Re: make: dealing with targets with multiple outputs

[David Holland <dholland-tech%netbsd.org@localhost>]

On Tue, Jun 11, 2019 at 02:00:39PM -0000, Christos Zoulas wrote:
 > In article <27887.1560240087%splode.eterna.com.au@localhost>,
 > matthew green  <mrg%eterna.com.au@localhost> wrote:
 > >"Simon J. Gerraty" writes:
 > >> David Holland <dholland-tech%netbsd.org@localhost> wrote:
 > >> > No, the special target should be for the unusual case.
 > >> > 
 > >> > The usual case is:
 > >> > 
 > >> >    y.tab.c y.tab.h: foo.y
 > >> >            $(YACC) -d foo.y
 > >> 
 > >> Sorry I've been away...
 > >> 
 > >> AFAIK yacc is normally handled via suffix rules.
 > >> How is that case handled?
 > >
 > >AFAICT, yacc handling is currently poor and is (part of)
 > >what dholland is trying to propose a fix for.
 > 
 > I think we should be trying to fix this by first implementing pattern rules
 > instead of introducing new syntax.

So it seems that in gmake, while if you write

   foo.h foo.c: foo.y
           yacc -d foo.y

it does the wrong thing, if you write a pattern rule of the same form:

   %.h %.c: %.y
           yacc -d foo.y

it does the right thing. Although this which should be equivalent:

   foo.h foo.c: %.h %.c: %.y
           yacc -d foo.y

causes a syntax error.

I don't particularly approve of this. I think the behavior of the
pattern rule and the non-pattern rule with the same syntax should be
the same. Declaring pattern rules of this form is well and good (and I
wonder when gmake added it; I'm pretty sure it didn't work back in the
day) but it isn't a substitute for handling the basic form correctly.

(I also still don't like gmake-style pattern rules at all, but I
suppose the world has me substantially outvoted.)

-- 
David A. Holland
dholland%netbsd.org@localhost