Daniel Borgmann <[EMAIL PROTECTED]> writes: > That sounds like a good plan. Any thoughts how this should work in > detail? I mean, should the translator just be something like a "pipe" to > the video memory or should it already provide drawing functions?
One could start with saying that it should support whatever /dev/fd supports on linux or on some other system with a sane framebuffer device. I don't know the details, but it must support mmapping the actual framebuffer, and it should use either ioctl:s or some read/write operations for doing other interesting things. It should also provide some mechanism for robust coordination between different processes using the frame buffer. I think it would make sense to implement acceleration by some simple protocol, "line x1,y1,x2,y2" where you write commands to the framebuffer, but I don't know if anybody else does framebuffers that way. Perhaps plan 9? > Should it allow access from every application? The ordinary filesystem permissions on the node to which the translator is attached determines who can access it. Ideally, whoever is allowed to use the (virtual) console should be allowed to use a corresponding frame buffer. > I just had the idea to look how svgalib is realized, maybe this > would be a good starting point? Might be. You'd also need a decent spec for the graphics hardware. Regards, /Niels -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
