[comp.lang.c] saber c and gnu make

emv@ox.com (Ed Vielmetti) (04/03/91)

apologies for discussion of commercial software on a gnu list.

environment: gnu make 3.57 
	     saber c 3.01 on a dec 3100, evaluation version expires
	       in 8 days.

problem:  gnu makefiles and saber c differ on comment handling.

in particular, gnu make treats a line starting a
pound sign as a comment, whereas saber c expects productions
beginning with a pound sign to be sent to the shell.  e.g.

saber_target:	$(STUFF)
	#sabercommand $(STUFF)

gnu make will treat the line as a comment, no-op out the line, and
make $(STUFF) with the default rules.  saber on the other hand expects
to be handed "sh #sabercommand foo.c bar.c bletch.c", realize that the # is
for it, and do its load.

workaround: ifdef all of the saber invocations in gnu make, use a
different comment leader, and do all of the things necessary in saber
to have it invoke gnu make with the proper invocation.

ifdef SABER_MAKE
saber_target:		$(STUFF)
	%sabercommand $(STUFF)
endif

problem:  saber c gets confused by gnu make rcsfile handling.

not precisely sure what's happening here yet, but if e.g. some piece
of things is out of date or missing, gnu make will check it out.  some
aspect of this gets echoed back to saber c, and saber tries to do some
shell commands.  (?!)  not entirely certain what particular
combination of events will bring this about, or how to deal with it.

workaround: none yet.

thanks for any help.  a copy has been sent to support@saber.com.
followups to comp.lang.c, though i'd hope the gnu make folks would
follow what's going on.

-- 
Edward Vielmetti, OTA Limited Partnership.  (emv@ox.com)

"With all of the attention and publicity focused on gigabit networks,
not much notice has been given to small and largely unfunded research
efforts which are studying innovative approaches for dealing with
technical issues within the constraints of economic science."  
							RFC 1216