Robert P. J. Day wrote: >> It works like this: When a build system target is called, the OPENWRT_BUILD >> variable is unset and that means, the preparatory work is to be done first. >> Package directories are scanned, prerequisites are checked, the .config is >> checked for if the target requests it, etc. >> At this point, the variable OPENWRT_BUILD is exported and when a real >> target is called, @$(MAKE) $@ will call the Makefile again, but because of >> the different set of variables, the real build system logic is included now >> and thus provides access to the real targets. >> Hope this helps. > > sorry, i probably sounded a bit dense when i first asked that > question. i'm actually familiar with two-pass makefiles, i just > didn't understand that exact rule command: > > @$(MAKE) $@ > > since $@ does, of course, simply refer to the target of that very > rule, which seemed pointlessly recursive. but now i'm assuming that > the variable $(MAKE) expands to more than just a simple "make" > invocation -- i'm guessing it has to include a change directory of > some kind, right? *then* it would make perfect sense. No, $(MAKE) is the regular make command, but because a variable has been exported by the first pass, this particular line will invoke the second pass....
- Felix _______________________________________________ openwrt-devel mailing list openwrt-devel@lists.openwrt.org http://lists.openwrt.org/cgi-bin/mailman/listinfo/openwrt-devel
