On Thu, 2017-06-22 at 08:35 +0100, Peter Maydell wrote: > On 22 June 2017 at 08:33, Gerd Hoffmann <kra...@redhat.com> wrote: > > diff --git a/ui/Makefile.objs b/ui/Makefile.objs > > > index 3369451285..08fb573a48 100644 > > > --- a/ui/Makefile.objs > > > +++ b/ui/Makefile.objs > > > @@ -6,6 +6,14 @@ vnc-obj-y += vnc-auth-vencrypt.o > > > vnc-obj-$(CONFIG_VNC_SASL) += vnc-auth-sasl.o > > > vnc-obj-y += vnc-ws.o > > > vnc-obj-y += vnc-jobs.o > > > +ifeq ($(CONFIG_VNC_SASL),y) > > > +# silent OSX SASL warnings (from https://stackoverflow.com/a/740 > > > 6994 > > > ): > > > +# because OpenSSL doesn’t offer API compatibility between > > > versions, > > > [...] Apple > > > +# can't provide security updates without breaking existing apps, > > > so > > > is migrating > > > +# from OpenSSL to Common Crypto. > > > +vnc.o-cflags := -Wno-deprecated-declarations > > > +vnc-auth-sasl.o-cflags := -Wno-deprecated-declarations > > > +endif > > > > Hmm, does clang understand "#pragma GCC diagnostic" ? > > > > If so, then this can be done in in the code not the makefiles, and > > also > > selectively for osx only. > > It does have some diagnostic pragma support. The awkward part is > that it has to be in force at the point where the deprecated > function is used, not where it's declared. So you can't just wrap > the #include of the ssl header in pragmas, you'd have to either > do it at every callsite or else over the whole .c file.
Do it on the whole file on osx only is still better because we continue to get the warnings on non-osx then, which should be enough to avoid the code bitroting unnoticed. cheers, Gerd