feat: function privacy leak validator

[mkhan45]

Part of #309

Adds a validation for leaking private types from function signatures

[mkhan45]

add .take(1) to show only a single diagnostic per function

[baszalmstra]

Thanks for this!

Would it be possible to highlight the region of the type thas has a more private type? So in the case of:

pub fn foo(ty: PrivateType) { 
               ~~~~~~~~~~~ <-- this part is highlighted: cant leak `PrivateType` 
 /// ...
}

[mkhan45]

Alright, I've changed the diagnostic to highlight the type, but it's a bit harder to make it actually display the type name in the diagnostic message. I don't think it's a big deal since the name is highlighted anyway though.

error: can't leak private type
 --> mod.mun:3:16
  |
3 | pub fn foo(ty: PrivateType) -> PrivateType {
  |                ^^^^^^^^^^^ can't leak private type
  |error: can't leak private type
 --> mod.mun:3:32
  |
3 | pub fn foo(ty: PrivateType) -> PrivateType {
  |                                ^^^^^^^^^^^ can't leak private type

[baszalmstra]

You can implement something in mun_diagnostic to also be able to give proper feedback on the name. In that crate you conversion happens from mun_hir::Diagnostics to mun_diagnostic::Diagnostic. A lot more context is known at that point (like the entire HIR database) which should make it fairly easy to get the name of the type. See https://github.com/mun-lang/mun/blob/master/crates/mun_diagnostics/src/hir/unresolved_type.rs as an example.

The goal of mun_diagnostic is also to provide fixes in the future, this would be a prime candidate.

Adding the name is not required perse but I think it will provide users with a much better experience down the road.

[baszalmstra]

Also works properly with the language server:

Nice!

[codecov]

Codecov Report

Merging #338 (f0a27e3) into master (1eb2a56) will increase coverage by 0.06%.
The diff coverage is 84.61%.

@@            Coverage Diff             @@
##           master     #338      +/-   ##
==========================================
+ Coverage   80.24%   80.30%   +0.06%     
==========================================
  Files         258      259       +1     
  Lines       15572    15637      +65     
==========================================
+ Hits        12496    12558      +62     
- Misses       3076     3079       +3     

Impacted Files	Coverage Δ
crates/mun_codegen/src/test.rs	100.00% <ø> (ø)
crates/mun_hir/src/ty/lower.rs	79.38% <0.00%> (-2.51%)	⬇️
crates/mun_runtime/tests/hot_reloading.rs	100.00% <ø> (ø)
crates/mun_runtime/tests/marshalling.rs	98.20% <ø> (ø)
crates/mun_runtime/tests/memory.rs	100.00% <ø> (ø)
crates/mun_runtime_capi/src/tests.rs	99.49% <ø> (ø)
crates/mun_hir/src/ty.rs	73.75% <75.00%> (+0.06%)	⬆️
crates/mun_hir/src/diagnostics.rs	64.87% <83.33%> (+0.46%)	⬆️
crates/mun_diagnostics/src/hir/exported_private.rs	92.30% <92.30%> (ø)
crates/mun_hir/src/expr/validator.rs	93.84% <92.30%> (-1.03%)	⬇️
... and 6 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1eb2a56...f0a27e3. Read the comment docs.

[mkhan45]

I've added a mun_diagnostic diagnostic so the type names show: https://github.com/mun-lang/mun/pull/338/files#diff-f6d9d08feaa1fdc33c5467abad598c17045b35ab9dc93d19b1bf0acf8e61fa6c

[baszalmstra]

Nice! 👍

[Wodann]

Looks good overall. I had one question

[Wodann]

We have duplication of this error message. Is there a way to reuse it to avoid code+paste errors?

Other occurance:
https://github.com/mun-lang/mun/pull/338/files#diff-7a57632f7acc9a0b1abad5ef6f9447f95fabecf03312d0ab7d0597807ae8e25dR24

[mkhan45]

I've updated it so that there's no duplication