Re: Gnus in Emacs is updated to No Gnus 0.7

[Daiki Ueno]

2007/10/29, Michael Albinus <address@hidden>:
> > I can't remember the previous discussion, but I think it's not so bad
> > for *applications* (such as Gnus) to use password.el.  I'd just say
> > that it's bad if it is implicitly used by low-level *libraries* such
> > as pgg.el.
>
> Tramp uses also password.el if it is in the load path. It is a
> convenient library if somebody uses an scp-like method, but does not run
> ssh-agent.
>
> It would be important to know whether there are deficiencies which would
> make the use of password.el impossible for Tramp.

My recommendation is simple: do password caching things at the
(single) place closest to a human user, or do nothing (just trust
ssh-agent/gpg-agent).  Otherwise the user will have to track down the
code between the UI and ssh/gpg to make sure where his secret is
really cached.

If Tramp had been designed to be mainly used by other programs rather
than human users, it might have caused the deficiencies.  However, I
believe it's not.  i.e. Tramp is an *application* rather than a
*library*.

Yes, password.el is a library itself.  But no one will be confused
about what it does.

Regards,
-- 
Daiki Ueno