feat!: Use env_logger for output #800
Conversation
|
Needs #799 for the CI, and I have to take a look where the nightly test is failing exactly |
| where | ||
| F: FnOnce() -> String, | ||
| { | ||
| if VERBOSE { |
There was a problem hiding this comment.
This was actually using generics to create two versions, one with the printlns disabled. that did save a bunch of time when I tested it.
There was a problem hiding this comment.
It really wouldn't surprise me if I accidentally removed a bit too much here.
| pub(crate) fn maybe_set_limit(limit: usize, verbose: u8) { | ||
| pub(crate) fn maybe_set_limit(limit: usize) { | ||
| JOBSERVER.get_or_init(|| { | ||
| if verbose > 1 { |
There was a problem hiding this comment.
this was using >1, so maybe trace?
|
A number of |
Well, maybe keep changing output method and changing output content separately. I did make laze output |
That's what I assumed the reason was for the prefix. I do think it should be applied consistently, otherwise messages that do not have the prefix might confuse the reader of the origin.
An opt out of the prefix? Should be possible to add in a follow up.
I already removed a 'warn' in the output content here and there, because that's duplicate now. |
|
Output should now be identical to current main. |
|
Bencher Report| Branch | pr/logger |
| Testbed | github-actions |
Click to view all benchmark results
| Benchmark | perf:task-clock | Benchmark Result msec x 1e3 (Result Δ%) | Upper Boundary msec x 1e3 (Limit %) |
|---|---|---|---|
| laze -C RIOT build --global --generate-only | 📈 view plot 🚷 view threshold | 9.34 x 1e3(-1.07%)Baseline: 9.44 x 1e3 | 9.94 x 1e3 (94.01%) |
3 participants
This refactors all (non-commented)
println!s to use a logger viaenv_logger. This allows for setting the log level via environment variables and retains the filtering of the messages via the command line arguments (--verboseand--quiet).A lot of the code is simplified because the
verboseargument doesn't have to be passed around in most cases.