Partial support for running UI tests with download-rustc
#103969
Conversation
|
r? @jyn514 (rustbot has picked a reviewer for you, use r? to override) |
rustbot
added
T-bootstrap
src/bootstrap/lib.rs
Outdated
| @@ -1034,13 +1034,17 @@ impl Build { | |||
| } | |||
|
|
|||
| fn debuginfo_map_to(&self, which: GitRepo) -> Option<String> { | |||
| if !self.config.rust_remap_debuginfo { | |||
| if !self.config.rust_remap_debuginfo && self.config.download_rustc_commit.is_none() { | |||
There was a problem hiding this comment.
why would this matter? if we're not remapping debuginfo, we shouldn't need this either way, right?
There was a problem hiding this comment.
Compiletest only replaces the /rustc/$sha1 paths with a placeholder when the CFG_VIRTUAL_RUST_SOURCE_BASE_DIR environment variable is passed to it at build time, and right now this is the way to set that.
I guess I could change the implementation and instead of changing this setting globally, pass to compiletest another prefix to normalize when download-rustc is enabled.
There was a problem hiding this comment.
ah I see, it needs to remap the source directory because the one we've downloaded from CI is not in the path the tests expect. That seems reasonable, can you add it as a comment?
There was a problem hiding this comment.
I guess I could change the implementation and instead of changing this setting globally, pass to compiletest another prefix to normalize when download-rustc is enabled.
Ended up with this approach, as there is way less risk of unintended side effects this way.
jyn514
added
S-waiting-on-author
This comment has been minimized.
This comment has been minimized.
This comment was marked as outdated.
This comment was marked as outdated.
pietroalbini
force-pushed
the
pa-download-rustc-ui-tests
branch
from
79a7621 to
8c5e6a8
Compare
|
☔ The latest upstream changes (presumably #104188) made this pull request unmergeable. Please resolve the merge conflicts. |
pietroalbini
force-pushed
the
pa-download-rustc-ui-tests
branch
from
8c5e6a8 to
470423c
Compare
pietroalbini
added
S-waiting-on-review
|
@bors r+ |
|
🌲 The tree is currently closed for pull requests below priority 1. This pull request will be tested once the tree is reopened. |
bors
added
S-waiting-on-bors
| // Source base on the sysroot (from the src components downloaded by `download-rustc`): | ||
| Some(self.config.sysroot_base.join("lib").join("rustlib").join("src").join("rust")), | ||
| // Virtual `/rustc/$sha` remapped paths (if `remap-debuginfo` is enabled): | ||
| option_env!("CFG_VIRTUAL_RUST_SOURCE_BASE_DIR").map(PathBuf::from), |
There was a problem hiding this comment.
Nice! Could use this in cg_clif too: https://github.com/bjorn3/rustc_codegen_cranelift/blob/d1e499aa2d5e16eab6b70c04b28bd04b67de3c60/scripts/setup_rust_fork.sh#L30-L47
Edit: done in bjorn3/rustc_codegen_cranelift@9723c79
…iaskrgr Rollup of 10 pull requests Successful merges: - rust-lang#103117 (Use `IsTerminal` in place of `atty`) - rust-lang#103969 (Partial support for running UI tests with `download-rustc`) - rust-lang#103989 (Fix build of std for thumbv7a-pc-windows-msvc) - rust-lang#104076 (fix sysroot issue which appears for ci downloaded rustc) - rust-lang#104469 (Make "long type" printing type aware and trim types in E0275) - rust-lang#104497 (detect () to avoid redundant <> suggestion for type) - rust-lang#104577 (Don't focus on notable trait parent when hiding it) - rust-lang#104587 (Update cargo) - rust-lang#104593 (Improve spans for RPITIT object-safety errors) - rust-lang#104604 (Migrate top buttons style to CSS variables) Failed merges: r? `@ghost` `@rustbot` modify labels: rollup
Right now trying to run UI tests with
download-rustcresults in a bunch of test failures, requiring someone who wants to only work on tests to also build the full compiler. This PR partially addresses the problem by solving a lot of the errors (but not all).rust-srccomponent on CI toolchains, since the test output depends on whether the standard library source code is installed; We can't just symlink the current source because therustc-devcomponent already includes the compiler sources, but not the library sources, and mixing things is worse IMO.$SRC_DIRnormalization done by compiletest correctly handles the source code added above.remap-prefixto the prefix used in the downloaded toolchain, to ensure compiletest normalizes it.