> >> mentioned on there are my decision to branch the Radeon driver
> >> and to contribute stuff back to the XFree86 project. How that is
> >> seen as a fork of the entire project, I'm unsure. I'll be the
> >> first to admit that I'd be completely unable to fork the entire
> >> XFree86 project on my own, and that it would be a very large
> >> amount of work without any guaranteed gains in return.
> >>
> >
> >Actually, there is a fork of XFree already - in dri.sf.net and another one
> >of ati driver only in gatos.sf.net.
> >
> >One more wouldn't hurt, but Mike - why do you want to fork Radeon driver ?
>
> Sorry to mix terminology in what some might feel is pedantical,
> however the term "branch" the Radeon driver seems more
> appropriate than fork, although it is possible I may have myself
> used the term fork somewhere. Branch is much more accurate, as I
> believe the term "fork" means something more hostile and ill
> intentioned, and "branch" means concurrant development which gets
> merged back to the trunk.
First it forks than it merges.. I am pretty sure the fork is more
noticable ;)
>
> You don't have CVS write access, however you've managed to put
> together a successful driver yourself too, which makes me a bit
> curious why GATOS has remained a separate (and rather succesful)
> project from XFree86.org for so long. I've never chatted with
> you before about it I don't believe, but I've often wondered why
> GATOS has never been integrated into XFree86 or DRI. It's useful
> as is nonetheless, but it'd be great to see such developmental
> efforts get incorporated into XFree86's main codebase so more
> users could benefit from one-driver-does-it-all so to speak.
Historically, GATOS first started with a standalone program and we had to
switch to messing with drivers when XFree86 4.x.x was released (as we
couldn't play the tricks with video driver by telling it to use less
memory than was available). Of course, there were a number of
other advantages to this as well (not the least being at least partial
compatibility with systems other than Linux).
>
> My assumption has been all along that you've had difficulty
> getting others to look into merging your changes, and decided it
> was easier to maintain your own codebase entirely than to twist
> people's arms to look at your code and provide feedback. I guess
> now would be a good time for me to just ask. ;o)
>
> So.. any chance of GATOS getting into XFree86? ;o)
This question gets asked very often. The answer is that part of GATOS is
already there. The part that isn't is TV-in and hardware Xv for mach64
cards. And it was submitted but it takes quite a while to get in. The
major reason for this is that ATI driver maintainer and me have different
priorities:
My impression is that ATI driver maintainer (Marc please correct me) is
primarily interested in
* clean and working code that is easy to maintain
which, IMO, is a worthy goal.
My priorities are:
* robust (non-crashing) driver that supports new features
* experimentation with new approaches to programming and new features
These two do imply "clean and easy to maintain" but in a significantly
different way.
Also both of us are quite short on spare time.
Lastly there is one more issue with GATOS radeon code - to make video
capture possible we need PCI DMA. Unfortunately, due to imprecision in
ATI documentation the early radeon DRM driver was designed in such a way that
it conflicts with PCI DMA into lower 64 meg of RAM (or whatever the size
of AGP aperture is). The changes required will break compatibility with
earlier modules, something that DRI maintainers are not willing to accept.
(and I do not know DRI part of things well enough to make a workaround..
though this might change with 4.3.0 code - the DRI code is a lot cleaner)
On the positive side GATOS code does present a valuable contribution in
the form of code that is known to work even if the maintainer decided to
rewrite the whole of it. Some of the hardware knowledge was not so easy to
come by (and is not well described in ATI docs).
best
Vladimir Dergachev
>
> I know ATI users would absolutely love this functionality, and I
> would too.
>
> Any comments, feedback, appreciated.
>
> Thanks,
> TTYL
>
> --
> Mike A. Harris
>
>
> _______________________________________________
> Devel mailing list
> [EMAIL PROTECTED]
> http://XFree86.Org/mailman/listinfo/devel
>
_______________________________________________
Devel mailing list
[EMAIL PROTECTED]
http://XFree86.Org/mailman/listinfo/devel
