Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 09 Apr 2002 14:46:36 -0600 (MDT)
From:      "M. Warner Losh" <imp@village.org>
To:        obrien@FreeBSD.org
Cc:        bright@mu.org, jhb@FreeBSD.org, kris@obsecurity.org, cvs-all@FreeBSD.org, cvs-committers@FreeBSD.org
Subject:   Re: cvs commit: src/sys/sys ioccom.h
Message-ID:  <20020409.144636.59877321.imp@village.org>
In-Reply-To: <20020409131837.D35868@dragon.nuxi.com>
References:  <20020408.222429.22925353.imp@village.org> <20020409064242.GJ93885@elvis.mu.org> <20020409131837.D35868@dragon.nuxi.com>

next in thread | previous in thread | raw e-mail | index | archive | help
In message: <20020409131837.D35868@dragon.nuxi.com>
            "David O'Brien" <obrien@FreeBSD.org> writes:
: While fixing screen is a good thing to do.  Nobody seems to be able to
: determine just WHICH commit broke building it.  I think we should determine
: that before "fixing" it.

The problem was tracked down.  The config programs are doing bogus
things to make sure that the prototype is defined.  That's why things
aren't working.  They are using a regular expression that doesn't get
all possible legal prototypes.  I sent a rather extensive message to
cvs-commmitters with the details last night.

I don't know about people, but I am able to determine which fix broke
it, why and what {screen,ddd,vim,hylafax} are doing that is bogus.
They all seem to have code that was copied in some way from some
original source (well, the ddd problem is due to bugs in autogen):

sed < $srcdir/osdef.h.in -n -e '/^extern/s@.*[)*        ][)*    ]*\([^ *]*\) __P.*@/[)*,        ]\1[    (]/i\\\
\\/\\[^a-zA-Z_\\]\1 __P\\/d@p' > osdef1.sed

is the root of the evil for screen.  This should read something more
like the following:

sed < $srcdir/osdef.h.in -n -e '/^extern/s@.*[)*        ][)*    ]*\([^ *]*\) __P.*@/[)*,        ][(]*\1[)]*[    (]/i\\\
\\/\\[^a-zA-Z_\\]\1 __P\\/d@p' > osdef1.sed

to have a chance of working.

I'm starting to think that too many people are doing too many bogus
things in their autoconfig stuff, so that we should revert back to the
1.10 version and fix the set of ports that that breaks.  This isn't a
'C' language issue, but rather a brain-dead automatic configuration
issue.

Warner

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020409.144636.59877321.imp>