"I think we need to revert the recent error-prone related work" -- Andrew, can you please give an example about this? Such as a JIRA that has this kind of failure in pre-commit build.
" Checkstyle's ImportOrder is one that always trips me up and no matter where I place the imports continues to complain." -- I had some struggles about this too, though, after I installed checkstyle plugin in intellij and used it before generating patches, it became less painful. I don't have any objection removing the check at the same time. Contributors should still try their best to organize imports cleanly and orderly. "" On Wed, Apr 3, 2019 at 12:02 PM Andrew Purtell <[email protected]> wrote: > I have been contributing to this project for more than ten years and have > noticed it is increasingly difficult to do so. > > For me the issues come down to precommit results. Precommit is a very > useful tool, but *only if committers are attentive to fixing breaking > changes immediately*. This has been an eternal problem. > > Also in fairness some problems I've thought are external to my patch have > turned out to be indirect consequences. Here the issue is I'm not able to > trust precommit so true positive results are still sometimes suspect. > > I think we need to revert the recent error-prone related work, this seems > to be the cause of some of the false failures in precommit jobs I've looked > at. > > In other cases we should adjust some static check settings. Checkstyle's > ImportOrder is one that always trips me up and no matter where I place the > imports continues to complain. I'm at a loss and it's really a trivial > matter. Let's just turn it off. > > The transient issues we sometimes face with Apache build infra are possibly > tolerable, I'm not referring to those. > > -- > Best regards, > Andrew > > Words like orphans lost among the crosstalk, meaning torn from truth's > decrepit hands > - A23, Crosstalk >
