This improves the compatibility and adds the ability to introduce
non-breaking changes by using a builder pattern.
Example:
```rust
fn main() {
match xplr::runner(None).and_then(|a| a.run()) {
Ok(Some(out)) => print!("{}", out),
Ok(None) => {}
Err(err) => {
if !err.to_string().is_empty() {
eprintln!("error: {}", err);
};
std::process::exit(1);
}
}
}
```
This works:
```lua
xplr.fn.custom.ping = function(app)
print("What's your name?")
local name = io.read()
os.execute('read -p "Hello ' .. name .. ', you are in ' .. app.pwd .. '"')
return {
{ LogSuccess = "pong" },
}
end
```
Then it can be called via `CallLua: custom.ping`.
The initial idea was to enable other tools to control `xplr` via the
input pipe. However, so far I didn't feel the need to use this feature.
And even if there is any need, it's much better to implement ad-hoc
services instead of wasting cpu resources.
- Optimize by avoiding cloning the whole app in each iteration of the main
loop.
- Increase the input poll timeout from 1 to 200. This works because the
poll will not apply to key hold.
- Do not read input pipe if it hasn't been modified.
```
Benchmarking focus next item: Collecting 100 samples in estimated 5.1972 s (126k itera focus next item time: [41.216 us 41.346 us 41.494 us]
change: [-28.669% -28.110% -27.551%] (p = 0.00 < 0.05)
Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
4 (4.00%) high mild
5 (5.00%) high severe
Benchmarking focus previous item: Collecting 100 samples in estimated 5.0576 s (116k i focus previous item time: [43.589 us 43.754 us 43.927 us]
change: [-29.506% -28.748% -28.039%] (p = 0.00 < 0.05)
Performance has improved.
Found 3 outliers among 100 measurements (3.00%)
1 (1.00%) high mild
2 (2.00%) high severe
Benchmarking focus first item: Collecting 100 samples in estimated 5.1765 s (116k iter focus first item time: [44.071 us 44.340 us 44.634 us]
change: [-26.739% -26.314% -25.885%] (p = 0.00 < 0.05)
Performance has improved.
Found 12 outliers among 100 measurements (12.00%)
8 (8.00%) high mild
4 (4.00%) high severe
Benchmarking focus last item: Collecting 100 samples in estimated 5.1522 s (116k itera focus last item time: [43.950 us 44.214 us 44.541 us]
change: [-27.571% -26.953% -26.337%] (p = 0.00 < 0.05)
Performance has improved.
Found 11 outliers among 100 measurements (11.00%)
5 (5.00%) high mild
6 (6.00%) high severe
Benchmarking leave and enter directory: Collecting 100 samples in estimated 5.4863 s ( leave and enter directory
time: [96.645 us 96.915 us 97.234 us]
change: [-28.720% -27.224% -25.666%] (p = 0.00 < 0.05)
Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
6 (6.00%) high mild
3 (3.00%) high severe
```
This PR targets 2 pain points.
1. The `Explore` message was async, which caused some unexpected
behavior. This was fixed by splitting `Explore` into `ExplorePwd`,
`ExplorePwdAsync` and `ExploreParentsAsync`. `ExploreParentsAsync`
is similar to the former `Explore`, which is mainly used when loading
`xplr` for the first time. However, what we'll be using frequently
are `ExplorePwd` and `ExplorePwdAsync` messages.
2. Files with spaces caused some unexpected behavior. This was fixed by
escaping the paths properly. This also fixed focusing of a file after
creating or renaming it.
Anothor breaking change is that `XPLR_PIPE_FOCUS_OUT` has been removed.
`XPLR_FOCUS_PATH` is all we need. So, the rule of thumb is if a variable
contains one liner value, it can be used directly from the env vars.
Variables that can contain multi-line values, will be exposed via the
pipes.
Minor changes are
- Add `switch_mode` mode to the global key binding help menu
- Moved some UI related code from config.rs to ui.rs.
- Fixed compilation issue on `rustc 1.50.0`.
From this commit, the app state will be written to the output pipes only
when invoking a command.
For auto refreshing pipes, we can brainstorm on `service`s concept.
Arijit Basu