On Thu, Jun 23, 2011 at 10:53:34AM +0100, Daniel Stone wrote: > Hi, > > On Thu, Jun 23, 2011 at 01:41:58PM +1000, Peter Hutterer wrote: > > On Thu, Jun 09, 2011 at 06:17:11PM +0100, Daniel Stone wrote: > > > Add a flags member which will be copied wholesale into the resultant > > > xXIRawEvent. > > > > sorry to flip-flop on this, but seing this patch and reading the XI2 proto > > definition of XIRawEvents makes me think that solving raw events properly > > is going > > to be trickier than just the flag setting. Technically the raw data is a > > button press but the transformed values are of a different number, etc. etc. > > > > If raw events should continue to be the driver-native data it may be better > > to just skip them altogether and not emulate them at all. What do you think? > > Hmm, it's hard to say. Certainly on the face of it it seems cleaner to > leave it as driver-native data, but I'm kind of worried that clients may > have come to rely on the principle that any DeviceEvent will show up > with a corresponding RawEvent. > > If we're going down this route, it might be best to cut raw events from > things like XTEST/WarpPointer as well?
WarpPointer certainly. XTEST not sure, because in the end the XTest device is a device, so the data could be regarded as raw data. Plus, xtest is currently the only sensible way to emulate devices from an X client, being able to use raw events here seems beneficial (and it doesn't require extra code anyway). Cheers, Peter _______________________________________________ [email protected]: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: http://lists.x.org/mailman/listinfo/xorg-devel
