emu/debug: Extended memory commands for region and space

[holub]

No description provided.

[Osso13]

Please remember to update debughlp.cpp, too.

[holub]

Sure, right after I see others have no concerns regarding this.

[ajrhacker]

I don't think this is quite justified. There may be memory regions and memory shares with the same names.

[holub]

I was thinking about it. It's quite immoral to expect the user to know about internal representation.
Maybe I can check if this conflict exists and request to supply an additional parameter which clarifies the type.

[holub]

...or alternatively we can have savem which successes in case no conflicts only + samer, saves, etc

[galibert]

Or an alternative alternative, assuming the parser is not insane, having a savem s:whatever/r:whatever/etc where the non-prefixed DWIM when it's not ambiguous and errors otherwise.

[holub]

Making it draft and working on @galibert 's version.

[cuavas]

Is it really worth making the syntax more complicated to avoid an extra command? I already need to take a trip to the documentation when I want to do anything interesting with expressions (this section).

Also you’re outright removing a pair of commands which is going to break a pile of scripts, and also any ROM-patching cheats that depend on them.

On top of that, in the existing expression syntax, m is used for memory regions, not shares (e.g. monitor.mb@78 means “the byte at address 78 in the memory region with the tag monitor). If you introduce syntax in another place where m represents a memory share, that’s going to confuse people.

[holub]

I can keep loadr and add loads. My concern is the user not expected to know what she really wants - she seeing the name in dropdown and uses the name. How do you expect her to know if it's share or region or banked?

[cuavas]

If you really want to avoid adding extra commands, you could probably extend save and load to allow a region or share to be specified with a suffix similar to the ones used in expressions. (Right now they only operate on address spaces.)

For example, use the .m suffix to access a region, as we already do in expressions. So this:

save data.bin,200:monitor.m,100

Would be equivalent to this:

saver data.bin,200,100,:monitor

You’d need to choose another suffix for shares that doesn’t clash with current expression syntax (perhaps .s).

Also other things could be extended to support suffixes for region/share:

• Expression syntax
• fill, dump, strdump and find commands

The thing to be aware of if you do this is that then the <addr>:<spec> syntax will have two variants: one that only supports address spaces (e.g. for wpset) and one that also supports other things with suffixes.

But I think getting rid of loadr/saver or changing them in an incompatible way would be a mistake.

[cuavas]

I can keep loadr and add loads. My concern is the user not expected to know what she really wants - she seeing the name in dropdown and uses the name. How do you expect her to know if it's share or region or banked?

I mean, it tells you:

See there it says “Region” for memory regions.

And it says “memory” for shares.

They already need to know what a region is to use the .m syntax in expressions.

[holub]

So... the plan is:

• Keep saver
• Add savee without exposing
• Allow to specify .m .e for save without auto-magical selection

"e" corresponds to m(e)mory in dropdown or shar(e) in internal representation.

same for load/fill/dump/...

[cuavas]

If you’re extending the syntax for one of the load, save, fill, dump, etc. commands, it needs to be done for all of them. If you add a suffix for accessing a memory share, it needs to be supported in expressions as well as commands. It would be confusing otherwise.

The source argument parsing is already encapsulated for things that take the <addr>:<spec> syntax for address/space. It would be a matter of adding a variant that can yield a region or share instead of an address space. There’s already some kind of support for using regions as well as address spaces in the expression handling code.

[holub]

Understood. I'll be working on all variants then.

Any concerns about "e" ?

[cuavas]

I’m not sure. I would’ve gone for s since people using the MAME debugger are kind of assumed to have at least glanced at the MAME source. Any opinion @galibert?

[holub]

The plan is clear and it will take some time.
You have time to pick the letter.

[holub]

I've implemented save/load but don't really like how is the code structured. If you have any ideas how to make it better let me know while I'm implementing the rest.

[holub]

That must be it