nami misfeature, cont.
utzoo!henry
utzoo!henry
Mon Apr 5 19:10:04 AEST 1982
We seem to have an inconsistency ahead of us. Like this:
The Unix Time-Sharing System:
As another limiting case, the null file name refers to the
current directory...
Unix 3.0 INTRO(2):
Unless specifically stated otherwise, the null path name is
treated as if it named a non-existent file...
I can't find any places in the 3.0 manual where anything is "specifically
stated otherwise".
What to do? I am ambivalent about this one. There is something to be
said for the V7 interpretation, although the do-something-reasonable-for-
strange-input principle is weaker for system calls than for programs that
users invoke directly. On the other hand, objecting to null filenames
would have caught a number of bugs I've seen at one time or another.
Note that the -1 pointer points to a null filename in V7. So does
the NULL pointer if you are running split-space. On the whole, I think
any use of a null filename is likely to be the result of a bug, so it
should be trapped.
In practice, the 3.0 interpretation will probably triumph by sheer
weight of numbers.
More information about the Net.bugs
mailing list