Ben Fennema wrote:
> Is there some kind of future plan for a common 'fork' access methidology?
>
> can I do something like inode->u.udf_i->streamdir = iget(stream dir ino num)
> then, using ioctl calls on the file, access the files in streamdir?
>
> (Keep using the kernel functionallity to access the files, but not have
> them part of the directory hierarchy)
That still has the problem where one files' contents (or metadata) is a
function of another. Race city.
At very least, the aliasing inodes should probably share locks or
something.
I don't think the dentry structure is intrinsically such a problem,
though the current code might be.
-- Jamie
