On Sun, 22 Jul 2001, Terry Lambert wrote:
> In fact, "man chflags", and look at the "-L" argument... I
> could make a good argument that it should operate on the
> link itself, if given a "-l" (currently unused) argument.
That was my expected result until I read the manpage completely and
followed-up through the code. Initial testing and a trawl through the code
shows that all ufs symlinks, at least, are first-class vnodes and support
flags. NFS returns EOPNOTSUPP, cd9660, union etc returns EROFS.
> Pushing the link following semantics into the kernel, instead
> of the C library, was a mistake in the first place; it precludes
> easy implementation of things like variant symbolic links,which
> are easily handled in a user space library routine that wraps
> the actual system call.
Possibly, but shouldn't we be wary of changing syscall semantics?
Especially in code that relates to securelevels.
I guess there is general agreement that it is desirable to be able to set
schg,sunlink etc on symlinks and fifos.
The consistency argument goes like this:
Currently exposed as accessor methods to VOP_SETATTR are:
chmod(2), fchmod(2), lchmod(2)
chown(2), fchown(2), lchown(2)
chflags(2), fchflags(2)
History records that the semantics for following symlinks in chown & chmod
date from 4.4BSD.
Since I have been a freebsd admin for some years without giving anything
back, I would like to put this together.
Joshua
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
