Hi, Constantine Kousoulos, le Mon 17 Jul 2006 14:35:51 +0300, a écrit : > However, since the linux kernel doesn't have a stable interface, it > is very likely that if we would like to update gnumach's drivers in > the near future, we would need to rewrite all over again the glue > code. In that way, writing glue code is not very reusable. I could try > writing gnumach native drivers, although i'm not sure this is a better > approach.
Writing glue code for getting a bunch of drivers is less work for more result than writing native drivers. Yes, driver interfaces change, but not that frequently. The problem here is just that nobody took the time to update the glue code, probably because this needs both gnumach and linux knowledge. Another solution is Xen. Porting GNU mach to Xen and writing Xen native drivers shouldn't be too hard, and should be relatively stable. Of course, this doesn't permit to run GNU mach natively on machines, but it permits a much broader support for people who might want to try the Hurd (note that more and more Linux distributions include Xen packages). Samuel _______________________________________________ Bug-hurd mailing list Bug-hurd@gnu.org http://lists.gnu.org/mailman/listinfo/bug-hurd
