RE: Extend gdb to filter registers

[Drew Adams <drew.adams@oracle.com>]

>   > That said I think that the current windowing model
>   > has a surprising semantic for dedicated, namely that
>   > a dedicated window is allowed to be split.
> 
> That does surprise me -- but with so many possible cases, can we be
> sure that it isn't sometime desirable to split those windows?
> 
> It should be easy to add a global variable to control this.
> Its default could tentatively be not to split dedicated windows.
> If anyone is disappointed with that change, we would find out.

I apologize for jumping in here out of the blue,
and without even have read the (long) thread.
(I don't use gdb.)

I'd just like to say this -

I use dedicated windows for all buffers, such as
`*Messages*', that have names like `*...*', and I've
long done so.  For me, they are "special-display"
buffers.

I've never had any idea that one should not be able
to split a dedicated window.  And occasionally I've
done so.

To me, a window being dedicated is only about the
association between a window and a buffer.  It's
not about also not being able to split or delete
the window.

I believe that's what the doc says too, and that's
always been the behavior.  Is there a good reason
why that should change?

I don't want to join a discussion about the question.
I just wanted to mention my experience and raise the
question, in case it hasn't been raised explicitly
in the thread.

If I had my druthers, I think I'd prefer that the
behavior remain as it's been, and if someone needs
something like a dedicated window that also cannot
be split then we create something else for that.

I know that Emacs has already created a second kind
of dedication - weakly dedicated.  Perhaps a third
kind should be created to handle the (apparent?)
use of case of strongly dedicated + unsplittable.
Just a thought, in passing.

The alternative of having a global variable whose
current value determines whether _all_ dedicated
windows are (currently) splittable, which is what I
guess Richard suggested, seems worse to me.  But I
haven't followed the discussion, so I don't really
appreciate the need, i.e., use case.

I'll butt out now.