Several little problems

Hi All. I have some little problems with RPCEmu.

1/ Fullscreen mode bug.
RPCEmu crashes on some PC, when going from windowed mode to fullscreen mode. I found exactly where the problem is.
It’s the “follow host mouse” option. When going fullscreen, the position in the host is unknown, and it crashes.

Solution could be to deactivate this option before switching to fullscreen mode. And to reactivate it after switching back to windowed mode. It works here. Unfortunately, I have no PC powerful enough to compile a version of RPCEmu with this possible patch.

2/ Menu bar in fullscreen mode
In fullscreen moden, the menu bar hides a part of the screen. Is it still needed? Especially if crashes in fullscreen<>windowed modes are solved…

3/ CPU load
Strange bug. Probably linked to the Wimp.
At start-up, the CPU Load is high for around 1 minute. Then it’s OK. Same effect on VRPC and RPC-Emu, with RISC OS 4 or 5.
If I don’t boot to Wimp, CPU load is low from start.

4/ Mouse
In non-wimp mode, mouse does not work. I used the needed code to activate it.
- If I boot in Wimp and launch a fullscreen task: it works
- If I boot in Wimp, leave it and launch my code: it works
- If I boot directly to Basic and launch my code: it does not work

5/ feature request: screen size
It’ll be very cool to have a SWI that would return the client size of the Window of RPCEmu. With Anymode, it could be possible to change the mode ‘on the fly’. And so to change of mode when going from windowed to fullscreen modes, or when resizing the window.

User mode networking, even if limited, would be useful too. But that’s probably a lot of work.

And I know that’s a strange request, but a ‘RISC PC – NewARM’ with more things (more memory, full FPA support, etc.) linked to a specific version OF RISC OS 5 could be a nice upgrade.

RPCEmu+Basic is for me more and more a viable option for my new software (boring IT things).
Just need a few minor enhancements.

3/ CPU load
Strange bug. Probably linked to the Wimp.

Yes, it’s the Wimp which is responsible for that. It waits about a minute before it starts doing any power-saving stuff. I’m not sure exactly what it does for that first minute – I think it just measures the system load so it’s able to tell the difference between an idle machine and a busy one.

Ah OK. Could it be shorter?
(you had to expect this question :) )

I’ve had a look at the code, and I think it should be possible to get rid of the delay completely. It might need tweaking a bit to make sure it still works correctly, but I don’t think it will be too difficult to fix.

(The Wimp uses a couple of self-adjusting thresholds to determine whether the machine is busy or not – so if the initial values of those thresholds are wrong then it might take a while before the algorithm starts making the right choices. The delay looks like it’s a bit of a bodge, to try and make sure the system is idle when it calculates the initial threshold values)

Thanks Jeffrey.

That would be good also for owners of real computers, as it’s probably slowing down the other apps.
For me it could solve the point 3 (and 4: no need not to run the Wimp).

CPU load issues should be fixed now.

https://www.riscosopen.org/viewer/revisions/logs?ident=1500924777-285394.html

Cooooool. I’ll test that. Thanks!