On Thu, 2013-10-10 at 10:00 +0100, Richard Purdie wrote: > On Tue, 2013-10-08 at 12:47 -0500, Richard Tollerton wrote: > > The biggest thing I can think of is gcc-native. My understanding of gcc > > documentation is that it's recommended to build a cross-compiler with a > > bootstrap, or at least the same version of compiler, and we're not doing > > that. (Right?) > > We don't build a gcc-native, no. Equally, we've not run in many issues > due to that so its not been something we've thought necessary.
I think there might be some difficulty with building a Java-enabled cross compiler if you don't have a matching host copy of ecj.jar on hand, but we currently force Java off by default anyway and as far as I know nobody is using it. If we did want to start supporting Java as a first class cross compiler target then it might be necessary to build some host-side gcc bits. Other than that, you're right, there is no particularly rational/compelling reason to build a native gcc before compiling the cross compiler. In theory it would be a good thing for folks who are building on systems where the installed native compiler is especially old or lame (because your cross compiler would run faster if you built a decent native compiler first and then used that to build the cross one) but, in practice, it would be impossible to run OE on such a system anyway without manually upgrading a bunch of tools and anybody who has gone to that trouble would almost certainly have installed gcc already. p. _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
