Re: gmake: how to avoid $$@ usage?

[Paul D. Smith]

%% Ernest <ecrvichNOSPAM@NOSPAMfootbag.org> writes:

  e> The GNU make manual, section 10.5.3, mentions the use of the
  e> SysV-style automatic variables $$@, $$(@D), and $$(@F), and that
  e> these "bizarre" uses are not necessary ("there are other ways to
  e> accomplish the same results").  How?

  e> For example, I need to do some stuff like what's shown below, where
  e> the prereq names are determined by a semi-complex manipulation of
  e> the target name (it's actually much more complex than this
  e> example).  This doesn't work, though.

Right, because you're trying to wrap $$@ in a function; that won't work
because while the expansion of $$@ is delayed, the expansion of the
FUNCTIONS are still immediate (before $$@ is expanded).

Without seeing a REAL example of what you're trying to do it's difficult
to determine how much power you need.  Typically there are three ways
you work around the need for $$@, etc.:

 1) Static pattern rules, for straightforward situations:

      target_types = build comp clobber foo bar biz baz boz
      $(target_types:%=%_all) : %_all : %

 2) Auto-re-exec: you can include a makefile, then provide a rule to
    generate that makefile where the rule constructs the complex data.
    Since this happens in a shell script you have all the power of the
    shell here:

      include all_targets.mk
      all_targets.mk: Makefile
                @echo <stuff> >> $@

 4) If you have GNU make 3.80, you can use the $(eval ...) function.

-- 
-------------------------------------------------------------------------------
 Paul D. Smith <psmith@gnu.org>          Find some GNU make tips at:
 http://www.gnu.org                      http://make.paulandlesley.org
 "Please remain calm...I may be mad, but I am a professional." --Mad Scientist