Re: how to kill a virtual group

[Eric Abrahamsen <eric@ericabrahamsen.net>]

Emanuel Berg <moasenwood@zoho.eu> writes:

> Eric Abrahamsen wrote:
>
>> I think it would be weird if Gnus let you edit a server
>> you'd defined in your .gnus.el file, and then those edits
>> were simply overwritten next time you restarted Gnus.
>
> But then you get the situation now where the same stuff has to
> be treated differently ... which should never happen.
>
> Something like this could be used from .gnus.el and *Server*,
> alike since they, while different, should differ on the
> interface level only ...
>
> (defun gnus-edit-server (srv params &optional redefine)
>  "Create or edit SRV with PARAMS.
> If REDEFINE is non-nil, overwrite existing data ..."
>  ... )
>
> If people then choose to ignore that and instead edit the data
> directly in .gnus.el then yes, that would reset any in-between
> session changes, be it from Lisp or *Server* or any other
> thinkable way for that matter - and that's completely fine, it
> is that way with everything, and so it should. Set the integer
> to something, change it to something else, if it is set again
> on the next init that's what it'll be - I mean, what else
> would one have it be and how ever would that work?
>
> So interactive interface from Emacs (buttons etc), interface
> from Lisp (Elisp code); they uses the same setters and
> functions one level below to do the actual business; and when
> done, no worry and no record where the data was set.

Well, I understand where you're coming from, but I don't think writing
all this code and changing Gnus' behavior is worth it for some
conceptual idea of consistency and purity. With proper messaging to the
user (and non-buggy code) it's just not a huge tragedy.