Re: scroll-in-place

[Nix <nix@esperi.org.uk>]

On 11 Apr 2011, Eli Barzilay said:

> I had this code for a while, and it's been working well -- recently
> Nick Alcock took it and made it more publishable, including changing
> the builtin functions.

Oh, I hardly did a thing. :)

> reprovide the functions from elisp code that can decide to simply call
> the builtins.  (Another option is to do this all in C, it's really
> pretty simple, but I can't offer serious help with that...)

I could do this, but it's probably pointless: scroll-{up,down}-command
isn't in C in the first place and is not really that speed-critical (as
long as it isn't visibly slow, vide the recent fixes to stop Emacs doing
massive amounts of line-by-line searching)

> otherwise it calls the primitive versions.  In addition, there is some
> old code that sets `scroll-in-place' when it wants to avoid the
> different behavior (at least Gnus and VM) -- so the code checks that
> variable too.  (That was just the simplest way to avoid problems with
> these things.)

Even though I suggested this I'm no longer sure it's a good idea, unless. This
means we end up with a situation where setting *either* of two variables
(`scroll-in-place' to nil, or `scroll-preserve-screen-position' to non-
`in-place') turns off scroll-in-place, which feels confusing.

It may be simpler to just change Gnus and VM to bind
`scroll-preserve-screen-position' where they bind `scroll-in-place',
changing its value only iff it is `in-place'. (This is trivial: I've
already written the code and can provide it if anyone agrees that this
is the right approach.)

> (Unfortunately I can't allocate more cycles to turn it into a patch,
> but that should be very quick for people here.)

I can do that if anyone wants.

-- 
NULL && (void)