> If it is only for DOS... mmm... so much work only for DOS?... OK, I will
>not going to implement anything of this anyway...
We only have to take it as far as we feel inclined. Writing the initial
driver framework will I think be quite tricky, but once that is working
it will just be a matter of filling in the blanks to support particular
chipsets, and that can be done if/when anyone feels inclined. It doesn't
have to be a 100% serious project, just a series of steps along a path
that may someday end up in a really cool place...
> The drivers must be GPL if them are being supposed to be used in Linux
>too, but this will not make fall our programs using those drivers under GPL.
This is a good point: what should be the licensing for these drivers? I
don't think GPL itself is good enough, because it is essential that
people be able to distribute the binary drivers along with their
programs, and not have to include full source as well. But basically I
like the GPL idea for something like this: I don't think a hardware
driver needs the same flexibility of usage that is important for
something like Allegro itself, and it would be a good thing if we can
force any modifications people make to be freely available like the
Perhaps it should be like djgpp itself: GPL, but with an added rider
that the binaries may be freely distributed as long as they are not
modified? That way anybody can include the drivers with their programs,
but if they want to extend or modify our work, their improvements must
remain available under the GPL.
> I think that all free software developers have to think of a way to
>implement some things (like this video drivers) in a plataform-independent
>way. So, one can do a video driver and then, it can help to many and not a
This is one reason why I really like VBE/AF. It isn't 100% perfect in
that area, but the flaws are few and far between, and SciTech claim that
the forthcoming Inertia spec will fix most of these small problems.
> X Windows uses and have a lot of accelerated drivers, maybe it
>is possible to take some ideas from them and do the drivers.
Those drivers are a good source of ideas, but the actual code is
useless: very closely tied in to the inner workings of their system.
Plus it is GPL, so we can't use it with Allegro unless we include full
driver source with every program.
> I encourage the people to help in this, but I think it has to be good
>planed and designed and consulted with other free developers and not just
>in the Allegro mailing list...
A noble ideal, but in practice I think this is a good way for a project
to go nowhere :-) The more people you get involved in planning
something, the bigger the ideas get, and the longer it takes before you
can actually start working. I want these drivers to be a reality next
week, or next month, not next year...
The opposite extreme is of course just as bad, where you design just a
little stop-gap solution for one specific problem, which is then no use
to anyone else. Which is why I think it will be a good idea to use some
external standard. If everyone just concentrates on dealing with the
particular small problems that are closest to them, but tries to do so
in the most generic and compatible way they can, it is usually possible
to slot all these different efforts together at a later date. This way
we can just sit down and start writing /AF drivers without any lengthy
planning, and who knows? Maybe a couple of years down the road some
XFree86 developers will be saying "hey, those djgpp people have a lot of
nice drivers for graphics hardware, maybe we can use some of that stuff
Shawn Hargreaves - shawn@... - http://www.talula.demon.co.uk/
"Pigs use it for a tambourine" - Frank Zappa