Wouter Verhelst wrote: > I very well remember that mail. The key word in Clint's mail that > prompted my reply was 'current' (problems). d-i has been ported to all > our architectures now, which required a lot of work, and doing a d-i > release still requires quite some work to synchronise that among all our > architectures, I presume. However, the point is that having many > architectures does not effectively stall our release today.
That mail listed a number of security holes that are not fixed in testing due to issues with one or more architecutes. Half of the packages on that list are still unfixed, and more have joined them. That mail said that d-i rc3 has been delayed for several months because of the need to first get updated kernel packages on all architectures. The last one of those finally got into testing today, for an overall delay about 2.5 months. We were thrown a week off our expected release date by problems with one architecture. I continue to spend time each day checking d-i install failures in my automated install lab, and spend time each week with the unavoidable sysadmin duties that come from maintaining a collection of hererogeneous hardware. From my POV, everything in that mail still applies as examples of real ways that supporting a great many architectures are impacting the release. BTW, apparently one of the key problems with getting the required security autobuilder setup, which is one of the main things that has delayed the sarge release for 6 months or more, is that all the connections from the buildds have been near to overwhelming newraff, and adding new queues would multiply that. It's another example of the way that small and seemingly insignificant effects can turn into real problems when the number of architectures is scaled up to as many as we're trying to support now. -- see shy jo
signature.asc
Description: Digital signature
