SjASMPlus от z00m

[NEO SPECTRUMAN]

meanwhile, if you have this construct in separate module, consider to switch off warnings globally by `--msg=err` level, producing only errors,

ну глобально я оно отключить не могу
так как у меня выводятся и мои важные предупреждения средствами lua (по крайней мере я так думаю : )
да и у меня тонны ошибок в коде...

так что я пока побуду на 1.15.0
пока не добавится ;ok

- - - Добавлено - - -

так же мне недавно попадалось


на нескольких вот таких таблицах на 256 значений

	pulse_frq2 = {}
					
	pulse_frq2[0] = ((0.50001545147716300 * pulse_fundamental_mlt) + (0.54171988752370500 * pulse_harmonic_mlt))
	pulse_frq2[1] = ((0.49965567733372600 * pulse_fundamental_mlt) + (0.51692156470823500 * pulse_harmonic_mlt))
	pulse_frq2[2] = ((0.49929590319028900 * pulse_fundamental_mlt) + (0.51590447856727700 * pulse_harmonic_mlt))
	....

правда все решилось
добавлением endlua lua allpass посредине

[Ped7g]

так же мне недавно попадалось

yes, the total buffer for lua text script to be executed is about 32kiB. I was refactoring this just for v1.15.1 to make it constant in the source, so if you are building sjasmplus from source, you should easily find the new constant and raise it (and disable the allpass warning), but I think you use only the binaries.

I'm not sure if, and by how much I should raise the script length limit, as I have little idea what limits are on the lua side (on sjasmplus side it's all about heap memory, so I can probably allocate also 10 or 100MB without issue on modern PC). Also the buffer is fixed size, allocating for every script in full length, but maybe I can check C++ `std::string` performance characteristics to see if it does deal with `append` effectively, not putting any fixed limit on the usage at all, leaving that to dynamic string allocator. Or maybe I can just use `std::vector` for this too.

There's lot of sjasmplus source which is not using basic C++ idioms and doing things manually like in C, but I never had time to rewrite some parts just to make them more C++, as I always have to hunt some bug or add feature, and also it took time to have enough test coverage to make such source rewrite reasonably safe (without test coverage the risk of breaking old stuff is too high). So maybe one day... ... in sjasmplus v2.x ...

[NEO SPECTRUMAN]

but I think you use only the binaries.

так и есть
собирать sjasm-ы пока не научились

[NEO SPECTRUMAN]

`if 1 <= aaa && aaa <= 3`
or
`if 1 == aaa || 2 == aaa || 3 == aaa`

это конечно хорошо
но хотелось бы использовать вот так

	if super_mega_very_long_and_big_stupid_variable_maybe_not = 0 || 1 || 5 || 7
	nop
	endif

или так

	if super_mega_very_long_and_big_stupid_variable_maybe_not = (0 || 1 || 5 || 7)
	nop
	endif

а приходится делать так

что абсолютно не читаемо

	if super_mega_very_long_and_big_stupid_variable_maybe_not = 0 || super_mega_very_long_and_big_stupid_variable_maybe_not = 1 || super_mega_very_long_and_big_stupid_variable_maybe_not = 5 || super_mega_very_long_and_big_stupid_variable_maybe_not = 7
	nop
	endif

еще могу предположить что кому нибудь может понадобиться такая конструкция

	if var_a || var_b || aaa || bbb = 1

- - - Добавлено - - -

так же только что нашел такую ошибку
коментарий написанный впритык уничтожает endif
сходу и не ясно где ошибка и почему

выдает такое
error: Label not found: label2
error: [IF] No ENDIF

aaa = 3

		jp label
				
		if aaa = 3
		nop
		endif
label

		jp label2
		
		if aaa = 0
		nop
		endif;	
label2

[Ped7g]

так же только что нашел такую ошибку
коментарий написанный впритык уничтожает endif

thanks, I will take a look on it, I guess it's about the cloned-parser-code used to parse some special cases like `endif/endlua/....` and similar, not using the main regular parser, so I guess every of those cloned ones is small original and not behaving the same way as regular, I will have to review them all.

но хотелось бы использовать вот так

The proposed syntax is already valid in sjasmplus, but has different meaning. The logical or is logical or.

But I get your idea, I will try to search for some nice common way how these things are done, the C++ doesn't have built-in operator for this kind of calculation IIRC, but some languages have them...

I think possible syntax may be like:

  IF long_expression isfrom(0, 1, 5, 7)
  ENDIF

I just can't recall what are the common names in other languages for such operator, and I need to check what kind of "array" enclosing is best fit for sjasmplus syntax, I think I can even go with`isfrom[0, 1, 5]` using brackets (I'm a bit worried about parentheses), the brackets can clash only with DEFARRAY, which should be quite safe usage, unless you decide to do `DEFARRAY isfrom`

BTW, if your "very long label" contains only very limited amout of numbers, like 20 to 29, you can in current/old sjasmplus use the DEFARRAY lookup-table trick:

very_long_label = 20
  DEFARRAY nextIfCheck 1, 1, 0, 0, 0, 1, 1, 0, 0, 0  ; true for: 20, 21, 25, 26
  DUP 10
    IF nextIfCheck[very_long_label - 20]
        DISPLAY /D, very_long_label, " -> true"
    ELSE
        DISPLAY /D, very_long_label, " -> false"
    ENDIF
very_long_label = very_long_label + 1
  EDUP

[Ped7g]

Tried to search for something similar in other languages, but nothing precisely fits sjasmplus current syntax, so I'm leaning toward "in_array" operator:

  ; "in_array" operator:

  ; variant with literal array directly in the expression
    ; syntax: <expression> in_array [0, 1, 4, 5]
      ; result is -1 when <expression> evaluates to 0, 1, 4, 5
      ; result is 0 when <expression> evaluates to anything else

  ; variant with array defined by DEFARRAY
    ; syntax: <expression> in_array <defarray name>
      ; result is -1 when <expression> does match (evaluated!) value in DEFARRAY
      ; result is 0 when <expression> evaluates to anything else

 ; examples:
  IF (very_long_label % 33) in_array [0, 1, 5, 7]
  ENDIF
  DEFARRAY freeday 6, 7
weekday = 2  ; tuesday
  IF weekday in_array freeday
    ; not hit
  ENDIF
weekday = 7  ; sunday
  IF weekday in_array freeday
    ; *hit*
  ENDIF

I need to review it a bit more to make sure this doesn't clash with current syntax in any edge-case, but so far it looks good to me and should be possible to implement it.

Any comments before I add it?

[Bedazzle]

I just can't recall what are the common names in other languages for such operator

For example, Python:

if variable in (1,2,3):
   ...

[Ped7g]

For example, Python

Thinking about using "in" instead of "in_array" (this one is from PHP), but "in" is already Z80 instruction, so it's adding to the visual ambiguity when reading the source (it's not a direct clash for assembler, as it's instruction vs operator, but I think I will prefer "in_array" any way, to make it distinct even at first look).

[Bedazzle]

Thinking about using "in" instead of "in_array" (this one is from PHP), but "in" is already Z80 instruction, so it's adding to the visual ambiguity when reading the source (it's not a direct clash for assembler, as it's instruction vs operator, but I think I will prefer "in_array" any way, to make it distinct even at first look).

if variable from (1,2,3,4,5)

?

[Ped7g]

https://github.com/z00m128/sjasmplus...es/tag/v1.16.0
* predefined defines extended with new ones (like __BASE_FILE__ or __LINE__)
* relocation data generator (inspired by SymbOS executables)
* some bugfixes/improvements in parser, some warnings are suppressible now

NEO: now the new lua warning about bytes emitted without "allpass" should be suppressible (you can put the "; ok" comment either at the beginning of lua block, or at the end of it, whichever way you prefer, both lines will count).