Re: Back to emacsclient/server

[Juanma Barranquero]

On 10/27/06, Jason Rumney <address@hidden> wrote:

> I don't think repeatability is realistic within Emacs,
> because a Lisp function cannot have any control over
> other functions that might have
> called random before it.

Still, if the initial seed is always the same and you control the code
that runs, you can have repeatability. Viz the current situation,
where I'm getting the same authentication string in different
invocations of Emacs :)

> If repeatability is desired, then we need a
> function to set the random seed to a known value,
> so Lisp code that requires repeatability can call
> it at the start of its psuedorandom number generation.

...which is exactly what I was proposing with (random-seed SEED) and
`with-tmp-seed' (or perhaps `with-current-seed') :)

But all this has no bearing on the emacsclient/server issue. All I
need to know is whether people thinks is better to call (random t) in
server.el, or note the need to do so in the documentation.

--
                   /L/e/k/t/u