On Fri, Nov 23, 2007 at 02:48:34PM +0100, andrzej zaborowski wrote: > > If that's really the case, I'm curious why it's distributed in its > > binary form rather than being built at runtime. > build-time?
Yes, my mistake :) > Perhaps because the (cross-)compiler used to build the BIOS is not > present in many distros and would be a quite inconvenient dependency. > This is for the PC bios from bochs. That's a reasonable point. > Now if all bioses shipped with qemu were built you would need some 4 > cross-toolchains. That's a good point. For us, however, the only alternative is to depend on other packages to provide these blobs (bochsbios, vgabios, openbios-sparc, etc). > Note that this code doesn't run in this form on the host Debian/Ubuntu > is on. That does not matter much. > It can well be treated as data, not code, like jpeg's shipped in > binary with a program. The difference is that a jpeg is usually the original work and there is no source to distribute that can be used to regenerate the jpeg. >> I'm also not familiar with the changes in the bios.diff. Are they >> specific to qemu or could the theoretically be applied to upstream >> bochs without any ill side effects? > There are only two tiny changes, one is generic, one is qemu specific. Ok. I'll change our bochsbios build to build a specific QEMU bios with these changes. That should make everyone happy. I sense that any response to this e-mail would be off topic for the ml (even this mail is borderline), so please follow-up privately. -- Soren Hansen Ubuntu Server Team http://www.ubuntu.com/
signature.asc
Description: Digital signature