Open ltabis opened 1 year ago
The tests are they way they are. If they don't run out of the box, figure out whether that's an incompatible change or wrong tests.
I think that the warning isn't strictly correct because it doesn't generate a default script in the absence of a particular script, this isn't even a reason to warn is it? I can imagine it being useful to warn if zero scripts were found, rather than just some. (update: What I was trying to say is that scripts are not required to exist, it is valid to provide only some rather than all possible scripts)
When I wrote the systemd unit maintainer script fragment injection stuff at the time cargo-deb had no concept of debug logging, but I wanted and still think there would be value in adding "debug" level logging to show where the logic is looking for and failing to find files, to help with the case that it doesn't do as you expect.
Just wanted to encourage to make this a hard failure instead of a warning.
After https://github.com/kornelski/cargo-deb/pull/11 was merged and our tooling picked those changes up (which seems to have taken some time :)) we discovered that the maintainer scripts were not included anymore as we were still specifying paths relative to the workspace.
I think cargo-deb should not only emit a warning but an error and stop processing when this happens (of course explaining nicely what the problem is :)). I'm not sure we would have noticed a warning among the tons of output from CI.
Regarding #69
When the
maintainer-scripts
field is set and a script is not found, cargo deb will emit a warning telling the user that it will supply the script itself.This generates the following logs for the vsmtp project: (which provides
postinst
,postrm
andprerm
)I failed to run the test suite on the following tests because there is a missing
test-resources/testroot/testchild/debian
directory.How can I get it ?