Open tesuji opened 1 week ago
Maybe we should deny this lint on CI.
This will happen naturally. We have a CI job that deny
s all warnings except deprecations. It uses a pinned version of Rust which gets updating through an automated PR process. When the lint hits stable, we'll start blocking on it automatically.
I went ahead and merged some items separately so this PR can better focus on the messy case of completions
I think that there is a related issue https://github.com/clap-rs/clap/issues/5540.
Setting TERM=dumb
in completest-pty
doesn't to work properly. I guess we could try TERM=xterm
or something more modern.
thread 'fish::complete' panicked at clap_complete/tests/testsuite/fish.rs:159:5:
--- Expected
++++ actual: In-memory
1 - % exhaustive
2 - action complete (Register shell completions for this program) hint pacman value
3 - alias help (Print this message or the help of the given subcommand(s)) last quote ∅
1 + % exhaustive ∅
thread 'zsh::complete' panicked at clap_complete/tests/testsuite/zsh.rs:163:5:
---- expected: tests/testsuite/zsh.rs:156:20
++++ actual: In-memory
1 - % exhaustive
1 + % exhaustive
2 2 | complete -- Register shell completions for this program
3 3 | help -- Print this message or the help of the given subcommand(s)
4 - pacman action alias value quote hint last -- ∅
4 + pacman action alias value quote hint last --
5 + % exhaustive ∅
The test passed with TERM=xterm
. I guess we forced to only support running completest on xterm compatible terminals.
They was not run on CI, so they're siliencely falling for a long time.
Close #5540.