Make #[ts(export_to = "...")] relative to TS_RS_EXPORT_DIR
#250
Conversation
|
This is so annoying, docs.rs passes on my computer, but not on ci |
|
That's weird, changing the order of the tests changed which one failed |
|
I don't quite understand what's going on, but it seems like in |
|
Running |
macros/src/lib.rs
Outdated
| let export = self | ||
| .export | ||
| .then(|| self.generate_export_test(&rust_ty, &generics)); | ||
|
|
||
| let export_dir = std::env::var("TS_RS_EXPORT_DIR").ok(); |
There was a problem hiding this comment.
Ahh, I think this here might be the issue!
This is evaluated when the macro executes, during compilation.
When later running this executable, the TS_RS_EXPORT_DIR is hard-coded in there.
That might also happen when just doing cargo test, since rust might not re-compile.
There was a problem hiding this comment.
I think you're right, I just commited a version using std::option_env! instead and that seems to have fixed it
There was a problem hiding this comment.
Still, it might be better to move reading the env variable to export.rs instead, because I think the way I just did it forces recompilation, which kinda sucks
There was a problem hiding this comment.
Agreed, both with the observation as well as the conclusion.
There was a problem hiding this comment.
The last change did cause a recompilation, but recompilation of ts-rs-macros.
The 3 possible behaviours we could get are
- Re-compile ts-rs-macros (which also re-compiles the consumer crate)
Useenv!ints-rs-macros - Re-compile the consumer crate
Load the environment variable inexport.rs, usingenv!() - Don't recompile anything
Load the environment variable inexport.rs, usingenv::var()
There was a problem hiding this comment.
Got it fixed! Sorry about the billion failed CI notifications 😅
There was a problem hiding this comment.
- Don't recompile anything
Load the environment variable inexport.rs, usingenv::var()
Ended up on this one, my cpu is already thanking me 😆
|
I finished moving |
|
I could add a method that returns the proper path to remedy this |
| let base = std::env::var("TS_RS_EXPORT_DIR") | ||
| .ok() | ||
| .as_deref() | ||
| .map(Path::new) | ||
| .unwrap_or_else(|| Path::new(".")) | ||
| .to_owned(); | ||
| let export_to = | ||
| T::get_export_to().ok_or(ExportError::CannotBeExported(std::any::type_name::<T>()))?; | ||
| let path = Path::new(&export_to); | ||
| T::EXPORT_TO.ok_or(ExportError::CannotBeExported(std::any::type_name::<T>()))?; | ||
| let path = base.join(export_to); |
There was a problem hiding this comment.
Could we re-use output_path here? Might be the case that we have to get rid of the path::absolute in output_path for that.
There was a problem hiding this comment.
I don't even think we need to remove path::absolute, this path will be fed into diff_paths, which uses absolute paths as well
|
The reason I was thinking about adding it to |
ts-rs/src/lib.rs
Outdated
| let base = std::env::var("TS_RS_EXPORT_DIR") | ||
| .ok() | ||
| .as_deref() | ||
| .map(Path::new) | ||
| .unwrap_or_else(|| Path::new(".")) | ||
| .to_owned(); | ||
|
|
||
| let exported_to = base | ||
| .join(T::EXPORT_TO.map(ToOwned::to_owned)?) | ||
| .to_string_lossy() | ||
| .to_string(); |
There was a problem hiding this comment.
Same thing here, might be nice to re-use output_path here as well
|
Yeah, that's a good point. If you think it makes sense, go for it! |
|
I agree that the function that returns the path should never be overwritten. Maybe a Making |
* document `#[ts(tag)]` and `#[ts(as)]`, reformat documentation * fix line breaks in README.md * put cargo features in a table * Fix typo in readme * Document change to export_to made in #250 * Add missing newline * Explain flatten * Document #[ts(untagged)] on enum variants * Document #[ts(rename_all = ..)] on enum variants * Add missing newline * Change interface to type * Link to mentioned items * Run cargo readme * Document TS::DOCS * Improve TS::ident so it doesn't panic * document TypeList --------- Co-authored-by: Gustavo <gustavo.shigueo@proton.me>
As discussed in #245, when using both
TS_RS_EXPORT_DIRand#[ts(export_to = "...")], theexport_topath should be relative toTS_RS_EXPORT_DIRinstead of the project'sCargo.toml