

[mikaelarguedas]

I think that's related to #51. Mu guess is that -DSECURITY=ON doesnt get it's end delimiter and thus the next argument gets merged into it

[wjwwood]

I think that's related to #51. Mu guess is that -DSECURITY=ON doesnt get it's end delimiter and thus the next argument gets merged into it

I was just getting to the same conclusion, but do you know why that is? (i.e. is it something you know how to fix off-hand, or should I dig into it?)

[mikaelarguedas]

Not sure what the right fix is, I didn't have time to investigate. My guess is that the logic in https://github.com/ros2/ci/blob/master/job_templates/ci_job.xml.em#L152-L173 and it's windows equivalent is fragile, and that the check for the end delimiter should be done either in a different place (make sure we don't append anything else to that list) or in a smarter way (identifying greedy argument and end-delimit them). Sorry that's not a very helpful answer

[wjwwood]

Looks like it was fixed last night:

http://ci.ros2.org/view/nightly/job/nightly_linux_coverage/418/

I wouldn't have thought this was a flakey kind of bug. Did anyone deploy or redeploy setting maybe (i.e. was master ok, but the latest had not been pushed somehow)?

[mikaelarguedas]

hmmm I guess that's on me, looks like I deployed the refactor of the paramete to that job last night (I thought I did only the ci_packaging jobs 😕 ). So it's been fixed by passing the additional ament build args before. I guess this one can be closed, we will need to reevaluate #51, I think there is still a conceptual issue on the delimiter logic

[wjwwood]

Ok, I'll close this for now and reopen it if it is an issue in the coming nightlies. I'll leave #51 to track the higher level issue.