issues in current heimdal

[Derrick J Brashear <shadow@dementia.org>]

1) if you configure krb4 support and you have a previous heimdal or
something else that installs roken at the same path (/usr/local/include)
you lose. this is because:

gcc  -DHAVE_CONFIG_H -I. -I. -I../include -I../include
-I/usr/local/include -I./../lib/krb5   -Wall -Wmissing-prototypes
-Wpointer-arith -Wbad-function-cast -Wmissing-declarations
-Wnested-externs -I../lib/roken -I../lib/roken -O -c kadmind.c
happens because of:
INCLUDES = -I$(top_builddir)/include $(INCLUDE_readline) $(INCLUDE_krb4)
-I$(srcdir)/../lib/krb5
in kadmin/Makefile
(from Makefile.am's
INCLUDES += $(INCLUDE_readline) $(INCLUDE_krb4) -I$(srcdir)/../lib/krb5
)

the generated COMPILE rule ends up e.g. 
COMPILE = $(CC) $(DEFS) $(INCLUDES) $(AM_CPPFLAGS) $(CPPFLAGS)
$(AM_CFLAGS) $(CFLAGS)

and the include of ../lib/roken comes from AM_CFLAGS

I'm not sure how to get the COMPILE command generated to be written
differently.

2) the libsl.a gets libeditline.a from the build problem still happens,
because 
libsl_la_LIBADD = @LIB_readline@
gets expanded literally to:
$(top_builddir)/lib/editline/libeditline.a $(LIB_tgetent)

i'm using cvs snapshots of automake, autoconf, and libtool when generating
the relevant autocrap and building. perhaps i need slightly newer
snapshots of these. this one isn't a big deal anyhow.

3) there's still the order-of-db_185.h-versus-db.h-inclusion issue in
hdb_locl.h but i can't give the exact details without undoing the patch
and doing a broken build. i will do so if i have time after doing the
other stuff i need to do.

-D