Support visualization of scopes, borrows and moves #387
Conversation
| rls-data = "0.7" | ||
| rls-span = { version = "0.4" , features = ["serialize-serde"] } | ||
| rls-vfs = { version = "0.4", features = ["racer-impls"] } | ||
| rls-analysis = { path = "../rls-analysis" } |
There was a problem hiding this comment.
Pending dependency updates mentioned in top-level comment.
| @@ -435,6 +435,35 @@ impl ActionHandler { | |||
| } | |||
| } | |||
|
|
|||
| pub fn borrow_info<O: Output>(&self, id: usize, params: TextDocumentPositionParams, out: O) { | |||
There was a problem hiding this comment.
Do you prefer another name?
| out.success(id, ResponseData::BorrowInfo(r)); | ||
| }, | ||
| Ok(Err(e)) => { | ||
| out.failure(id, &format!("BorrowInfo not available for that location: {:?}", e)[..]); |
There was a problem hiding this comment.
I'm betting this is not the way to handle a "nothing to see here" type of message. How should I fix this?
There was a problem hiding this comment.
We should probably be able to return something morally equivalent to out.success(id, None) and also use debug! to output a message for devs
| out.failure(id, &format!("BorrowInfo not available for that location: {:?}", e)[..]); | ||
| } | ||
| Err(_) => { | ||
| out.failure(id, "BorrowInfo failed to complete successfully"); |
There was a problem hiding this comment.
Other code wasn't logging errors. Should we start to do that? Is there a file we can write to in case writing to stderr interferes with the IDEs?
There was a problem hiding this comment.
Just using debug! is fine. The client can hide, display, or redirect to a file. We should be logging all errors with debug!.
There was a problem hiding this comment.
Should I output the object in the Err too? I noticed some other were following this pattern of throwing away the actual error, e.g. hover.
There was a problem hiding this comment.
I've been trying to output the error in the debug! log, but not in the failure message we send back to the LSP
src/cmd.rs
Outdated
| @@ -258,4 +277,8 @@ fn help() { | |||
| println!(" hover file_name line_number column_number"); | |||
| println!(" textDocument/hover"); | |||
| println!(" used for 'hover'"); | |||
| println!(""); | |||
| println!(" borrows file_name line_number column_number"); | |||
| println!(" textDocument/borrowInfo"); | |||
There was a problem hiding this comment.
I just noticed I forgot to update this to rustDocument.
| @@ -0,0 +1,77 @@ | |||
| use lsp_data::Range; | |||
There was a problem hiding this comment.
Since the language server types crate seemed to only be for types that are part of the spec, this seemed to be the best place for these structs. Are there any concerns with this setup?
There was a problem hiding this comment.
I'm not clear what the difference between these types and the rls-data types is? Ideally we'd just use one set of types. If that is not possible, then this setup seems fine.
There was a problem hiding this comment.
If I remember correctly it's the serialization derives. One is RustcSerialize and the other is Serializable. I'll double check after work.
There was a problem hiding this comment.
We ought to be able to derive both if we need to (I think we have feature gates for that, or maybe that is span and we could add them for data)
|
@nrc should we close this one too until the issues in the other crates are resolved? If we do close it, could I get responses to some of the questions above so I can get them fixed before the next PR? |
|
Answers inline. I don't mind leaving open or closing, whichever is easier for you. |
|
So given where some of the conversations have gone in rust#42733 I think we should close this. I took a look at the non-lexical lifetimes (NLL) RFC, and the way it manages lifetimes is very different. The front end code I have shouldn't change much since I'm already accounting for all sorts of edge cases there, but I think the compiler code and rls code will require changes to most of the design. Given that, it doesn't make sense to try to push this through, especially since it might complicate the implementation of NLL. The good news is @nikomatsakis asked to mentor me in the implementation of NLL! I'll do my best to take the visualizer into consideration when working on that project. |
This depends on rls-data#6, rust#42733, and rls-analysis#71.