Josh Stone

Ah, I suppose that's non-trivial, and probably rules out my hope of getting this into `std`'s own backtracing. But for general use, a new cargo feature sounds reasonable to me.

> I think this is subsumed by #7916 It doesn't seem so. I tried by running `cargo vendor >.cargo/config`, then removing some of the dev-deps. I could still build with...

Please note, I've already identified the likely LLVM fix in that Fedora bug report. It's also fine with LLVM 9.

So the solution is to patch your LLVM, as I intend to do in Fedora 30. It's not really a bug in compiler-builtins -- it just started using functionality that's...

> I don't know if rustc returns different error codes in those cases. It does not -- I get exit code `1` for regular errors and for that PGO error....

I gather you have applied this patch (or something like it) to your Firefox builds. How has that worked out in practice? Any more instances like @glandium's where you would...

Another way to use it is to simply `clone()` the `AutoCfg`, enable the feature and test with it, then drop the clone. If we did a `with_features` thing, I would...

Does cargo communicate this in any way? Maybe an environment variable?

Hmm, normally that output is captured and hidden, and even `cargo build -v` doesn't show anything, but `cargo build -v -v` does. https://github.com/rust-lang/cargo/blob/8e403e5e77e1fb5ebdba3094c5cf424bfc473f9e/src/cargo/core/compiler/custom_build.rs#L353-L370 We could add some debug/verbosity control in...

I'm open to it. We could add some nightly detection in the version part of the API (and dev is the equivalent), but I would also want to see some...