Stream the dep-graph to a file instead of storing it in-memory.
This is a reimplementation of #60035.
Instead of storing the dep-graph in-memory, the nodes are encoded as they come
into the a temporary file as they come. At the end of a successful the compilation,
this file is renamed to be the persistent dep-graph, to be decoded during the next
compilation session.
This two-files scheme avoids overwriting the dep-graph on unsuccessful or crashing compilations.
The structure of the file is modified to be the sequence of `(DepNode, Fingerprint, EdgesVec)`.
The deserialization is responsible for going to the more compressed representation.
The `node_count` and `edge_count` are stored in the last 16 bytes of the file,
in order to accurately reserve capacity for the vectors.
At the end of the compilation, the encoder is flushed and dropped.
The graph is not usable after this point: any creation of a node will ICE.
I had to retrofit the debugging options, which is not really pretty.
Fix `redundant_clone` fp
fixes: #5973fixes: #5595fixes: #6998
changelog: Fix `redundant_clone` fp where the cloned value is modified while the clone is in use.
panic early when `TrustedLen` indicates a `length > usize::MAX`
Changes `TrustedLen` specializations to immediately panic when `size_hint().1 == None`.
As far as I can tell this is ~not a change~ a minimal change in observable behavior for anything except ZSTs because the fallback path would go through `extend_desugared()` which tries to `reserve(lower_bound)` which already is `usize::MAX` and that would also lead to a panic. Before it might have popped somewhere between zero and a few elements from the iterator before panicking while it now panics immediately.
Overall this should reduce codegen by eliminating the fallback paths.
While looking into the `with_capacity()` behavior I also noticed that its documentation didn't have a *Panics* section, so I added that.
Update cargo
5 commits in 1e8703890f285befb5e32627ad4e0a0454dde1fb..3c44c3c4b7900b8b13c85ead25ccaa8abb7d8989
2021-03-26 16:59:39 +0000 to 2021-03-31 21:21:15 +0000
- Fix semver docs for 1.51. (rust-lang/cargo#9316)
- Add `cargo config` subcommand. (rust-lang/cargo#9302)
- Give one more example for the --featuers CLI (rust-lang/cargo#9313)
- Bump to 0.54.0, update changelog (rust-lang/cargo#9308)
- Make the URL to the tracking issue for `--out-dir` into a link (rust-lang/cargo#9309)
Rollup of 5 pull requests
Successful merges:
- #83015 (Add regression tests for #79825 and #81555)
- #83699 (Add a regression test for issue-68830)
- #83700 (Fix documentation of conversion from String to OsString)
- #83711 (Clarify `--print target-list` is a rustc's option)
- #83712 (Update LLVM with another wasm simd fix)
Failed merges:
r? `@ghost`
`@rustbot` modify labels: rollup
Internal `if_chain!` lints
changelog: none
We use `if_chain!` a lot. So this enforces some style rules around it, internal only.
Lints when...
* Nested `if`/`if_chain!` can be collapsed
* An `if_chain!` starts with `let` or ends with `let ..; then {..}`
* An `if_chain!` has only one `if`
* An `if_chain!` contains `if .. && ..;` that spans multiple lines
lintcheck: warn if we get a bad exit status while running clippy
Right now we won't notice if a crate fails to build.
Print a warning message to indicate that there is a problem of some sort.
I'll still have to do more investigation on why this actually happens.
I suspect that the problem is that `clippy fix` might run --all-targets but when we download the crate source from crates.io, some path deps (used for internal tests etc...) are not available (which is usually not a problem because the internal tests are not needed when using the crate as a lib..?)
changelog: none
Destructure args in `methods`
changelog: none
This changes the main pattern in `methods` to match and destructure the method call args at the same time as the method name, and pass individual arg `Expr`s to the lint impls.
```rust
// before
["expect", ..] => expect::check(cx, expr, arg_lists[0]);
// after
("expect", [arg]) => expect::check(cx, expr, recv, arg);
```
This makes the code safer since there is no risk of out of bounds `args[n]` everywhere. There will be no more collecting `method_names`, `arg_lists`, `method_spans` as a separate step - everything comes out of the `match`es. Chained methods are parsed in a nested `match`. This makes the code more verbose in some ways, but IMO it is much easier to follow.
~Definitely should wait for #6896. Just putting out the idea.~
Refactor `Binder` to track bound vars
c.c. `@rust-lang/wg-traits`
This is super early (and might just get closed at some point), but want to get at least an initial idea of the perf impact.
r? `@ghost`
Lint: filter(Option::is_some).map(Option::unwrap)
Fixes#6061
*Please write a short comment explaining your change (or "none" for internal only changes)*
changelog:
* add new lint for filter(Option::is_some).map(Option::unwrap)
First Rust PR, so I'm sure I've violated some idioms. Happy to change anything.
I'm getting one test failure locally -- a stderr diff for `compile_test`. I'm having a hard time seeing how I could be causing it, so I'm tentatively opening this in the hopes that it's an artifact of my local setup against `rustc`. Hoping it can at least still be reviewed in the meantime.
I'm gathering that since this is a method lint, and `.filter(...).map(...)` is already checked, the means of implementation needs to be a little different, so I didn't exactly follow the setup boilerplate. My way of checking for method calls seems a little too direct (ie, "is the second element of the expression literally the path for `Option::is_some`?"), but it seems like that's how some other lints work, so I went with it. I'm assuming we're not concerned about, eg, closures that just end up equivalent to `Option::is_some` by eta reduction.
