Mode change behaviour

I have tried again with your fixes, Jeffrey. AnyMode aside, it is better. Thanks for that.

I tried again going into MODEs 3 and 6, then doing things like *Modules and there is some odd repetition or redraw issue. But it isn’t really a problem – who on earth would be using those modes these days?!

Jeffrey, did you add a botch to get AnyMode working? Or fix an issue?

Just wondering if I need to figure out how to get it working, as it’s not going to be straightforward.

Just this fix from a few days ago: https://www.riscosopen.org/viewer/revisions/logs?ident=1542668558-714024.html

If you request a mode that’s in a supported colour depth, you should get an AnyMode mode.

If you request a mode that’s in an unsupported colour depth, Service_ModeTranslation will be invoked, so you’ll end up with whatever mode the kernel or ScreenModes Service_ModeTranslation implementation thinks is best (which won’t be an AnyMode mode, since the kernel + ScreenModes only try modes which they personally know about).

This thread does pose the question that maybe there should be one generic Service_ModeTranslation implementation that knows how to use “any” mode as a replacement. Although it would rely on the Service_EnumerateScreenModes output to pick fallback resolutions (which would cause it to fail to pick AnyMode’s on-the-fly generated modes), the routine could be coded so that it first tries the originally requested resolution (in different BPPs) without checking to see whether it exists in the “known resolutions” list – which should restore AnyMode to full working order.

Just this fix from a few days ago

Cool, bug fixes.

If you request a mode that’s in an unsupported colour depth, Service_ModeTranslation will be invoked, so you’ll end up with whatever mode the kernel or ScreenModes Service_ModeTranslation implementation thinks is best

That’s fine…I think :/

The reason AnyMode exists is to cover the mismatch between what resolution the hardware can support, and what the OS thinks it supports – MDF/EDID do not lend themselves to upscalers. Unless someone is using AnyMode to get a random resolution at a bpp that’s not supported then as tested with the 25-11-18 build, I think AnyMode is now working acceptably.

New version of the diagram/flowchart is done.

https://www.riscosopen.org/forum/forums/3/topics/3135?page=3#posts-86408