Allow {symbol} interpolation outside of strings

[Rangi42]

Fixes #629

Closes #631

[ISSOtm]

Unfortunately, this implementation suffers from a bug that macro arguments used to have (#63).

issotm@sheik-kitty ~/rgbds% ./rgbasm - <<< 'S = 42                        [interpolation|✔] 
quote> PRINTV 0{d:S}0
quote> '
ERROR: <stdin>(2):
    syntax error
error: Assembly aborted (1 errors)!
$0

The proper fix for that issue was to expand arguments before everything else (hence peek itself performing the macro arg expansion).

[Rangi42]

Now peek is solely responsible for doing symbol interpolation outside of strings. (readString has its own readInterpolation call because its peek calls need to have interpolation disabled.)

This fixes the above issue; interpolation will work inside of numeric literals.

[ISSOtm]

Why do strings need interpolation disabled?

[Rangi42]

Inside readString, peek needs to not do interpolation itself, since that can lead to invalid syntax. For example, in PRINTT "{__FILE__}\n", __FILE__ is "\"file-sym.asm\"", so without the lexerState->disableInterpolation = true, it would see PRINTT ""file-sym.asm"".

[ISSOtm]

Alright, I'm not a fan of this special case, but it's necessary for compatibility, and the implementation is acceptable.

Checkpatch notes two style mismatches.

Finally, as this is a touchy part of the codebase, I'll page @AntonioND for a second pair of eyes; if he's satisfied with the change as well, he can merge this.

[Rangi42]

This, plus more printf-like format specifiers, will really simplify pokered's code.

TechnicalMachines:
n = 1
REPT NUM_TMS
	db TM{02d:n}_MOVE
n = n + 1
ENDR

;\1 source map
def_warps_to: MACRO
; output and purge each _WARP_TO_NUM_<N> warp_to, from N=1 to _NUM_WARPS
_WARP_TO_WIDTH = \1_WIDTH ; used in each _WARP_TO_NUM_<N>
n = 1
	REPT _NUM_WARPS
		_WARP_TO_NUM_{d:n}
		PURGE _WARP_TO_NUM_{d:n}
n = n + 1
	ENDR
ENDM

The "output and purge" comment there is basically self-explanatory.

[AntonioND]

Ooops, I didn't see the review request. I've received so many notifications from this repo in the last few days that I've just started to get rid of most of them...

My problem with this PR is that it is just not clear what it does. There is no documentation and the only test is so long and complicated that is not useful as an example.

[ISSOtm]

The lack of documentation is certainly a problem. I would define the content of this PR as "expanding the contents of symbols, like in strings, but in code".

@AntonioND If you're receiving too many notifications, you may want to tune your "Watch" settings?

[AntonioND]

The lack of documentation is certainly a problem. I would define the content of this PR as "expanding the contents of symbols, like in strings, but in code".

Yeah, true.

@AntonioND If you're receiving too many notifications, you may want to tune your "Watch" settings?

Well, most of the time this isn't a problem because there isn't that much activity, and I like the instant notifications in that case. In situations like the current one, when there are so many changes, just ping me on Discord to make sure I see it.

[Rangi42]

I've simplified the test case file and added it to the man documentation. It covers the possible uses of {interpolation}:

• Applied to strings
• Applied to numbers
• Nested
• Using format specifiers
• Interpolating within ordinary code context
• Interpolating within a numeric literal
• Interpolating within a name to be defined
• Interpolating within a name to be purged

Please let me know if the example should be edited further.

[AntonioND]

Yeah, that's better.

[ISSOtm]

Alright, I'm OK with the new changes as well, so we'll (finally!) merge this.