Closed lzap closed 2 hours ago
I did split the OTK hunks into a separate PR: https://github.com/osbuild/images/pull/1051
Would it make sense to add a regression test here? Also in the "test-as-documentation" sense this might be useful(?)
It would make sense to add test for sure... I'd prefer a follow-up tho, I wouldn't like to block this, unless someone disagrees.
Would it make sense to add a regression test here?
I can do a followup, I actually tried by running a tiny HTTP proxy via library but it was not great test to be honest.
Would it make sense to add a regression test here?
I can do a followup, I actually tried by running a tiny HTTP proxy via library but it was not great test to be honest.
Cool, we don't need to block on this, I had a quick look and opened https://github.com/osbuild/images/pull/1052 - it's not ideal as it's not a full integration test but OTOH I think the extraction of the helper is beneficial and it does check that the proxy is correctly set (but it does know a bit too much about the implementation details for my taste).
Finally found the ostree resolving problem - it was in proxy parsing. The configuration value is not in URL form, it is in
hostname:port
form. Andurl.Parse
silently parses such URI setting a nonsense scheme tohostname
. TIL thatURL
in Go actually is technically URI: https://pkg.go.dev/net/url