Coverage reported as 0% when using cargo test --all

[euclio]

See report here: rust-rosetta/rust-rosetta#563

[roblabla]

Looking into it

[roblabla]

https://travis-ci.org/Hoverbear/rust-rosetta/builds/205573979#L4329 Cargo coveralls isn't running any tests :|

[roblabla]

could you try cargo coveralls --all ?

[euclio]

Build log with cargo coveralls --all: https://travis-ci.org/Hoverbear/rust-rosetta/builds/205580675

[roblabla]

So at least this time it ran the tests. The problem is, kcov seems to choke when trying to open some directories. I cloned your repo and managed to reproduce the issue. I'll play around with it and report back when I get it to work :).

[roblabla]

So this problem is due to the way workspaces work. The "target" directory isn't next to the cwd() in this case. I'm refactoring the code to use cargo methods to recover the location of the "target" directory. This should be much more robust, and allow it to also support things like CARGO_TARGET_DIR

[roblabla]

0.0.3 is published on crates.io. If you run the builds again, it should work.

[euclio]

Ah, thank you for the quick work! It recorded coverage, but also took the coverage of some library dependencies (...? I see glium, glutin etc. in the covered files). So, it's giving incorrect coverage. Coverage log here: https://coveralls.io/builds/10337936

[roblabla]

You're right. It seems that it's more problems with workspaces. I'm investigating.

[roblabla]

Fixed with 2581789 . 0.0.4 released.

[roblabla]

Woops. This wasn't quite fixed, and I forgot to reopen it.

https://coveralls.io/builds/10347968 It's really weird, coveralls seems eager to include files that are not in the include-path. I need to investigate this more. I'll try to go over it this week-end.