The `LifetimeParam` and `Local` variants use `source()` to find their
range. Now that `source()` returns an `Option` we need to handle the
`None` case.
In #6901 some special case handling for proc-macros was introduced to
prevent panicing as they have no AST. Now the new HasSource::source
method is used that returns an option.
Generally this was a pretty trivial change, the only thing of much
interest is that `hir::MacroDef` now implements `TryToNav` not `ToNav`
as this allows us to handle `HasSource::source` now returning an option.
Added the error message to the doc for the UnresolvedProcMacro
diagnostic, explaining that either enabling the procMacro setting
or disabling this diagnostic should make the warnings go away.
7123: Add support for Rust 2021. r=lnicola a=m-ou-se
This adds `2021` in all places where rust-analyzer already knew about `2015` and `2018`.
The only edition-specific behaviour I could find in the source code was gated on a direct comparison with `Edition2015`, so `Edition2021` should (correctly) behave the same as `Edition2018`:
56a7bf7ede/crates/hir_def/src/nameres/path_resolution.rs (L132)
Co-authored-by: Mara Bos <m-ou.se@m-ou.se>
7116: Fix deep syntax tree bug generated by proc-macro r=jonas-schievink a=edwin0cheng
This PR fixed a bug from `semver-parser` and `pest_derive` crates which generate a very deep syntax tree such that serde reject to de-serialize. To fix this bug, we disabled recursion limit in `serde` (by calling [`Deserializer::disable_recursion_limit`](https://docs.rs/serde_json/1.0.61/serde_json/struct.Deserializer.html#method.disable_recursion_limit))
I have a feeling that we still need some way to protect against bad proc-macro generating huge syntax node, but I have no idea right now.
r? @jonas-schievink
Fixes#7103
Co-authored-by: Edwin Cheng <edwin0cheng@gmail.com>
7117: Use stable instead of beta on macos-11 release runner r=lnicola a=lnicola
Because beta was promoted to stable yesterday.
bors r+
Co-authored-by: Laurențiu Nicola <lnicola@dend.ro>
7102: Fix completion of Default struct update syntax r=Veykril a=nick96
Previously the inserted text was always `..Default::default()` which ends up as `...Default::default()`
if `.` was typed. Now checks if the current token is `.` and inserts `.Default::default()`
if it is, so `..Default::default()` is correctly completed.
I think there's probably a better way to implement this context aware completion because I've seen it in other parts of rust-analyzer as a user but I'm not sure how to do it.
Fixes#6969
Co-authored-by: Nick Spain <nicholas.spain@stileeducation.com>
7071: Pass --all-targets to "cargo check" when discovering external resources r=matklad a=WasabiFan
There is a repro case and background in the linked issue.
In short, the goal of this MR is to allow rust-analyzer to discover proc-macros which come from your tests (including, most importantly, dev-dependencies).
By default, `cargo check` implies the equivalent of `--lib --bins`, meaning it'll check your libraries and binaries -- but not tests! This means proc-macros (or, I guess, build scripts as well) weren't discovered by rust-analyzer if they came from tests.
One solution would be to manually add `--lib --bins --tests` (i.e., just augment the effective options to include tests). However, in this MR, I threw in `--all-targets`, which [according to the docs](https://doc.rust-lang.org/cargo/commands/cargo-check.html#target-selection) implies `--benches --examples` too. I have absolutely no idea what repercussions that will have on rust-analyzer for other projects, nor do I know if it's a problem that build scripts will now be discovered for tests/examples/benches. But I am not aware of a reason you _wouldn't_ want to discover these things in your examples too.
I think the main drawback of this change is that it will likely slow down the `cargo check`. At a minimum, it'll now be checking your tests _and_ their dependencies. The `cargo check` docs also say that including `--tests` as I have here may cause your lib crate to be built _twice_, once for the normal target and again for unit tests. My reading of that caveat suggests that "building twice" means it's built once for the tests _inside_ your lib, with a test profile, and again for any consumers of your lib, now using a normal release profile or similar. This doesn't seem surprising.
Very minor caveat: `--tests` will not include tests within a binary if it has `test = false` set in `Cargo.toml`. (I discovered this manually by trial-and-error, but hey, it actually says that in the docs!) This is likely not an issue, but _does_ mean that if you are -- for whatever reason -- disabling tests like that and then manually specifying `cargo test --package <...> --bin <...>` to run them, rust-analyzer will remain unaware of proc-macros in your tests.
I have confirmed this fixes the original issue in my sandbox example linked in #7034 and in my own project in which I originally discovered this. I've left it configured as my default RA language server and will report back if I notice any unexpected side-effects.
Fixes#7034
Co-authored-by: Kaelin Laundry <wasabifan@outlook.com>
7107: add working dir to cargo metadata fail messages r=matklad a=lf-
Context: I was having an error in my workspace config that I couldn't figure out without this added output. In particular, I have a Code workspace with a docs folder and one of my top level crates [which each have their own custom target], which was picking up the Cargo workspace level Cargo.toml incorrectly [which in my project should not ever be used for the editor], and ultimately had to override it with `linkedProjects`.
Here's a sample of the changed output:
```
[INFO rust_analyzer::main_loop] handle_event(Workspaces([Err(Failed to read Cargo metadata from Cargo.toml file /home/jade/dev/mu/Cargo.toml, cargo 1.50.0-nightly (75d5d8cff 2020-12-22)
Caused by:
0: Failed to run `cargo metadata --manifest-path /home/jade/dev/mu/Cargo.toml` in `/home/jade/dev/mu`
1: Error during execution of `cargo metadata`: error: target path "../../riscv64imac-mu-shoo-elf.json" is not a valid file
Caused by:
No such file or directory (os error 2)
), Err(Failed to read Cargo metadata from Cargo.toml file /home/jade/dev/mu/shoo/Cargo.toml, cargo 1.50.0-nightly (75d5d8cff 2020-12-22)
Caused by:
0: Failed to run `cargo metadata --manifest-path /home/jade/dev/mu/shoo/Cargo.toml` in `/home/jade/dev/mu/shoo`
1: Error during execution of `cargo metadata`: error: target path "../../riscv64imac-mu-shoo-elf.json" is not a valid file
Caused by:
No such file or directory (os error 2)
)]))
```
Co-authored-by: lf- <lf-@users.noreply.github.com>
7055: Install rustfmt if needed during formatting request r=matklad a=kinnison
In order to fix#6052 and to fix#4249 this PR installs `rustfmt` using `rustup` if `rustfmt --help` fails.
In theory we ought to memoise the attempts (both the `--help` and the installation) so that we don't keep retrying something which will fail (e.g. if nightly is missing rustfmt), but this is a first-pass for opinions.
In order to make it possible to notify the user of what happened, I added a facility for `GlobalStateSnapshot` to send *messages* to the client indicating what happened. There may be a cleaner way to do this as well but I wasn't sure exactly what might be best given this is my first time in this codebase.
It may be worth, longer term, working with `rustup` to provide a way to detect a missing component binary since `rustup` returns `1` if `rustfmt` is not installed, which is not ideal.
Co-authored-by: Daniel Silverstone <dsilvers@digital-scurf.org>
In an attempt to fix#6052 and #4249 this attempts to detect
if rustfmt is a rustup proxy which isn't installed, and reports
the error message to the user for them to fix.
In theory this ought to be memoised but for now it'll do as-is.
Future work might be to ask the user if they would like us to
trigger the installation (if possible).
Signed-off-by: Daniel Silverstone <dsilvers@digital-scurf.org>
Previously the inserted text was always `..Default::default()` which ends up as `...Default::default()`
if `.` was typed. Now checks if the current token is `.` and inserts `.Default::default()`
if it is, so `..Default::default()` is correctly completed.
Fixes#6969
