RE: Genericity, strong typing, SOAP-RPC, GRASP, and defining the Web

Jeff Bone wrote,
> > That's a good example, but it also does a good job of showing up
> > the limitations of that sort of extremism. What are all those 
> > ioctls for if the Plan 9/UNIX model is really so uniform? Can you 
> > really accept() on a _file_ descriptor as opposed to a listening 
> > socket descriptor? Is the uniformity reality or ideology?
> Notice that I said Plan [9, woops] *vis-a-vis* UNIX.  Plan 9 doesn't >
have an ioctl call, AFAIK:

It might not have something spelled "ioctl", but it must have its
moral equivalent: there's a difference between a descriptor associated
with a file on a tape drive and one associated with a listening 
socket, and that's got to be reflected _somewhere_ (either that or 
Plan 9 is lousy at dealing with one or other of tapes or sockets ;-)



Received on Tuesday, 26 March 2002 21:20:08 UTC