ralph@mit-atrp.UUCP (Amiga-Man) (10/13/87)
I have noticed the following bug when running several programs with various screens, and with some in interlace; am I alone or do others find this ? I'm working with, lets say AegisDrawPlus, uEmacs, popCLI, performance monitor, and the workbench (I have 2.5Meg on a A1000). Now, I pop the uEmacs screen forward, do a little text action, then pop it to the back. The highres DrawPlus screen come forward as only it's upper half, displayed in non-interlace (200lines). It looks like the screen information got walked on. If, I only use left-amiga-N to bring the workbench forward everything lives and keeps running. BUT, if I try to use the mouse to push this broken screen back, lockup&reboot. The really bad news is that I've seen this with several different programs, And also without any uEmacs running. It almost seems like it's a bug in intuition, since it appears independent of the exact programs. It seems to be correlated with poping between interlace and non-interlace screens. If I'm not alone on this, maybe I'll try to get a specific reproduction of the setup and operations which cause it reliably. It is a major problem, and makes it hard to depend on the machine (which I really do depend on these days). AmigaTeX and it's previewer and uEmacs all together, with the workbench in interlace and uEmacs not in interlace seems to do it really often. It was so bad I was forced to make the previewer stay as a window on the workbench instead of being another screen, and using TxEd, so essentially everybody is on one screen. Then, since I never pop screens it doesn't happen. Bummer, Ralph
keithd@cadovax.UUCP (Keith Doyle) (10/16/87)
In article <1636@mit-amt.MEDIA.MIT.EDU> ralph@mit-atrp.UUCP (Amiga-Man) writes: >I'm working with, lets say AegisDrawPlus, uEmacs, popCLI, >performance monitor, and the workbench (I have 2.5Meg on a A1000). >Now, I pop the uEmacs screen forward, do a little text action, then pop >it to the back. The highres DrawPlus screen come forward as only it's upper >half, displayed in non-interlace (200lines). It looks like the screen >information got walked on. I've seen exactly this effect, when using a lo-res screen that has been modified to a high res screen by modifying the view and viewport structures, and doing a MakeView MrgCop LoadView etc. The information has not been walked on, somehow intuition seems to be confused as to just how many lines are supposed to be on screen. If the hi-res application later does another LoadView etc., the lower part of the screen will come back intact. I also found that if you go to the workbench and drag the workbench screen down in front of the hi-res screen, similar effects would occur. >If, I only use left-amiga-N to bring the workbench forward everything >lives and keeps running. BUT, if I try to use the mouse to push this >broken screen back, lockup&reboot. Haven't tried this, I was using a screen that didn't have a screen-to-back gadget. >The really bad news is that I've seen this with several different programs, >And also without any uEmacs running. >It almost seems like it's a bug in intuition, since it appears independent >of the exact programs. It seems to be correlated with poping between interlace >and non-interlace screens. If I'm not alone on this, maybe I'll try to get >a specific reproduction of the setup and operations which cause it >reliably. It is a major problem, and makes it hard to depend on the machine >(which I really do depend on these days). As I've said, it's happened to me. Unfortunately, what I'm doing is not exactly a Commodore approved mechanism for changing the resolution of a screen (is there one? as far as I know, no.) so CATS might just say it's not a bug but a feature. However, it seems that maybe the problem is that I somehow need to tell Intuition that the screen resolution has been modified, and if that is doable, it implies the bug is in DrawPlus and not in Intuition. Let me know what you find out. With my program, the problem is not at all critical, because it dosen't hang (no screen-to-back gadget) and it does its own LoadView etc. often enough to cure itself pretty quick. Still, if I knew what to do to fix it I certainly would. Keith Doyle # {ucbvax,decvax}!trwrb!cadovax!keithd Contel Business Systems 213-323-8170
rokicki@rocky.STANFORD.EDU (Tomas Rokicki) (10/17/87)
> Let me know what you find out. With my program, the problem is not at > all critical, because it dosen't hang (no screen-to-back gadget) and it > does its own LoadView etc. often enough to cure itself pretty quick. > Still, if I knew what to do to fix it I certainly would. To reproduce this reliably with a crash, write a program which does the following: Opens an interlace screen Opens another interlace screen Opens a non-interlace screen Now hit the to-back gadget about three or four times; the system will crash. That's all it takes. -tom