Control: tags -1 - moreinfo On Wed, 2019-07-24 at 11:26 -0300, Chris Lamb wrote: > Parallel to my comment on #932862, what would you say to simply also > emitting testsuite-autopkgtest-missing in this case (and naturally > updating the description, etc). I would say that superficial tests do provide a small amount of test coverage (command-line parsing and options processing) so they do provide some value, so I think that this case is distinct enough from autopkgtests being entirely missing that it should be a separate tag. In addition I think this superficial-only-tests tag should be of a different severity to testsuite-autopkgtest-missing, perhaps raised to info or warning level instead of pedantic. > What I really mean to say is that I wonder whether we should zoom out > a bit and get a good picture about what we want in this area without > any duplicative code or effort. :) I agree that we should zoom out and get a more global view of the problems that could potentially arise in autopkgtests. This is my first foray into autopkgtest QA and so it is the first thing I ran into. I came into it through the uhubctl RC bug, which reports failure of a test that runs `uhubctl -v` via a script, rather than from the debian/tests/control file. I think unless lintian folks want to write some Haskell based on shellcheck or OCaml based on morbig (see Ralf's talks in recent DebConfs about shell script parsing), lintian will never be able to detect the uhubctl case as superficial. As a result I prepared an item for Misc Dev News asking people to tag their tests and to write some non-superficial tests: https://wiki.debian.org/DeveloperNews?action=diff&rev1=608&rev2=609 -- bye, pabs https://wiki.debian.org/PaulWise
Attachment:
signature.asc
Description: This is a digitally signed message part