This commit adds support to the makefiles, configuration script, and build
system to understand MUSL. This is broken up into a few parts:
* Any target of the form `*-musl` requires the `--musl-root` option to
`./configure` which will indicate the root of the MUSL installation. It is
also expected that there is a libunwind build inside of that installation
built against that MUSL.
* Objects from MUSL are copied into the build tree for Rust to be statically
linked into the appropriate Rust library.
* Objects for binary startup and shutdown are included in each Rust installation
by default for MUSL. This requires MUSL to only be installed on the machine
compiling rust. Only a linker will be necessary for compiling against MUSL on
a target machine.
Eventually a MUSL and/or libunwind build may be integrated by default into the
build but for now they are just always assumed to exist externally.
This adds a new `#[cfg]` matcher against the `target_env` property of the
destination target triple. For example all windows triples today end with `-gnu`
but we will also hopefully support non-`gnu` targets for Windows, at which point
we'll need to differentiate between the two. This new `target_env` matches is
provided and filled in with the target's environment name.
Currently the only non-empty value of this name is `gnu`, but `musl` will be
shortly added for the linux triples.
Fix `make_command_line` for the case of backslashes at the end of an
argument requiring quotes. We must encode the command and arguments
such that `CommandLineToArgvW` recovers them in the spawned process.
Simplify the logic by using a running count of backslashes as they
are encountered instead of looking ahead for quotes following them.
Extend `test_make_command_line` to additionally cover:
* a leading quote in an argument that requires quotes,
* a backslash before a quote in an argument that requires quotes,
* a backslash at the end of an argument that requires quotes, and
* a backslash at the end of an argument that does not require quotes.
I figure I'd start easy with fixing a simple documentation bug. This is also a test to see that I got everything right w r t the fork/pull request process.
This changes the `ToTokens` implementations for expressions, statements, etc. with almost-trivial ones that produce `Interpolated(*Nt(...))` pseudo-tokens. In this way, quasiquote now works the same way as macros do: already-parsed AST fragments are used as-is, not reparsed.
The `ToSource` trait is removed. Quasiquote no longer involves pretty-printing at all, which removes the need for the `encode_with_hygiene` hack. All associated machinery is removed.
New `Nonterminal`s are added: NtArm, NtImplItem, and NtTraitItem. These are just for quasiquote, not macros.
`ToTokens` is no longer implemented for `Arg` (although this could be added again) and `Generics` (which I don't think makes sense).
This breaks any compiler extensions that relied on the ability of `ToTokens` to turn AST fragments back into inspectable token trees. For this reason, this closes#16987.
As such, this is a [breaking-change].
Fixes#16472.
Fixes#15962.
Fixes#17397.
Fixes#16617.
An actual typeck error is the cause of many failed compilations but an
unrelated bug is being reported instead. It is triggered because a typeck
error is presumably not yet identified during compiler execution, which
would normally bypass an invariant in the presence of other errors. In
this particular situation, we delay the reporting of the bug until
abort_if_errors().
Closes#23827, closes#24356, closes#23041, closes#22897, closes#23966,
closes#24013, and closes#23729
**There is at least one situation where this bug may still be genuinely
triggered (#23437).**
Transplant the relevant changes (turns out to be all of them) to `grammar.md`, and remove all grammar talk from `reference.md`. Sorry for the chaos.
The second commit, further below, goes over the comments and whitespace sections.
r? @steveklabnik
This changes the `ToTokens` implementations for expressions, statements,
etc. with almost-trivial ones that produce `Interpolated(*Nt(...))`
pseudo-tokens. In this way, quasiquote now works the same way as macros
do: already-parsed AST fragments are used as-is, not reparsed.
The `ToSource` trait is removed. Quasiquote no longer involves
pretty-printing at all, which removes the need for the
`encode_with_hygiene` hack. All associated machinery is removed.
A new `Nonterminal` is added, NtArm, which the parser now interpolates.
This is just for quasiquote, not macros (although it could be in the
future).
`ToTokens` is no longer implemented for `Arg` (although this could be
added again) and `Generics` (which I don't think makes sense).
This breaks any compiler extensions that relied on the ability of
`ToTokens` to turn AST fragments back into inspectable token trees. For
this reason, this closes#16987.
As such, this is a [breaking-change].
Fixes#16472.
Fixes#15962.
Fixes#17397.
Fixes#16617.
At https://doc.rust-lang.org/book/vectors.html, there should be a link to
Generics page but the link address is ommitted and thus link is not functioning
well. So I added a link definition to the vectors.md.
r? @steveklabnik
Fix `make_command_line` for the case of backslashes at the end of an
argument requiring quotes. We must encode the command and arguments
such that `CommandLineToArgvW` recovers them in the spawned process.
Simplify the logic by using a running count of backslashes as they
are encountered instead of looking ahead for quotes following them.
Extend `test_make_command_line` to additionally cover:
* a leading quote in an argument that requires quotes,
* a backslash before a quote in an argument that requires quotes,
* a backslash at the end of an argument that requires quotes, and
* a backslash at the end of an argument that does not require quotes.
At https://doc.rust-lang.org/book/vectors.html, there should be a link to
Generics page but the link address is ommitted and thus link is not functioning
well. So I added a link definition to the vectors.md.
r? @steveklabnik
