[Bug 69] Generalize SSH_ASKPASS

[David Woodhouse]

On Wed, 2003-03-26 at 20:32, Jim Knoble wrote:
> Circa 2003-03-26 12:59:49 +0000 dixit David Woodhouse:
> 
> : > as to typing "YES", this is intentional, since users should _think_
> : > not click.
> : 
> : I agree. And it's not really hard for the ssh-askpass program to
> : recognise the question text and do something other than the normal 'ask
> : for textual input' anyway.
> 
> Actually, i would rather not have ssh-askpass try to do anything with
> the text it's given except display it. 

Yeah, I suppose you're probably right. Parsing text strings isn't really
ideal, especially if they're localised. I'm not too bothered about
typing 'yes', and I can sort of see the point in forcing the user to do
that explicitly although I suspect it's a little OTT -- but I really do
want ssh-askpass to handle s/key for me :)

> If you want a separate behavior, that request should be explicit:
> either an explicit option to ssh-askpass (e.g., 'ssh-askpass --yesno')
> or (probably better) a separate program (e.g., 'ssh-confirm').
> 
> Care to come up with a specification?

Hmmm. We really do have to be careful about backwards compatibility. So
a separate program probably accompanied by a separate environment
variable for it (SSH_CONFIRM?) is likely to be the best way forward.

-- 
dwmw2