-
Notifications
You must be signed in to change notification settings - Fork 2.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
lighttpd: update to 1.4.73 #46978
lighttpd: update to 1.4.73 #46978
Conversation
next time just amend and force-push the previous PR instead of opening another one
this PR isn't very useful if it wasn't run-tested at all and wasn't built on void (or do you mean in xbps-src on openwrt?) |
@classabbyamp I am a lighttpd developer. I have tested this release on multiple platforms (Linux, FreeBSD, OpenBSD, NetBSD, MacOS) and multiple architectures (under openwrt), though not specifically on void linux. If this PR is not acceptable, then please close it. |
The CI tests built the package on multiple architectures under void linux. Is that not sufficient for build testing? Regarding run-testing: lighttpd source supports |
@classabbyamp the void linux CI does perform lighttpd
|
it is, but running it on void is the more important testing
yes, it runs check on all non-cross arches In general, we'd prefer the package get tested in a real void environment by someone who uses the package on void (in addition to the CI and |
If someone comes along and tests it in situ that's fine too, this PR doesn't need to be closed |
Sounds like you think it more important for joe-user to say "yep, lighttpd serves my static file index.html on void linux." than the more extensive lighttpd test suite, which performs both unit tests and integration tests (e.g. full web request to a fastcgi server, an scgi server, CGI scripts, and more). For a different package with minimal tests, I could see the desire for in-person manual testing of a package. For more mature packages with more extensive test suites, I find that argument to be weaker that a user manually running a few smoke tests is better than the mature test suite which runs on multiple distros and multiple platforms and multiple architectures. |
Tested here and it works
It's && not ||; test failure is fatal in CI and will block merging of a PR until the cause can be found and fixed, unless the test is found to be broken in the first place or needing fixtures that can't run in CI. Manual testing is still required because of the variety of ways different packages do tests, and the extra work that would be required to verify that each one of them has enough tests and the right kind of tests that would catch runtime problems. It is also important to check that things like the services Void ships for packages do in fact continue to work as packaged after the update. |
@0x5c of course. My responses in this PR are due to @classabbyamp saying
Do you agree with @classabbyamp? Do you not want encourage submissions to maintain packages?
@0x5c Can that be added to the CI? If a package includes a service file, there can be an automated test which starts and stops the service. If I spin up a void VM, what is the "manual testing" that you suggest before I say "I tested the changes in this PR: yes" in the PR? (You're trusting people not to lie here. I am at least asking what should be "manually" tested, since I can probably automate it.) |
lighttpd: update to 1.4.73
Testing the changes
I tested the changes in this PR: NO
(build-tested elsewhere: on openwrt)