Idea for the next version of Stella (if possible)

[karokoenig]

http://home.comcast....2600-daptor.htm

 

Thanks for that tip. I contacted him. It's probably gonna be a very expensive thing for me. Anyone an idea how to get something similar in Europe?

[Tjoppen]

Please include this patch. I changes the background color on the playfield/GRPx widgets to COLUBK instead of black, which makes debugging easier.

0001-Use-COLUBK-instead-of-kBGColor-as-background-for-GRP.patch.txt

[+stephena]

OK, this patch is applied. It's a good idea, and I didn't consider it before. This will be in the 3.9.1 release.

 

There are several other good ideas in this thread, and I may get to some of them for the 3.9.1 release ...

[+stephena]

Tell DASM to create a symbol file. The name has to be identical to the binary created, except for the extension which has to be ".sym"

 

So if your file is test.bin use: -stest.sym

 

Also, Stella 3.9.1 will (partially) support DASM list files, which you generate very similar to symbol files:

So if your file is test.bin use: -stest.sym -ltest.lst

 

This is very useful to differentiate symbolic constants from symbolic addresses. Eventually, the lst support may be further extended (theoretically, the lst file contains the entire disassembly, so Stella could use it to make the output very accurate, essentially source-level debugging). But that is some time away, if ever ...

[Tjoppen]

OK, this patch is applied. It's a good idea, and I didn't consider it before. This will be in the 3.9.1 release.

Thanks

[+stephena]

Actually, I may have to modify the patch slightly, since it's still somewhat confusing when taking object precedence into account. But I understand your intent, and that the having black meaning 'nothing' (and not the colour black) is confusing. I think I'll just leave it blank in that case.

[Tjoppen]

it's still somewhat confusing when taking object precedence into account

It is? COLUBK is always furthest back..

 

Random idea and note to self for a potentially better debug display: show the whole line as the TIA would generate it if no registers were changed. Maybe some little arrows pointing to where all graphics objects are, and where the beam is.

[enthusi]

Random idea and note to self for a potentially better debug display: show the whole line as the TIA would generate it if no registers were changed. Maybe some little arrows pointing to where all graphics objects are, and where the beam is.

 

That'd be wicked indeed.

[+stephena]

If, for example, GRP0 contains 'nothing' (ie, a zero), then what will be 'drawn' for that register is the stuff under P0 (which will differ depending on the priority currently in use). So always filling it with COLUBK is also misleading, since P0, PF, or BK are possible values. I think it makes more sense in this case to simply make the pixel 'transparent' in the UI, since that's more like what's actually happening. I make it transparent by using the background dialog colour (ie, it simply blends into the background).

 

There are a few cases where it makes sense to use COLUBK. One is in normal priority mode, when PF is empty. In that case, it indeed should show the value of COLUBK, since that's the only thing under it in priority.

 

As for showing a full TIA line and/or arrows, etc., mockups of desired functionality are welcome. I have a little trouble picturing how this would look and/or work, but I'm open to suggestions.

[+SvOlli]

I found a problem due to the current change of the bank switching: I'm calling different banks through a subroutine that analyses the top three bits of the PC. So bank 0 is called at $1xxx, bank 1 at $3xxx, ..., bank 7 at $Fxxx.

 

But the debugger assumes that the ROM as always used at $Fxxx, which makes it hard to set breakpoints. I usually forget to set the correct address by hand, right mouse button -> set breakpoint doesn't work. Maybe just masking out $1fff during the comparision would do the trick.

[Tjoppen]

 

stephena: What about displaying the actual GRPx/PFx bits as well? Colored boxes + 0/1. And yeah, if I find the time I might mock something up for an even more awesome graphics debug display.