Apart from making the build system determine the LLDB version, this PR also fixes an issue with enums in LLDB pretty printers. In order for GDB's pretty printers to know for sure if a field of some value is an enum discriminant, I had rustc mark discriminant fields with the `artificial` DWARF tag. This worked out nicely for GDB but it turns out that one can't access artificial fields from LLDB. So I changed the debuginfo representation so that enum discriminants are marked by the special field name `RUST$ENUM$DISR` instead, which works in both cases.
The PR does not activate the LLDB test suite yet.
The old version switched in between examples from the value `5i` to `"Hello"` and back.
Additionally, the code generated by `rustc print.rs --pretty=expanded` is not as verbose anymore.
LLDB doesn't allow for reading 'artifical' fields (fields that are generated by the compiler). So do not mark, slice fields, enum discriminants, and GcBox value fields as artificial.
This provides a way to pass `&[T]` to functions taking `&U` where `U` is
a `Vec<T>`. This is useful in many cases not covered by the Equiv trait
or methods like `find_with` on TreeMap.
This provides a way to pass `&[T]` to functions taking `&U` where `U` is
a `Vec<T>`. This is useful in many cases not covered by the Equiv trait
or methods like `find_with` on TreeMap.
Currently, the Guide says tuples "are only equivalent if the arity, types, and values are all identical", before presenting an example that uses `==` to compare two tuples whose arity and contained types match. This is misleading, because it implies that `==` can dynamically check whether two tuples have the same arity and contained types, whereas trying to do this would lead to a compiler error.
I tried to avoid destroying the flow of this section, but I'm not sure if I've been successful.
Because my '30 minute intro' was originally a blog post, the tone was a bit too light. It also was written a long time ago, and deserves a bit of a refresher for modern Rust. now that my work on the Guide is wrapping up, I want to give it a quick re-write as well.
This is not yet done, but I'm submitting it for feedback so far. I'd really like some comments on the ownership part in particular, which gets lower level than before, but is not strictly 100% accurate. Trying to strike a balance.
In general, I'm not sure I go into enough detail for those without systems experience, but am afraid of too much detail for those that do.
Rendered view: https://github.com/steveklabnik/rust/blob/intro_redux/src/doc/intro.md
/cc @wycats @nikomatsakis @brson etc
The old version switched in between examples from the value `5i` to `"Hello"`
and back. Additionally, the code generated by `rustc print.rs
--pretty=expanded` is not as verbose anymore.
Adds a high-level discussion of "what collection should you use for what", as well as some general discussion of correct/efficient usage of the capacity, iterator, and entry APIs.
Still building docs to confirm this renders right and the examples are good, but the content can be reviewed now.
For example, this matcher: `fn $name:ident( $($param:ident : $pty:ty),* )` would fail when parsing `fn foo()`, because macro parser wouldn't realize that an ident cannot start with `)`.
This resolves#5902, and at least partially mitigates #9364 and #3232.
[Previously](e5da6a71a6), the `Any` trait was split into a private portion and an (empty) public portion, in order to hide the implementation strategy used for downcasting. However, the [new rules](e9ad12c0ca) for privacy forbid `AnyPrivate` from actually being private.
This patch thus reverts the introduction of `AnyPrivate`.
Although this is unlikely to break any real code, it removes a public trait and is therefore a:
[breaking-change]
[Previously](e5da6a71a6),
the `Any` trait was split into a private portion and an (empty) public
portion, in order to hide the implementation strategy used for
downcasting. However, the [new
rules](e9ad12c0ca)
for privacy forbid `AnyPrivate` from actually being private.
This patch thus reverts the introduction of `AnyPrivate`.
Although this is unlikely to break any real code, it removes a public
trait and is therefore a:
[breaking-change]
This PR adds support in rustdoc for properly naming lifetimes in bounds, instead of just showing `'static` for everything. It also adds support for unboxed function sugar bounds, which were also previously rendered as `'static`.
Previously, external code might call `markdown::render()` without having
called `markdown::reset_headers()`, meaning the TLS key
`used_header_map` was unset. Now `markdown::render()` ensures that
`used_header_map` is set by calling `reset_headers` if necessary.
Fix#17736
