>
>
> On Sun, 20 Jun 1999, Alexander Viro wrote:
> >
> > Huh? You mean that the right case here being *not* the regular symlink?
>
> No, the normal case being _nothing_. Neither symlink nor other wormhole.
Accepted, but in that respect both implementations are identical.
> > Up to you. But IMO you are mixing two seriously different objects
> > together and the cost is pretty high.
>
> Why?
>
> They are NOT different.
>
> They are only different because you _think_ they are different. Get over
> your hangups with old implementations of symlinks, and you will notice
> that it's _exactly_ the same issue as with any other "magic wormhole".
If you hope to get the meaningful behaviour with the same rules for
following wormholes and following symlinks I would like to hear details.
Really. That's where they differ big way.
Another thing about wormholes being: life would be much simpler if
we considered devices and FIFOs that way. We don't need a namespace to
keep a bunch of dentries. But that's another story...
> I agree about code duplication, but if that's your concern then you should
> just create a "generic_file_symlink()", the same way we condense the
> common code for "generic_file_read()" and "generic_file_write()". That has
> nothing to do with whether it is recursive or not.
Umm... Almost so, except that amount of stack wastage *does* matter here.
OK, I'll rewrite the patch into that form (not a big problem).
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/