On Mon, 2019-07-29 at 12:39 +0100, Daniel P. Berrangé wrote:
On Fri, Jul 26, 2019 at 05:46:38PM +0200, Andrea Bolognani wrote:
> On Fri, 2019-07-26 at 16:21 +0100, Daniel P. Berrangé wrote:
> > On Fri, Jul 26, 2019 at 05:19:56PM +0200, Andrea Bolognani wrote:
> > > [...]
> > > > +REMOTE_DAEMON_LD_ADD += \
> > > > + ../gnulib/lib/libgnu.la \
> > > > + $(LIBSOCKET) \
> > > > + $(NULL)
> > >
> > > As an aside, it looks like $(LIBSOCKET) is a leftover of days long
> > > gone and no longer used for anything.
> >
> > It is something that gnulib defines. Whether it expands to a non-empty
> > string on any of our supported build platforms though, I don't know.
>
> I tried removing it and ran it through the full gauntlet without
> getting any failures, so I'm pretty confident we don't need it. It'd
> be pretty weird if we did, since we have at least two other daemons
> already and neither of those is using it...
LIBSOCKET is set to -lws2_32 on Windows builds.
We don't build libvirtd on mingw though which is why you don't see
a problem. Previously we would have needed this for cygwin I expect
but that's not a supported build target
I guess anyone who wanted to (re?)introduce support for non-MinGW
builds will have to address many more issues, so as long as it's not
needed on any of our target build platforms we can safely drop it.
--
Andrea Bolognani / Red Hat / Virtualization