doc: discuss build script instruction order
### What does this PR try to resolve?
It is currently not documented that the order of build script `cargo:` instructions may be relevant to linking.
This has caused issues such as: https://github.com/rust-lang/rust/issues/96328
### How should we test and review this PR?
Build/view documentation.
### Additional information
- Cargo maintainers should fact check my wording.
- We may need to discuss if this should also be documented for `rustc`
- Maintainers should ensure that this change does not prevent a change in what is currently unspecified behavior. Perhaps `cargo` will want to rearrange link arguments itself to resolve issues in the future?
Fix use of .. in dep-info-basedir
### Summary
This allows setting, in .cargo/config's dep-info-basedir, some relative path that goes above the crate's directory.
### Motivation
In a setup like this:
```
repo_root
├── Makefile
├── some_c_things
│ └── foo.c
└── rust_things
├── Cargo.toml
└─── src
└── lib.rs
```
If you want the generated .d files to be includable directly in the Makefile (without post-processing), you need them to mention paths relative to the root, like:
rust_things/target/....: rust_things/src/lib.rs
### Implementation
For this you need to have relative paths with parent directories (in this case ..) in dep-info-basedir, which does not work without the change in this PR (due to render_filename doing only strip_prefix, while the basedir still contains literal ..s).
Let me know if this change is acceptable. Another implementation could be to canonicalize in ConfigRelativePath::resolve_path instead, especially since that struct outputs absolute paths. But that would have it access the filesystem, while it currently doesn't.
This patch implements more complete logic for applying rustflags to
build scripts and other host artifacts. In the default configuration, it
only makes build scripts (and plugins and whatnot) pick up on
`rustflags` from `[host]`, which fixes#10206 but otherwise preserves
existing behavior in all its inconsistent glory. The same is the case if
`target-applies-to-host` is explicitly set to `false`.
When `target-applies-to-host` is explicitly set to `true`, rustflags
will start to be applied in the same way that `linker` and `runner` are
today -- namely they'll be applied from `[target.<host triple>]` and
from `RUSTFLAGS`/`build.rustflags` if `--target <host triple>` is given.
The tests in question all tail to even build the build script, so the
contents of `main` don't matter, and make the tests seem more complex
than they really are.
Remove `build_script::build_script_scan_eacces` test case because cargo
ignores it and returns its path during a `cargo build`. The caller still
has a chance to hit the IO error if they does access it.
This works by introspecting rustc's error output, using the JSON format
to determine whether it's a warning or error, then skipping it
altogether if it's a summary of the diagnostics printed.
Before:
```
src/main.rs:1:10: warning: trait objects without an explicit `dyn` are deprecated
src/main.rs:1:1: error[E0601]: `main` function not found in crate `wrong`
src/main.rs:1:9: error[E0038]: the trait `Clone` cannot be made into an object
error: aborting due to 2 previous errors; 1 warning emitted
error: could not compile `wrong`
```
After:
```
$ cargo check --message-format short
src/main.rs:1:10: warning: trait objects without an explicit `dyn` are deprecated
src/main.rs:1:1: error[E0601]: `main` function not found in crate `wrong`
src/main.rs:1:9: error[E0038]: the trait `Clone` cannot be made into an object
error: could not compile `wrong` due to 2 previous errors; 1 warning emitted
```
Configure hosts separately from targets when --target is specified.
This prevents target configs from accidentally being picked up when cross compiling from hosts that have the same architecture as their targets.
closes#3349
Sample `shoo.d` file prior to this change is below, note the `build.rs`
at the end, which was not from my package.
From booping the debugger, I found this was coming from
`compiler_builtins`. This is not really their bug though: if a build.rs
asks for rerun-if-changed on some crate relative path, this will happen
in general. So I've fixed it in Cargo and added a test to prevent it
regressing.
```
target/riscv64imac-mu-shoo-elf/release/shoo: /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/core_arch_docs.md /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/macros.rs /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/mod.rs /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/simd.rs /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/simd_llvm.rs crates/build_bits/src/lib.rs shoo/src/main.rs shoo/src/task.rs shoo/src/vectors.s build.rs
```
This change fixes it so it's like:
```
target/riscv64imac-mu-shoo-elf/release/shoo: /home/jade/.cargo/registry/src/github.com-1ecc6299db9ec823/compiler_builtins-0.1.39/build.rs /home/jade/.cargo/registry/src/github.com-1ecc6299db9ec823/log-0.4.14/build.rs /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/core_arch_docs.md /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/macros.rs /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/mod.rs /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/simd.rs /home/jade/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/../../stdarch/crates/core_arch/src/simd_llvm.rs crates/build_bits/src/lib.rs shoo/src/main.rs shoo/src/task.rs shoo/src/vectors.s
```
This prevents target configs from accidentally being picked up when
cross compiling from hosts that have the same architecture as their
targets.
closes#3349
"exit code" is wrong terminology on Unix. I am trying to fix this in
Rust stdlib in
https://github.com/rust-lang/rust/pull/83462
but this currently breaks the cargo test suite.
See that MR for full explanation of the change.
Signed-off-by: Ian Jackson <ijackson@chiark.greenend.org.uk>
By doing so, rustdoc will also emit workspace-relative filenames for the doctests.
This was first landed in #8954 but later backed out in #8996 because it changed the CWD of rustdoc test invocations.
The second try relies on the new `--test-run-directory` rustdoc option which was added in https://github.com/rust-lang/rust/pull/81264 to explicitly control the rustdoc test cwd.
fixes#8993
If a package is tested and the library for the package wasn't built
(e.g. only tested or it wasn't present) then the `links` env vars from
dependencies weren't showing up to the build script by accident. This
was an accidental regression from #8969.
The intention of #8969 was to exclude connections to build scripts
connected via dev-dependencies, but it only applied a heuristic because
the unit graph doesn't retain information about dev-dependencies. The
fix here is to instead actually retain information about
dev-dependencies which is only used for constructing the unit graph and
connecting build script executions to one another.
Closes#9063