On Thu, 2011-08-25 at 11:01 +0200, Samuel Thibault wrote: > Svante Signell, le Thu 25 Aug 2011 10:58:43 +0200, a écrit : > > On Thu, 2011-08-25 at 10:48 +0200, Samuel Thibault wrote: > > > Svante Signell, le Thu 25 Aug 2011 10:20:44 +0200, a écrit : > > > > > > 1) Where to find the specification resulting in i486-pc-gnu? > > > > > > I don't understand the question. Why would you want that ? > > > > Some packages are configured with i486-pc-gnu, > > For instance?
See Guillems reply. I see other examples too where i486-gnu is not forwarded by debian/rules: mathgl, nbd. They use dh_auto_configure and in config.log --build=i486-gnu is already set as an option to configure. I have not yet found out where it is set, though. > > > > A common denominator of $host is > > > > > > > > c) *-gnu* > > > > > > This one: don't care the actual architecture, don't care the actual > > > version of GNU. > > > > It is not unique if * means wildcard, > > Yes, but that's the best we can do. It looks like the triple CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM e.g. i686-unknown-gnu is not unique but CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM e.g. i686-unknown-hurd-gnu would. Or should the kernel be gnumach? Has this ever been proposed and/or discussed with upstream autoconf? /usr/share/misc/config.sub amd64-linux gives x86_64-pc-linux-gnu while /usr/share/misc/config.sub gives x86_64-pc-linux-gnu And a config.guess entry on kfreebsd would be: i686-unknown-kfreebsd8.1-gnu > > > > 6) Are these only local to the rules file? > > > > > > I don't understand the question. They are dpkg-architecture output, > > > that's all. > > > > Yes, and they are used for logic in the rules (Makefile) right?, and not > > exported in any way to configure*/outside ? > > When cross-compiling, it would be. See Guillems reply here too. Thanks Guillem.