Hi, Quoting Raphael Hertzog (2017-06-26 20:22:01) > On Mon, 26 Jun 2017, Johannes Schauer wrote: > > > Because when I wrote that script, I was probably not aware that > > > --apt-distupgrade was the default. > > > > so does the existence of that option help your use-case somehow? > > I modified Kali's build script to pass it, yes. But my request remains. > > > > I understand this. But you can react differently to the failure in the > > > various places where you are calling "$resolver->distupgrade". > > > > What do you mean exactly here? > > I mean that you can turn it into a warning in Sbuild/Build.pm and an > error into bin/sbuild-update. > > > So how do you want to solve the issue now for kali? By just passing > > --no-apt-distupgrade and upgrading the chroots manually from time to time > > like > > the Debian and Ubuntu buildds do it? > > That's what I will do, yes. But I believe that this should be avoidable > and letting sbuild dist-upgrade the chroot should have been enough and should > not have the current downsides.
If you agree, then I will see this bugreport as a feature request to implement the following: Let APT_DISTUPGRADE be a tristate option: 0, 1, 2 In addition to --apt-distupgrade and --no-apt-distupgrade add --maybe-apt-distupgrade. In case of --apt-distupgrade, fail if sbuild removes build-essential. In case of --maybe-apt-distupgrade, allow "partial" distupgrades by putting build-essential on hold before distupgrading. Please signal your agreement to this plan by lowering the severity of this bug to wishlist. Thanks! cheers, josch
signature.asc
Description: signature
