On Sun, Feb 15, 2004 at 06:12:14PM +0100, Thomas Viehmann wrote: > Steve Langasek wrote: > > ... your argument against the use of debconf is that questions at low > > priority will sometimes be asked? Um, anybody who runs debconf at low > > priority gets exactly what they ask for. Trying to create a separate > > priority with the explicit intention that the questions will *never* be > > asked would inappropriately overload debconf. > My argument is that debconf priority low is defined to be used in user > interaction. If, in the opinion of the maintainer, there is a resonable > default, policy instructs him hat the interaction should be avoided. > If CDDs want to use debconf in these cases, priority low is not appropriate. There is no reason why users should not be *given the option* of adjusting these settings through debconf interaction, if the code is already there to use debconf. Policy describes perfectly well the user's package configuration experience when running debconf at high priority (assuming no per-package bugs). Everything that can be configured at medium or below is icing anyway. > (For your reference of the NTP howto: The howto & faq on ntp.org > contains dead links in the server selection section. My guess that it > was written before there was pool.ntp.org.) Actually, this is fundamental to the NTP protocol; you get a more accurate clock if you can cross-reference the timestamps between multiple servers. Pointing to pool.ntp.org may give you a consistently accessible *single* server, but you don't get the assurance of having multiple referents. (You'll just have a single referent, which may be a different server each time you poll it.) -- Steve Langasek postmodern programmer
Attachment:
signature.asc
Description: Digital signature