tape stops logging when using mock-fs

[nemoxps]

I'm not sure what's going on, but if mock-fs can't handle fs.writeSync(1, content), then it might be a bug in that library.

Take a look at tape-testing/tape#354.

[tschaub]

Thanks for the report @nemoxps.

[ryanblock]

This bit me today (in Windows, not on Mac or *nix), any thoughts on how we might approach a fix?

[tschaub]

Here is one approach: 80d318f. This only works for stdin/out/err streams connected to a text terminal. For non-TTY environments, the process never exits – I didn't spend enough time to figure it out, but maybe someone else can. There are likely other ways to bypass the mock filesystem for the special 0, 1, and 2 file descriptors.

[ryanblock]

I'm back because I was once again bitten by this bug! (lol)

No opinion on that specific fix but in theory I'm happy to help write some tests around this.

[filmaj]

I have also been bit by this bug 🙈

Same here, I am not familiar enough with mock-fs internals to say which approach should be taken, but for what it is worth, the std branch of mock-fs fixes the issue! (I tried it out on a PR in our project: architect/logs#311)

[filmaj]

Friendly bump 😄
I hit this bug again today haha. I always forget about this, and then remember, and end up back here.
What are the chances of getting the std branch that is pushed up to the origin of this repo merged into mainline?

[tschaub]

See how things go with mock-fs@4.14.

[filmaj]

Awesome, thanks so much @tschaub ! Works great for us over in the @architect project!