wdc@ATHENA.MIT.EDU (Bill Cattey) (04/20/89)
> I just took the newly cleaned up X11R3 Xibm server and > ran Andrew on it. I was told that the Save Under and > Backing store code was buggy, so I disabled it. > When I posted my first menu under Andrew, the server hung up. Thanks Bob for the speedy response. Saying that the server had to be providing the Expose events. The problem is that Xibm has an argument naming conflict in controlling the backing store. -bs disables backing store in the dix layer, and enables it in the ddx/ibm layer! The server, if queried would say the window did not have backing store enabled, and would then perform backing store operations for you, (which includes not sending expose events). The quick dirty solution is to use just the "-nobs" command line argument and to absolutely not use the documented "-bs" argument to disable backing store on Xibm. Perhaps IBM and the X Consortium can iron out how they want to name this crud. ---- Now my server is running, but my andrew windows are sill missing certain parts of redraws, and some of the redraws are missing bits. This is indeed upsetting, since, after working for two days with our server people merging in all the known Xibm fixes, I'd hoped the thing would be working by now. At this point, I suspect the Andrew stuff is at fault. -- 'You mean people enjoy doing this??' -wdc
dba+@ANDREW.CMU.EDU (David Anderson) (04/20/89)
>Excerpts from andrew.x11: 19-Apr-89 >Resolution on Xibm Backing .. Bill Cattey@athena.mit.e (1344) > >Now my server is running, but my andrew windows are sill >missing certain parts of redraws, and some of the redraws >are missing bits. This is indeed upsetting, since, after working >for two days with our server people merging in all the >known Xibm fixes, I'd hoped the thing would be working by now. >At this point, I suspect the Andrew stuff is at fault. > >-- 'You mean people enjoy doing this??' No, in fact, I hated it, which is why I've given up on X11R3 for the RT and gone back to running WM, the ITC's own window system. I might suspect the Andrew stuff too, except that it runs fine on Suns and Vaxen running R3. Most, if not all of the remaining apa16 (RT) bugs stem from synchronization errors between the CPU and display processor. One thing that helps is to only run on non-APC RT's (they're only half the speed) -- that's what Palo Alto developed/debugged the server on. --david P.S. For those keeping track, I still see what appear to be line drawing glitches, backing store failures, and font cache corruption.