On Saturday 09 February 2008, Jérémy Bobbio wrote: > Chris' proposal would gives us the same situation as ppp-udeb. > There would be some code depending highly on how d-i is working as a > whole (its components might change along the development) which would > not be easily modifiable by the d-i team. I disagree. Splitting this in two would give us _two_ new extremely small udebs instead of one. The functionality in the postinst script of cpuburn is trivial. > The removal of "ifconfig" made me aware of this situation for ppp-udeb, > and I think it would be better if we could avoid it in the future. It > would also enable Chris' to maintain the installer component without > having to bother Aurélien everytime he would like to improve the stress > test without touching cpuburn. That's only a very limited disadvantage as the cpuburn udeb is not likely to require much maintenance. The only moment such a split would make sense to me if a stress-test udeb would become a kind of wrapper around multiple hardware test tools, but IMO the split could always be done at that time. Cheers, FJP P.S. I also don't really see the problem with ppp-udeb and in that case the udeb-specific stuff is even interwoven a _lot_ tighter with the ppp stuff itself which means that it would be likely that you'd need coordinated changes in two places.
Attachment:
signature.asc
Description: This is a digitally signed message part.