Chen Jun ???? wrote: > Well Brian, the example presented by me is just a fabricated one, not > for my real world use. I use $(shell ) just to introduce a make variable > or a make function so that my question becomes concrete. My real world > case regarding this problem is quite a bit complex so I'd like not to > talk about to it directly.
Regardless, I think the same principle holds: if you need logic that considers factors that change during the execution of a recipe, then code that logic using the shell rather than make's functions. I can't think of a situation where this would not be possible; you can pass through to shell variables the contents of any makefile variables that might be required by that logic. And if all else fails there's always the universal hammer of recursively invoking a sub-make. Brian _______________________________________________ Help-make mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-make
