>> foo.c: foo
>> 	fsm < foo > foo.c

>> Default rules are all very well in the absence of explicit
>> instructions to the contrary, but when (as here) there _are_
>> explicit instructions, I maintain they should be obeyed.

> The make source code has not changed in that respect, but the
> ``shuttle'' rule from the .c to .NULL suffix was added in the default
> rules.

Ah, okay, so that's the critical change.

> I agree with you that explicit instructions should be obeyed, but
> even if we fix it to work in the NetBSD make, it will not work
> anywhere else.

It'll work everywhere I need it to, if necessary by my carrying the
netbsd make with me. :-)  (I also am a strong believer in "if it's
broken, make it obvious that it's broken", because brokenness that
isn't obvious and blatant usually never gets fixed.  Anywhere I find a
make that misbehaves on such constructs in the Makefile, I report it to
whatever bug channel is available.)

> I don't think that it is very easy to fix either.

It seems conceptually easy to me to fix: when a loop occurs that
involves both explicit transformations and transformations derived from
suffix rules, disable the suffix rules for the duration.

Since it seems to be either diking out the suffix rule or doing
something of the sort, I suppose I'll try to find the time to implement
this and see how it works....

					der Mouse

			    mouse@collatz.mcrcim.mcgill.edu

------------------------------------------------------------------------------