> Are you certain this happens with a clean tree and using sun4u OBJTYPE as
> well?
> Please recheck with hg tip.
sdf1 % hg identify
e39eeddcc295+ tip
sdf1 % cat config.mk
# Customize to fit your system
# paths
PREFIX = /usr/local/plan9
MANPREFIX = ${PREFIX}/share/man
VERSION = 7
#OBJTYPE = 386
#OBJTYPE = arm
#OBJTYPE = x86_64
OBJTYPE ?? ?? = sun4u
# Linux/BSD
#CFLAGS += -Wall -Wno-missing-braces -Wno-parentheses -Wno-switch -c -I.
-DPREFIX="\"${PREFIX}\""
CFLAGS += -c -I. -DPREFIX="\"${PREFIX}\""
LDFLAGS += -static
# Solaris
#CFLAGS = -fast -xtarget=ultra -D__sun__ -c -I. -DPREFIX="\"${PREFIX}\""
#LDFLAGS = -dn
# compiler
AR = ar rc
CC = cc
YACC = ../yacc/9yacc
sdf1 % make
"config.mk", line 11: Need an operator
Fatal errors encountered -- cannot continue
sdf1 % file config.mk
config.mk: UTF-8 Unicode make commands text
When I replace config.mk with an otherwise identical, ASCII format file,
compilation is able to proceed, but it still dies in the same place as
noted in my original e-mail. This has been the same since the first time
I tried to compile sun4u early last year.
> The main reason I dislike using the host yacc is simply that it will
> pretty much diverge from Plan9's yacc.
> And hence I don't trust it to produce a compatible parser in most
> cases. For ex. does it produce a proper rc?
I've been using 9base compiled with the host's yacc for about a year
now and I've never run into any bugs. Rc seems to behave as expected.
-sl
