[RFC] Should CUSE4BSD move to the FreeBSD source tree?

[Hans Petter Selasky]

Hi,

The CUSE4BSD, character device in userspace, library and kernel module, 
has become an extremely popular driver choice for supporting a whole lot 
of USB devices under FreeBSD. There are now many clients using CUSE4BSD, 
cx88, uhidd, webcamd, and someone even mailed me a python wrapper for 
CUSE4BSD, so actually you can now implement a character device driver in 
a high level scripting language!

When upgrading the kernel and stuff like that, it would be more easy for 
people if CUSE4BSD was part of the kernel. Of course this puts more 
restrictions on the API, and CUSE4BSD sometime has API additions. Also 
that might mean that CUSE4BSD will go through some API changes, as input 
from other FreeBSD developers. One change might be to change lengths 
from "int" to "ssize_t" although, it is very unlikely to read more than 
a few megabytes at a time from a character device.

Also I wonder if putting CUSE4BSD in src, means that it is then also 
accepted as new *BSD standard?

Some other questions are: Where should CUSE4BSD reside in svn? Should it 
have it's own vendor branch or simply just @ head + MFC to xxx-stable?

Input and comments are appreciated!

--HPS

Reference:
http://www.selasky.org/hans_petter/cuse4bsd
/usr/ports/multimedia/cuse4bsd-kmod

+++ Happy new year to all of you +++