Re: [9fans] clone files and Srv->clone

[Francisco J Ballesteros <nemo@lsub.org>]

srv->clone clones a client fid. You can think of it like
doing a dup of a file descriptor. You obtain a new fid that is a clone
of the old one. Try man 5 clone.

Walk1 walks a name. For example, you might have a fid for your
/ file. After a walk1 "a" on that fid, your fid now points to "/a". See
man 5 walk.

You might do a clone file but that has nothing to do with walk and clone.
For example, you might arrange for your srv->open to create a new file
each time, and then make the fid point to it. That way, each open(2)
would obtain a cloned file.

Run for example ramfs with -D, play with touch ls and the like, and look
at the 9p messages printed by ramfs. Section 5 describes the whole
protocol.

hth

On 12/10/05, Sascha Retzki <sretzki@gmx.de> wrote:
> Evenin',
>
>
> I've got a question about the process of cloning in the 9p(2) sense and
> in the sense of e.g. webfs does: Is it the same?
>
> Or, to explain my problem from the top-down perspective, I want my lil
> file-server to have a clone-file. On read, a new directory with a tree
> is created and the name of this new directory is returned. Is it that I
> have to do that mostly manually, or can Srv->walk1 and Srv->clone help
> me? If yes, how (rougly)? Or am I mistaken, and there is some all-in-one
> solution to my problem (because many file-server do it in this way, I
> assume there is some library-calls for that?)
>
>
> Thanks, Sascha
>
>