> 3) Make isdisk() and istty() machine-independent, and add istape().
   > It occurs to me that perhaps these should just be deleted instead,
   > though, and their usages changed to directly access the switch table.

   actually, it seems they'd make rather good macros...  8-)

I thought of that.  But...

I made the changes in my personal tree, and found that all but one of
the uses (in vfs_bio.c) already pokes at the switch table.  Now I'll
admit the switch table should become dynamic at some point, but it
isn't going to add noticably to that work to explicitly peek at it
right now, and I have visions of 2 dozen is*() macros appearing over
time.

Therefore, I say torch the macros.  B-)