acassis commented on issue #17914: URL: https://github.com/apache/nuttx/issues/17914#issuecomment-3755676383
> Github labels were introduced primarily so that CI logic could be later based on them. Some optimizations are already implemented thanks to Lup. Working with CI is difficult, implementing complex test logic is not easy, and any CI tests are slow. > > I proposed this PR [#17718](https://github.com/apache/nuttx/pull/17718) so that users could test changes on their forks, not wasting Apache CI resources for incomplete changes, but unfortunately it doesn't work as it should. > > I've said it many times before, but I'll say it again: we have too many configurations being built in CI. Most of them could be optimized into one large configuration that CI builds to ensure coverage. More and more users are adding new configurations to demonstrate simple features - all of this must be built via CI which is a waste of CI resources. It doesn't make any sense, especially for boards with a lot of FLASH memory @raiden00pl but I think those board profiles are useful for new users to get started to NuttX, unfortunately it is not easy for a new user to just copy a board profile from other board and in many cases even an "equal" board profile from other board X will fail in the board Y. See the recent issues with ESP32 WiFi that works fine in others ESP32 chips. But I agree with you that a board with more memory available could have a board profile called "full", "jumbo", etc and the CI should test only it. We don't need to test all board profiles, maybe we should just test "nsh" and some "full" board profile for each board. -- 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]
