Re: Suggested dual human/binary interface for proc/devfs

[Jamie Lokier]

Alexander Viro wrote:
> > - Mapping the structure inherent to the USB messages to a flat
> >   stream of bytes means the need for a parser in kernel space
> 
> Erm? Even if you s/ioctl()/write()/?

Or indeed, sendto/recvfrom.

A USB tree is a little network, just like IRDA.  The interface is there.
There are even good generic tools that work with it.

-- Jamie