Fix error message in test, run tests on AppVeyor #120
Conversation
fpoli
changed the title
|
This should fix #113, I think. |
|
Yes, but it doesn't solve the failing tests on AppVeyor |
|
No, I think I can fix that though. You're right, it's related to spaces in paths. |
|
Ok, I think if you do: --- a/test-project/tests/tests.rs
+++ b/test-project/tests/tests.rs
@@ -9,7 +9,7 @@ fn run_mode(mode: &'static str) {
config.mode = cfg_mode;
config.src_base = PathBuf::from(format!("tests/{}", mode));
- config.link_deps();
+ config.target_rustcflags = Some("-L target/debug -L target/debug/deps".to_string());
config.clean_rmeta();
compiletest::run_tests(&config);It should work. I know it's a hack, but I'd rather have these tests running at the moment, than trying to solve the bigger underlying problem in compiletest (namely #81). |
| //~| expected type `std::boxed::Box<Foo + std::marker::Send + 'static>` | ||
| //~| found type `std::boxed::Box<Foo + 'static>` | ||
| //~| expected type `std::boxed::Box<(dyn Foo + std::marker::Send + 'static)>` | ||
| //~| found type `std::boxed::Box<(dyn Foo + 'static)>` |
There was a problem hiding this comment.
This fix seems simple, but I don't really understand why this is the fix?
There was a problem hiding this comment.
It's just what the latest nighty rustc outputs... they probably changed the text of the message.
There was a problem hiding this comment.
Oh, interesting... That means that even if you write Box<Trait> the error message is going to write Box<dyn Trait> (or is it Box<(dyn Trait)>?). I wonder if that's intentional.
There was a problem hiding this comment.
That looks strange, indeed... I assumed that here we are testing that compiletest matches errors correctly, not that rustc produces the correct message.
There was a problem hiding this comment.
To be honest, I just copied some tests from rustc's test suite when developing the test project - it was just to verify that things worked as expected, and for use as a future benchmark.
I think it's a result of the "dyn everywhere" approach: rust-lang/rust@4b18085. In particular: https://github.com/rust-lang/rust/blob/master/src/test/ui/mismatched_types/trait-bounds-cant-coerce.stderr.
|
Nice, 🤞 that the tests work. Thank you so much for doing this! Once this is merged I'll put out a new version. |
src/runtest.rs
Outdated
| @@ -211,8 +211,12 @@ impl<'test> TestCx<'test> { | |||
|
|
|||
| fn check_correct_failure_status(&self, proc_res: &ProcRes) { | |||
| // The value the rust runtime returns on failure | |||
| const RUST_ERR: i32 = 1; | |||
| if proc_res.status.code() != Some(RUST_ERR) { | |||
| let rust_err: i32 = if cfg!(feature = "stable") { | |||
There was a problem hiding this comment.
I was afraid of this. The whole "support stable Rust" is starting to become unmaintainable :-/ The check_correct_failure_status() function is somewhat different in rustc/compiletest so this fix isn't that back-portable.
I suppose it's fine for now, but I think a bigger sync-up with the current rustc/compiletest would be a good idea. I just worry that it would stable support.
Would it be better to skip the tests for the stable feature?
There was a problem hiding this comment.
OK, thanks, I've also filed #121. Once this passes, we should be good to go. Thanks again!
|
0.3.13 is out! |
This fixes tests on Travis and enable tests on AppVeyor.
However, tests on AppVeyor are failing: https://ci.appveyor.com/project/fpoli/compiletest-rs/build/1.0.1
It's probably some Window-related issue (path separators...). I suggest to open a new issue, as I can't debug this easily.
Closes #113