Re: Framework extending window functions for Follow Mode (etc.).

[Alan Mackenzie]

Hello, Martin.

On Sat, Nov 07, 2015 at 04:18:18PM +0100, martin rudalics wrote:
>  > I don't think that would be the Right Thing (if I've understood it
>  > properly, that is).  With Follow Mode active, which window start
>  > window-start returns is primarily about the context of the window-start
>  > invocation; it's not about some windows always doing one thing, some
>  > always doing another.

> I don't understand why we would want to hardcode a GROUP argument.

It would appear to be a good way of solving the immediate problem, that
is, of being able to address a Follow Mode group of windows and a single
window in a unified fashion.  It thus avoids having explicitly to code
Follow Mode stuff into random .el (or .c) files.

The immediate use case is in Isearch, where currently Isearch doesn't
play well with Follow Mode.  See the stuff relating to bug #17453 in
debbugs.

> When we encounter a similar problem with foo-mode what are we supposed
> to do?  Use the GROUP argument?  Add a FOO argument?

Yes, I would say.  I'm guessing your concern is that such arguments on
some functions could proliferate, giving an unmanageable number of
them.  But what is the likelihood of this happening?  Or, in particular,
of it happening to these particular window handling functions?

> martin

-- 
Alan Mackenzie (Nuremberg, Germany).