Alex Lyon

Yeah, I'm just writing down some notes. The current version is a good enough jumping off point though.

Can you test the coverage on the `sh-parser` branch? It no longer uses `nom` so I'm curious if that helped at all.

Not sure what the problem is then :/

There’s no guideline per se at the moment, but generally what you want to do is implement a type that derives `Fail` (such as `ChmodError`) and then map errors from...

I mean the method is generally pretty consistent (I don’t think it’s possible to make the setup fully general as the utilities have widely different errors) except for a few...

Everything using `display_err!()` should switch to `display_msg!()` as we are getting rid of the `error: ` prefix. `cat` also prints errors (on phone so I can’t link). If the utilities...

It should be `|=`, so there must have been a typo :/ I’d need to spend some time to review but IIRC `chmod` returns `Err()` on fatal errors whereas `Ok(1)`...

We might want to open this as an issue in `assert_cmd`. It's hard to replicate though as it only seems to occur on Travis.

Haven't really checked to see what is different between the targets but the `COVERAGE=1` target never has this issue (at least for the 5 or so builds I've checked).

Ah, that makes sense.