Date: Wed, 6 Oct 1999 12:01:02 -0400 (EDT) From: Thomas Valentino Crimi <tcrimi+@andrew.cmu.edu> To: freebsd-alpha@freebsd.org Subject: New Signals - breaks Alpha makeworld Message-ID: <4ryr8y_00UwG0PvkY0@andrew.cmu.edu>
next in thread | raw e-mail | index | archive | help
There is a slight #include problem with the Alpha header files, it all
boils down to u_int* being used used /sys/alpha/include/alpha_cpu.h In
building yacc..
cc -O -pipe -c /usr/src/usr.bin/yacc/main.c
In file included from /usr/include/machine/frame.h:34,
from /usr/include/machine/ucontext.h:34,
from /usr/include/sys/ucontext.h:34,
from /usr/include/sys/signal.h:238,
from /usr/include/signal.h:44,
from /usr/src/usr.bin/yacc/main.c:52:
/usr/include/machine/alpha_cpu.h:299: syntax error before `alpha_ldbu'
Now, shoudl this be fixed with an #include <sys/types.h> as being a
necessary pre-requisite to include signals.h, with that we're placing a
requirement on programmers that they'll never catch when testing on
i386, meaning running into this more than once (already more than a few
programs will break compiling, yacc is merely the first in the make
world chain.
I've currently included sys/types.h in alpha_cpu.h, which is hackish
but at least gets the world built. I'm now doing the buildworld to see
what breaks, but I can't see it's clear what the optimial solution is.
I would say to put in the extra #include in alpha_cpu or possibly higher
up in the chain, but requiring it in all applications probably means
we're going to be doing a few bouts of patching.
The kernel built without a hitch.
Tom
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4ryr8y_00UwG0PvkY0>