Re: 3.80: Unix<->NT rule inference differ

[Cyrus Klug]

Thank you Eli. Having used this -p switch
on NT, I see that there are built-in rules on
for ".o" files, but no built-ins for
".obj" files, thus explaining the discrepency.

Is there some switch for generating built-in .obj
rules when building gnumake on NT? I could adapt the
gnumake source code and rebuild, but if there is an
easier way I would appreciate knowing it.

Bye,
CK

 --- Eli Zaretskii <address@hidden> wrote: > >
Date: Mon, 3 Feb 2003 14:36:42 +0000 (GMT)
> > From: =?iso-8859-1?q?Cyrus=20Klug?=
> <address@hidden>
> > 
> > CC=cl
> > MAKELIB= lib.exe /OUT:$@
> > 
> > 
> > objFiles1= foo.obj
> > objFiles2= foo.obj bar.obj
> > 
> > lib: foo.lib bar.lib
> > 
> > foo.lib:  $(objFiles1)
> >    $(MAKELIB) $?
> > 
> > bar.lib:  $(objFiles2)
> >    $(MAKELIB) $?
> > 
> > foo.obj: foo.c
> >    $(CC) -c $< -o $@
> > 
> > bar.obj: bar.c
> > 
> > ----------------------snip------------------
> > 
> > Here is the issue: under NT, gnumake 3.80
> determines that the rule
> > for bar.obj has no command line. Contrast Unix,
> with essentially the
> > same Makefile and gnumake release, the command
> line $(CC) -c $< -o
> > $@ is inferred for target bar.obj based on the
> rule for
> > foo.obj.
> 
> Perhaps the Windows port of Make has some built-in
> rule for *.obj
> files, whereas the Unix version doesn't.  Run "make
> -p" on each
> platform and see whether the differences give a
clue. 

__________________________________________________
Do You Yahoo!?
Everything you'll ever need on one web page
from News and Sport to Email and Music Charts
http://uk.my.yahoo.com