On Mon, 3 Apr 2017 09:47:50 -0400 (EDT) Olivier Fourdan <[email protected]> wrote:
> Hi Quentin, > > > I think it may be worth it to add a little safety net: > > “Compositors are required to restrict this interface to Xwayland alone, > > and raise a protocol error for native Wayland clients.” > > Yes, agreed, good point! But do we really need to kill a client that would > dare to try to bind to the grab interface? > > Right now my implementation simply ignores the bind request... Hi, you cannot ignore a wl_registry.bind request. Did you ever try to see what happens? :-) If you do not create a wl_resource when the protocol spec says "this creates a new object", then if the client ever refers to the non-created object, it will hopefully be automatically shot dead for referencing a bad object id. Furthermore, I think it also violates the object id allocation scheme, so if the client tries to create any new object after the ignored bind request, that would also explode. So you might just let the client out of its misery before anything more confusing can happen. At least you can then deliver a descriptive error message. ;-) As for the protocol wording, I think it should be enough to say that compositors should restrict the interface to Xwayland. How they do that is up to them. Thanks, pq
pgpHfEPme63NN.pgp
Description: OpenPGP digital signature
_______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: https://lists.x.org/mailman/listinfo/xorg-devel
