On Thu, Aug 17, 2000 at 04:10:53PM +0100, Philip Hands wrote: > It's almost impossible to remember all the little things that might go > wrong as well, so encapsulating that knowledge in a regression test > suit is definitely the way to go.
In which vein, it might be helpful to have test machines around. Given enough warning, I'd normally be able to lend Phil a sparc for the duration of the CD test cycle, for example. Or it's possible that next year I'll have a sparc on a fast connection; one of the two. I suggest that before next time we go through this rigmarole we try to isolate definite testing volunteers for each architecture (for slink, for example, Steve not only burnt some sparc CDs for me to test, but gave me a SCSI CD to test them with...). Jules
