lupyuen commented on PR #17197:
URL: https://github.com/apache/nuttx/pull/17197#issuecomment-3442684876

   > However, it's a good idea, so it might be worth making this change, but 
first we also need to think about how to bypass the check step.
   
   @simbit18 Yeah I'm a bit worried about doing any "emergency bypass", like 
when the CI fails during NuttX Release and we scramble to fix the CI. Also we 
have a Critical Expertise problem in NuttX CI, only you and I understand how 
the CI works :-(
   
   > The goal here is to reduce how often those parallel jobs run unnecessarily 
during normal PR development (for example, on drafts or failed checks), not to 
remove or serialize them completely.
   
   @trns1997 Actually running "unnecessary" jobs might not be a bad thing. 
Compare the Current CI vs Proposed CI:
   
   - __Current CI:__ Suppose all 3 steps fail: Check , Lint and Build. The 
developer sees all 3 failures immediately and fixes all 3 failures altogether 
in a Single PR Update. Which requires only 1 re-run of the CI.
   
   - __Proposed CI:__ Suppose all 3 steps fail: Check , Lint and Build. The 
developer only sees Check Step failure, fixes it, updates the PR, re-runs the 
CI. Then the Lint Step fails, we fix it, resubmit and re-run the CI. Finally 
the Build Step fails, we re-fix, re-submit, re-run the CI. Which means more CI 
Jobs, and probably more developer frustration ;-)
   
   - __I'm also thinking:__ The Current CI runs all Build Jobs in parallel: 
arm-01, arm-02, .... And we don't cancel the Entire CI Workflow immediately if 
any build fails, we always wait for the Build Job to complete. This will help 
our developer to fix all build errors in One Single PR Update. The Proposed CI 
seems to go against the current thinking?
   
   So sorry I find it a bit difficult to accept the Proposed CI changes. If we 
had more folks to maintain NuttX CI (especially during the critical period of 
NuttX Release), then we might be able to accept the changes. But right now, 
it's just too risky sorry :-(
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to