Quoting Helmut Grohne (2023-02-24 07:19:41) > As such, I propose a generic DEB_BUILD_OPTIONS=nowerror modelled after the > original observation, but meant to also match other checkers such as > shellcheck. The general idea should be that a warning should that can be > non-fatal should be non-fatal if possible.
Are there packages that run shellcheck in other places than those disabled by DEB_BUILD_OPTIONS=nocheck? If yes, should running shellcheck not be moved to there? In contrast to -Werror, shellcheck is not involved in literally building something. Otherwise, other linter tools like black come to mind that also recently broke my packages [1] when black got upgraded to 23.1.0-1 two weeks ago. But black (like shellcheck I'd assume) is usually disabled with nocheck or by not running autopkgtests. In what places is shellcheck called such that it cannot or should not be moved to a place where it could be disabled by DEB_BUILD_OPTIONS=nocheck? Thanks! cheers, josch [1] Versions of mmdebstrap and diffoscope that work with the new black in their autopkgtest have already been uploaded to unstable and are waiting to transition to testing so that black can transition as well. Maybe the timing of this upload of black was a bit unfortunate this late into the freeze because other packages are using it as part of their build process and now FTBFS. Like another package of mine (botch) which is also fixed now and waiting: https://bugs.debian.org/1031469
