Re: implicit rules

[Noel Yap]

Paul D. Smith wrote:

> %% Noel Yap <address@hidden> writes:
>
>   ny> I see.  You wouldn't happen to have a patch that'll allow make to
>   ny> use the implicit rule for the former example, would you?
>
> Don't know which example you're talking about when you refer to "the
> former example", but this:

I think I meant "latter" (ie "%.mk: %.m").  How can I get make to use the "%.mk: 
%.m" implicit rule to build aoeu/aoeu.mk?

>   >>> %.mk: %.mk
>   >>>       cp $(<) $(@)
>
> Just doesn't make sense.  It's like saying:
>
>     foo.o : foo.o
>
> How can a target depend on itself as a prerequisite?  That requirement
> cannot be satisfied given our current understanding of the laws of
> space/time :).

The reason one might want something like "%.mk: %.mk" (rather than "foo.o: foo.o" which is different) is because of how implicit rule matching treats directories.  For example, except for the spurious "Cyclic dependency" errors, in the end, make did 
exactly what I intended.

IIUC, make can use "%.mk: %.mk" to match "aoeu/aoeu.mk: aoeu.mk".  OTOH, IIUC, make should use 
"%.mk: %.mk" to match "aoeu/aoeu.mk: aoeu.m" so I'm clearly missing something here.

> You _COULD_, however, do this:
>
>     aoeu/%.mk : %.mk
>             cp $< $@
>
> and then all would be right with the world.  Or at least make would be
> happy.

This rule is now not as generic as before; one would have to write a new rule 
depending on where the target will be created.

Noel