ATI video comes out of the closet

[Bruno Wolff III]

On Sat, Sep 08, 2007 at 17:30:10 -0500,
  Les Mikesell <lesmikesell at gmail.com> wrote:
> 
> Is there some reason to think this can't be done within driver modules 
> having a stable interface to the kernel?  In spite of the fact that 

There are costs to maintaining legacy interfaces as well as benefits.
Currently it seems the thinking is that the benefits maintaining legacy
interfaces are not that high or are external, so the interfaces are changed 
when it is felt it would improve the kernel.

> other popular OS's can do that?  Personally, I think that working 

I don't think Windows has the rate of change that Linux does. By the time
a version is released, development of that version has pretty much ended.

> drivers for most popular hardware would have been provided by the 
> vendors for Linux ages ago if binaries were acceptable and could be 
> expected to work for several years unchanged.  Apparently, that doesn't 
> suit someone's politics.

Or needs. There are real problems with binary drivers. You are relying on
the manufacturer to fix bugs, which they don't have a lot of incentive
to do for old hardware and it adds another party you trust when using a
system.

Fedora is probably the worst distro with regards to binary drivers with its
emphasis on freedom and fast change. Binary kernel modules work against both
those goals.

> >Also see the new EPEL repository, where Fedora developers build RHEL
> >versions of Fedora packages.  And of course, several independent repos,
> >such as ATRPMs and RPMforge.
> 
> This is a step in the right direction, but why back into it piecemeal? 
> Build a distro that installs that way in the first place.

You are free to go do it. You can probably even find some people to help
without looking too hard.