Re: console update

[Marcus Brinkmann]

On Tue, Jun 18, 2002 at 07:15:23AM +0200, Stephan Trebels wrote:
> Why you didn't create separate matrixes for attr, 1st char, second
> char, or for simplicity mandatory vs. optional matrixes.

Several reasons:
Because then if you change a character, you need to send two change
notifications if you don't want to add the semantics to the protocol that
only change notifications for the first "plane" are sent.  This multiplies
the RPC overhead.
Because it distributes the data for one cell over several memory pages, and
updating a single cell could result into several page faults.
Because it probably would make resizing the screen/scrollback buffer more
complex (not sure about this, as this isn't implemented ;)

> I agree, it'd look weird,

Not really, I considered it.  Multiple planes like this are a common in
pixel devices.

> but it'd remove the interleave problem
> immediately.  And you can support a huge cellsize without ever
> comitting the memory unless it's used.

This is true.  You can even add another plane with full backwards
compatibility.  I didn't evaluate this possibility under the issues raised
here, so maybe it's worth to reconsider it.

Thanks,
Marcus

-- 
`Rhubarb is no Egyptian god.' Debian http://www.debian.org brinkmd@debian.org
Marcus Brinkmann              GNU    http://www.gnu.org    marcus@gnu.org
Marcus.Brinkmann@ruhr-uni-bochum.de
http://www.marcus-brinkmann.de