Feature Request: FailFS

[0xDEC0DE]

The MemFS is great for unit tests, and I absolutely adore your implementation. But it does present challenges when attempting to test failure modes, since it, you know, works really well.

I was able to achieve limited success testing failure-to-write scenarios by loading an RoFS view over the MemFS and running tests, and I thought perhaps there would be utility in taking that idea further.

So I came up with this for my tests:

retr0h/gilt@6293c13

...and some snippets of how it is implemented/used:

retr0h/gilt@52f0dda

It's fidgety, cantankerous, and prone to panics if you don't set it up right, but there might be the precursor to something useful if left in more capable hands.

Would this be worth developing further?

[psadac]

Your idea and your implementation is very clever, but it's too panicky for my taste. So I have implemented a much dumber one, probably less panic friendly, which only necessitates only one function to generate failures. However I am not sure it will work for you.