Proposal: expose the drive functions in the filepath package

[Neil Mitchell]

Hi

> Neil is on holiday.

Neil is, but just saw this thread - although I think this makes it
clear why we need two week discussion periods - the deadline expires
before I get back to the UK!

I'm happy to expose the drive functions, its a trivial change, since
they get exposed for testing anyway. The first version of the filepath
library did expose them, and people complained, so I hid them.

Looking at the implementation of many of the higher level filepath
functions, to get them just right, you need the drive distinction.
This makes me suspect that to implement other functions on top of
filepath, you will also need the drive functions.

I'm broadly in favour of this - Igloo has a concrete use case that can
be done no other way so we definately need to solve it somehow.
Whether his particular function wants adding is another question.

Thanks

Neil