gordon@maxwell.waterloo.edu (Gordon R. Strachan) (11/27/90)
Hello there. We have recently acquired an HP 700/X X terminal (model C2300B). While, all in all this is a nice terminal and we a quite happy with it there is one severe problem we are having. This terminal seems to be completely allergic to starbase. We use starbase a lot on our system, and we have a number of graphical programs which we wish to display on the X terminal. In theory this is possible through with the starbase sox11 device driver. Indeed, I have not had any trouble doing this on other remote X servers (both HP servers and others such as Sun and NCD terminal). However, on the 700/X, the terminal will randomly lock up and die. There seems to be a critical timing bug associated with the opening of the window and the call to gopen. In theory, this shouldn't happen because I wait for server to notify me that the window is open before I attempt to call gopen. Also, the terminal seems to view the bitmap_to_file function as particularly evil. It sometimes will work, but generally causes the terminal to die. Now I would like to think that HP would have tested this terminal with their own software before releasing it so what gives? Has anyone else seen this type of problem with this terminal? Any suggested work arounds? I would submit a software report on this one but it is not clear to me who it should be sent to, the HP-UX guys or the HP 700/X guys. Actually I don't even now how to submit bug reports on that terminal now that I think of it. We are using HP/UX v 7.0 and the 700/X server version is listed as B.03.00. I don't suppose by any miracle there is a later version of the server which might fix this bug. Gordon
stroyan@hpfcso.HP.COM (Mike Stroyan) (11/28/90)
> There seems to be a critical timing bug >associated with the opening of the window and the call to gopen. In theory, >this shouldn't happen because I wait for server to notify me that the window >is open before I attempt to call gopen. Also, the terminal seems to view >the bitmap_to_file function as particularly evil. It sometimes will work, >but generally causes the terminal to die. The older MIT sample servers, (and possibly the terminal's software), would not consider a window id valid for other connections until the window was mapped for the first time. The sox11 driver has a separate connection to the display, so it could run into this problem. To avoid the problem map the window and wait for a MapNotify event before calling gopen. You actually have to wait for a MapNotify because the window manager can redirect an XMapWindow request and delay the mapping indefinitely. The bitmap_to_file function uses block_read to get data from the server. There was a correction to the block_read function in the sox11 driver made last January. It corrected block_read data problems on the four plane HP X terminals. The first release with that fix was 7.03. Neither of these problems are known to make the server die. >I would submit a software report on this one but it is not clear to me who >it should be sent to, the HP-UX guys or the HP 700/X guys. Actually I don't >even now how to submit bug reports on that terminal now that I think of it. Go ahead and send it to the HP-UX guys. They can talk to the HP 700/X guys. Full version numbers and short source code examples are much appreciated. Mike Stroyan, mike_stroyan@fc.hp.com
gordon@maxwell.waterloo.edu (Gordon R. Strachan) (11/28/90)
In article <7370261@hpfcso.HP.COM> stroyan@hpfcso.HP.COM (Mike Stroyan) writes: >> There seems to be a critical timing bug >>associated with the opening of the window and the call to gopen. In theory, >>this shouldn't happen because I wait for server to notify me that the window > >The older MIT sample servers, (and possibly the terminal's software), >would not consider a window id valid for other connections until the >window was mapped for the first time. The sox11 driver has a separate >connection to the display, so it could run into this problem. To avoid Hmmm.... I had suspected that it had did this thanks for the confirmation. >the problem map the window and wait for a MapNotify event before calling >gopen. You actually have to wait for a MapNotify because the window Double checking my code I did find that I wait for the MapNotify event. I used to call XSync and this didn't work at all (obviously because of the different display connection). Still, even with the MapNotify I will occasionally get into trouble. >manager can redirect an XMapWindow request and delay the mapping >indefinitely. > >The bitmap_to_file function uses block_read to get data from the server. >There was a correction to the block_read function in the sox11 driver >made last January. It corrected block_read data problems on the four >plane HP X terminals. The first release with that fix was 7.03. > >Neither of these problems are known to make the server die. > >>I would submit a software report on this one but it is not clear to me who >>it should be sent to, the HP-UX guys or the HP 700/X guys. Actually I don't >>even now how to submit bug reports on that terminal now that I think of it. > >Go ahead and send it to the HP-UX guys. They can talk to the HP 700/X guys. >Full version numbers and short source code examples are much appreciated. > Tell you what, I'll talk to my SE and try to get v7.03 of the sox11 driver. That might fix my problems. If it doesn't I'll try to put together some sample code. But like I said it's kind of random so its hard to figure out the minimum amount of code required to hit it. By the way, we used to kill the 7.0 server running one the workstation consoles a lot when doing starbase graphics using the 300h device driver. I grabbed the 7.03 server off the net a while back and besides being a lot faster, it seemed to fix that problem. All in all, the 7.03 server seems to be quite stable. >Mike Stroyan, mike_stroyan@fc.hp.com Thanks for the help Gordon
jbb@hpcvlx.cv.hp.com (Jim B. Byers) (11/30/90)
> I grabbed the 7.03 server off the net a while back and besides being a > lot faster... Great! Glad you found it useful. I want to clarify one point. You probably have the server that is made available from: 15.255.72.15 hpcvaaz.cv.hp.com (an open subnet machine). This ftpable server has a what string as follows (note the D): X.7.03.300: X Window System, Version 11 HP-UX 7.03 A.03.02 $Revision: 9.30.1.3 $ $Revision: 64.4 $ hpa1096 R703 $Date: 90/05/03 14:55:04 $ hpa1416 R703 $Date: 90/05/07 10:19:49 $ 350.1.7.1 05/10/90 libxd300h.a 350.1.7.1 05/10/90 libxd300l.a 350.1.7.1 05/10/90 libxd98550.a 350.1.7.1 05/10/90 libxd98720.a 350.1.7.1 05/10/90 libxd98730.a 350.1.7.1 05/10/90 libxd98704.a 350.1.7.1 05/10/90 libxd98735.a 350.1.7.1 05/10/90 libsdiutils.a This server is identical to the server that ships with the 7.03 instant Ignition Series 400s except that it knows how to run fast on the 98550 (1280x1024) graphics card. If you take the standard 7.03 server and run it on the 98550 card, X11 performance will be about 10% faster than the 7.0 server. If you take the server from hpcvaaz you will get the performance on the 98550 card that is the same as available on a 7.03 Series 400 Color VRX. This combination is noticabily faster than the 98550/7.0 combination. The 98550/post 7.03 server (hpcvaaz combination) is a supported combination on HP-UX 7.0 and 7.03. In short, ftp this server if you have this graphics card and want a boost in X11 performance. Jim Byers Interface Technology Operation Lab/Marketing Team "The X11/Motif/Vue/Architect folks in Corvallis."
stroyan@hpfcso.HP.COM (Mike Stroyan) (12/03/90)
> Tell you what, I'll talk to my SE and try to get v7.03 of the sox11 driver. > That might fix my problems. If it doesn't I'll try to put together some > sample code. But like I said it's kind of random so its hard to figure out > the minimum amount of code required to hit it. You can't, in general, mix Starbase drivers from one release with libraries from another release. If you update the driver you should update all of the system. Mike Stroyan, mike_stroyan@fc.hp.com