Joey Hess <jo...@debian.org> writes: > Adam Borowski wrote: >> Could you please mention which ones do not? And if so, how are they >> relevant/are they fixable? > > As one of the maintainers of debootstrap, I am perhaps more aware than > some how broadly it's used. Ok.. > > They use it on Android (41,600 hits including > http://evilzone.org/android/debian-on-android/) > They use it on Nokia (96,600 hits) > They use it on Nook (14,000 hits) > They use it on headless old Red Hat systems in a datacenter somewhere > They use it on Debian oldstable systems, where xz-utils is not even packaged. > They use it on absolutely modern peices of unusual kit that ship with some > crufty busybox binary (no source naturally) from far up the supplier > chain, that was built well before xz support entered busybox in 2010.
How are they relevant? Where do they download and unpack udebs? Where is busybox used to unpack debs? >> Special-casing base packages would be a lot of complexity, let's avoid that >> if possible -- but still preferred to letting gzip stay. > > Base packages can be identified at build time by their priority. > if ($priority ne 'important' && $priority ne 'required') { > } > > Although I do think that rebuilding the entire archive at this point in > the release process is probably going to result in a lot of .. > complexity. For one, d-i relies on being able to unpack firmware .debs > The code that does this doesn't support data.tar.xz. There are probably > plenty more problems where that came from. Can someone set the default to xz and recompile all of Debian or at least base and create a repository from that for install tests? MfG Goswin -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/87zk98pa11.fsf@frosties.localnet