[Rpcemu] RPCEmu pre release test version 0.8.99

Peter Howkins rpcemu.howkins at marutan.net
Sat Oct 21 11:25:24 PDT 2017

On Sat, Oct 21, 2017 at 12:19:04AM +0200, Steffen Huber wrote:
> I did a quick test.
> Apart from the reported HostFS problem, I encountered a
> reproducible crash, with both interpreter and recompiler.
> I did a full CDVDBurn recompile, which means a lot of I/O.
> I took the taskwindow where the compile ran and put it to
> the bottom of the screen (a few pixels off-screen). I then
> clicked the window resize button and dragged it quickly down
> -> crash. I will try if RPCEmu 0.8.15 also exhibits this.

Hi Steffan, I've not been able to reproduce this crash yet, as such, 
here's a lot of questions.

Does it require the compilation to be happening in the task window or is 
it all the time?
Does it need to be the task window, or any window?
Were you in fullscreen, windowed mode, or does it happen in both?
Which wimp mode are you in, does it happen in other colour depths or 
If you're in windowed mode were you using 'Follow Host Mouse', or the 
capture mode when that is turned off?

> I like the new warning dialog that RPCEmu will be reset if you change
> the settings. One minor niggle: the warning dialog is centered to
> the display and not to the RPCEmu window.

I can look into that

> Is it still sensible to have RPC with ARM610 being the default
> machine? Does the recompiler work correctly with this config - ISTR
> reading that it needs StrongARM config to be active?

For the last couple of years the recompiler has supported all the ARM CPU
models. There no speed benefit for StrongARM as we emulate them in the 
same way, so we stick to ARM610 as this is compatible with all versions of
RISC OS from 3.5 onwards.

> Host mouse following mode still exhibits a minor problem when
> e.g. a RISC OS error window moves the RISC OS mouse pointer inside
> the restricted area. It is irritating that the host mouse pointer
> is not also silently moved. On the other hand, I agree that the
> host mouse pointer must not be restricted to the RO pointer
> rectangle.

This has actually changed from 0.8.15 where we did capture the mouse in 
the error box, but as I was changing the code for Qt I couldn't make up my 
mind if this was the correct behaviour. There's not really a good option 
here :(


Peter Howkins
peter.howkins at marutan.net

More information about the Rpcemu mailing list