> > The part of stage2 that loops through the whole package set could > > instead loop through all of the script names in that directory. (If > > build order is an issue for future archs, we could use sysvinit-style > > order numbers: /stage2/buildscripts.d/45-gawk). > > I'm for using a numbered build script approach with sysv ordering. One > of us could hack this up, but since it's it's DJ's script let me copy > him and give him an opportunity to figure out how he wants this done. I'm a big fan of "if you're willing to do the work, go for it!" :-) Don't feel you have to get my permission if you want to do something better, but PLEASE keep this in mind: It's stage *TWO*. You also have to update stage ONE since it builds the stage2 rootfs, so must copy everything into it that stage two needs. The whole point of these scripts is to start FROM SCRATCH so please don't break that! Absolutely everything in the stage2 rootfs MUST come from stage1, including the build scripts and source trees. I don't think stage1 needs a more elaborate design, and in fact it might be quite hard to coordinate the kernel/glibc/gcc juggling with separate scripts, but if you want to do stage1 also go ahead. Also, if anyone wants to move my bootstrap.git to fedorahosted, I'm OK with that too. I suspect we're well past the point where I need to own it - once we're into the repetetive package building cycle, you folks know more about the process than I do ;-) _______________________________________________ arm mailing list arm@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/arm
