On Wed, Jul 04, 2007 at 08:27:40PM +0200, Lucas Nussbaum wrote: > On 08/03/07 at 08:02 +0100, Lucas Nussbaum wrote: > > On 07/03/07 at 23:07 +0100, Christian Perrier wrote: > > > > I would really like to see that happening at the beginning of the lenny > > > > release cycle. Packages that prompt without using debconf make it > > > > unnecessary difficult to test them using piuparts. > > > > > > > > Looking at my piuparts results (testing packages in etch), most packages > > > > that prompt the user already do that through debconf, so it would not > > > > result in more than 50 or 100 bugs (and that's the worst case scenario).
> > > Hmmm, that many? > > > [..] > > > I'm still puzzled by the number of package you announce, Lucas. > > Well, that's really the worst case scenario. I would have to run > > piuparts again to get better numbers, since: > > - I'm running piuparts on etch, not sid, and packages > > in-sid-but-not-in-etch are likely to be less well maintained, so > > changes are higher that they do bad stuff with /dev/tty > > - I use a workaround (make /dev/tty a copy of /dev/null) so that most > > packages reading /dev/tty don't block during the test > > I'll try to give a better estimation soon. > Indeed, there are probably less packages to fix that I thought. > Christian, would you be interested in pushing a lenny release goal about > this? Last I knew, policy said packages *were* allowed to depend on the availability of /dev/tty during configuration, even if they're not supposed to be doing direct prompting by way of it. This seems to have been changed, but isn't mentioned in the policy upgrading-checklist? Given that this has changed, I don't personally have any objections to this proposal (and my tentative objection only had to do with the possible methodology of fiddling with /dev/tty). -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. [EMAIL PROTECTED] http://www.debian.org/ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
