Re: Double colon rules with phony prerequisites

[Kaz Kylheku (gmake)]

On 2021-05-06 06:15, Benjamin Fogle wrote:
> Oh, I see. I misinterpreted the docs: Phony targets shouldn't be
> _direct_ dependencies of real files, but indirect dependencies work as
> expected. I'm glad that this wasn't some obscure behavior I was
> relying on.

The "needs update" dependency relationship in Make
is not transitive; there is no indirection in it.

If we have A -> B -> C, it is not the case that A -> C,
if the -> symbol means "needs update due to".

A's rule is triggered if A is missing or older than B.
That's it; there is no concern for B. A doesn't care
how B comes into existence. B's dependency is not A's
dependency.  If B is a phony target, then "needs update
due to" is always true.

Suppose C is newer than B, but the rule for updating B
is not simply time-stamp-based: the rule notices that
the content of C has not changed and avoids touching B.
In that case, A will not update.

If we have another symbol => which means "comes before
in a goal chain", then A -> B -> C implies A => B => C,
and that has transitivity A => C.