Hi, Quoting Andrey Rahmatullin (2014-09-12 18:14:55) > On Fri, Sep 12, 2014 at 03:11:08PM +0200, Johannes Schauer wrote: > > The common fallacy is that the "foo" in "any-foo" is the name of a Debian > > architecture while in fact it is the name of the CPU which is mapped to one > > or > > more Debian architectures by /usr/share/dpkg/triplettable > Indeed, maybe we need to spell this more explicitly in the Policy?
Either that (footnote 99 [1] gives some more explanation) or change the meaning of architecture wildcards into something more intuitive like os-debianarch as it is probably understood by most developers because the existing values for "cpu" happen to be (or used to be) debian architecture names. Another relevant bug about this topic is #694630. I think this and the apt bug #748936 give a good overview about how the current system works, why it exists and how alternatives could/should look like. Lintian successfully warns when things like "any-armel" are used in an attempt to let the second part be a debian architecture instead of a cpu name, so we don't have many of these in the archive. This is the relevant thread where I tried to identify packages with invalid architecture wildcards: http://lists.debian.org/20140523140000.20924.60609@hoothoot cheers, josch [1] https://www.debian.org/doc/debian-policy/footnotes.html#f99 -- To UNSUBSCRIBE, email to debian-devel-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140912225415.3685.31745@hoothoot