On Sun, 2007-01-21 at 04:20 -0800, Steve Langasek wrote: > On Wed, Jan 17, 2007 at 02:55:30PM -0800, Thomas Bushnell BSG wrote: > > On Wed, 2007-01-17 at 13:47 -0800, Steve Langasek wrote: > > > > Of course, I have a conflict of interest here as an alpha porter, so > > > ultimately I'll defer to Andi if he thinks it's become a problem; but in > > > general we're unlikely to cut a port from the release at this late stage > > > without some pretty serious, long-term problems. > > > What if a security build needs to be made in these ten days? > > Then we're in the same situation we've been in any number of times before -- > a bad one, but not one that's been deemed a reason to exclude an > architecture from release per se. This is the first time we have the "must have two buildds" rule. This rule had a purpose behind it, which was *exactly* the situation we are now in. We are experiencing exactly the kind of problems for which this rule was created. This is not some kind of strange and exceptional case which is unlike the problems the rule was created to avoid; this is exactly the case the rule was created to avoid. So, why do we have the rule? If the rule exists so that it can be ignored in the central case the rule was about, then let's get rid of the rule. Thomas
Attachment:
signature.asc
Description: This is a digitally signed message part