Re: bug#736: C-l interacts badly with scroll-conservatively

[Chong Yidong <cyd@stupidchicken.com>]

"Drew Adams" <drew.adams@oracle.com> writes:

>> emacs -q
>> M-: (setq scroll-conservatively 100) RET
>> C-h n
>> C-v
>> C-l
>> C-l
>> C-l
>> 
>> The repeated C-l's don't move to the top or bottom of the 
>> window, unlike the usual behavior (new to Emacs 23) of
>> recenter-top-bottom.
>
> Maybe I'm misunderstanding the bug report, but by design
> `scroll-conservatively' modifies the behavior of
> `recenter-top-bottom'. As the doc string says:
>
>   "Top and bottom destinations are actually
>    `scroll-conservatively' lines from true window top
>    and bottom."

That's true; I didn't see that.

But why does recenter-top-bottom use scroll-conservatively?  AFAICT,
people often use a large nonzero number for scroll-conservatively so
that when point is at the bottom of the window, C-n scrolls down by one
and keeps the cursor at the bottom; and similarly, if point is at the
top of the window, C-p keeps the cursor at the top.  It doesn't make
sense to make scroll-conservatively affect C-l.

I think we should use scroll-margin for this purpose, rather than
scroll-conservatively.

What do people think?