On Tue, Sep 04, 2012 at 09:18:20AM -0600, Eric Blake wrote:
On 09/04/2012 08:57 AM, Jasper Lievisse Adriaanse wrote:
>>From bafcb4ed2b90b5ba845ca6b61861e3caa548b16a Mon Sep 17 00:00:00 2001
> From: Jasper Lievisse Adriaanse <jasper(a)humppa.nl>
> Date: Tue, 4 Sep 2012 16:57:09 +0200
> Subject: [PATCH] Include some extra headers needed for OpenBSD.
>
> ---
> src/util/virnetdevbridge.c | 9 +++++++++
> 1 files changed, 9 insertions(+), 0 deletions(-)
Please show the compiler errors that you got without these includes. I
can't help but wonder if you have instead uncovered a bug in the gnulib
headers, but knowing which symbols were not declared makes a difference
in answering that question.
>
> diff --git a/src/util/virnetdevbridge.c b/src/util/virnetdevbridge.c
> index 7b11bee..8559223 100644
> --- a/src/util/virnetdevbridge.c
> +++ b/src/util/virnetdevbridge.c
> @@ -30,6 +30,15 @@
> #include "intprops.h"
>
> #include <sys/ioctl.h>
> +
> +#ifdef HAVE_SYS_PARAM_H
<sys/param.h> is non-standard; what are we using that requires us to
probe for its existence? Should gnulib consider guaranteeing this
header in spite of it being non-standard?
Nope, it was an error on my side to
include it. I thought it was needed by
socket.h, but after trying to show you the error...there was none.
> +# include <sys/param.h>
> +#endif
> +
> +#ifdef HAVE_SYS_SOCKET_H
This line shouldn't be necessary; gnulib guarantees a working
<sys/socket.h> on all architectures.
OK. Could you please push this one
then?
--
Eric Blake eblake(a)redhat.com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org
--
Cheers,
Jasper
"Stay Hungry. Stay Foolish"