Previously, the borrow mode of each upvar was inferred as part of regionck. This PR moves it into its own separate step. It also employs the `ExprUseVisitor`, further simplifying the code. The eventual goal is to support better inference of `Fn` vs `FnMut` vs `FnOnce` that is not based on the expected type, as well as supporting individual by-move upvars.
r? @eddyb
and which uses EUV. For now, upvar inference is not any smarter than
it ever was, but regionck is simpler because it doesn't have to do as
many things at once.
`UnboxedClosureTyper`. This requires adding a `tcx` field to
`ParameterEnvironment` but generally simplifies everything since we
only need to pass along an `UnboxedClosureTyper` or `Typer`.
which should always result in an error.
NB. Some of the hunks in this commit rely on a later commit which adds
`tcx` into `param_env` and modifies `ParameterEnvironment` to
implement `Typer`.
expr-use-visitor) early. Turns out I was wrong to remove this; it
causes a lot of pain trying to run EUV etc during typeck without
ICEing on erroneous programs.
The `make_hash` function is used to prevent hashes of non-empty
buckets to collide with `EMPTY_HASH = 0u64`. Ideally this function
also preserve the uniform distribution of hashes and is cheap to
compute.
The new implementation reduces the input hash size by one bit, simply
by setting the most significant bit. This obviously prevent output
hashes to collide with `EMPTY_HASH` and guarantees that the uniform
distribution is preserved. Moreover, the new function is simpler (no
comparisons, just an OR) and (under the same assumptions as the old
function, i.e. only the least significant bit will contribute to the
bucket index) no additional collisions are caused.
After 8b3c67690c the `make install`
command fails if docs are not disabled through CFG_DISABLE_DOCS,
because now the `install` target uses
../../tmp/dist/$(DOC_PKG_NAME)-$(CFG_BUILD)/install.sh
In 714a2c678c the `prepare_install`
target wwas changed to conditionally depend also on the doc archive,
but did not modify `prepare_uninstall`.
Instead of explicitly depending on
dist/$(PKG_NAME)-$(CFG_BUILD).tar.gz, the `prepare_[un]install`
targets now depend on `dist-tar-bins`, which packages the appropriate
dist archives depending on the configuration.
This modifies `Parser::eat_lt` to always split up `<<`s, instead of doing so only when a lifetime name followed or the `force` parameter (now removed) was `true`. This is because `Foo<<TYPE` is now a valid start to a type, whereas previously only `Foo<<LIFETIME` was valid.
This is a [breaking-change]. Change code that looks like this:
```rust
let x = foo as bar << 13;
```
to use parentheses, like this:
```rust
let x = (foo as bar) << 13;
```
Closes#17362.
This corresponds to the JMM memory model's non-volatile reads and writes to shared variables. It provides fairly weak guarantees, but prevents UB (specifically, you will never see a value that was not written _at some point_ to the provided location). It is not part of the C++ memory model and is only legal to provide to LLVM for loads and stores (not fences, atomicrmw, etc.).
Valid uses of this ordering are things like racy counters where you don't care about the operation actually being atomic, just want to avoid UB. It cannot be used for synchronization without additional memory barriers since unordered loads and stores may be reordered freely by the optimizer (this is the main way it differs from relaxed).
Because it is new to Rust and it provides so few guarantees, for now only the intrinsic is provided--this patch doesn't add it to any of the higher-level atomic wrappers.
