Strange math evaluator bug. print([y,mode,15]) => [?,?, not 15]

[Reptorian1125]

This doesn't look like I can reproduce in other code, but do you find this strange?

[gmic]./ Start G'MIC interpreter (v.3.4.3).
[gmic_math_parser] [y,mode,15] = (uninitialized) (mem[68]: vector3)
[gmic_math_parser] [y,mode,15] = [ 0,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 1,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 2,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 3,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 4,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 5,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 6,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 7,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 8,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 9,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 10,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 11,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 12,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 13,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 14,1,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 15,0,15 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 16,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 17,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 18,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 19,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 20,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 21,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 22,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 23,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 24,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 25,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 26,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 27,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 28,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 29,1,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 30,0,1 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 31,1,2 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 32,1,2 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 33,1,2 ] (size: 3)
[gmic_math_parser] [y,mode,15] = [ 34,1,2 ] (size: 3)
[gmic]./ Start interactive display of 2D image [0]

That's not supposed to happen, is it? Look at the 15, and see the 1 and 2.

[dtschump]

Could you please copy/paste a minimal G'MIC command that reproduces this bug ?

[dtschump]

This works for me:

foo :
  1,35,1,1,"const mode = 1; print([y,mode,15])"

[Reptorian1125]

Could you please copy/paste a minimal G'MIC command that reproduces this bug ?

Without knowing what happened, I will post the one without a parameter. You can see 15 becomes 1 for some reason. I will see if I can find the minimum example.

Well, this is the current minimal example I have:

foo:
35,1,1,3,"begin(
	v=vectors();
	const dec_max_sum=15-1;
	const end_pos=s-1;
	scanline_pos=end_pos;
	mode=1;
	started=0;
	limit=15;
	old_mode=1;
 );
 started?(
	mode?(
		if((++v[end_pos])==limit,
			mode=0;
		);
	):(
		v[scanline_pos]==15?(
			copy(v[scanline_pos--],0,s-scanline_pos,1,1);
			v[scanline_pos]=1;
			mode=1;
			limit=dec_max_sum;
		):(
			mode=1;
			temp_sum=0;
			for( p = scanline_pos, p < s, ++p,
				temp_sum+=v[p];
				if(temp_sum==limit,
					mode=0;
					break();
				);
			);
			mode?(
				limit=15-temp_sum;
			):(
				copy(v[p],0,s-p,1,1);
				++v[scanline_pos];
			);
		);
	);
 );
 print([y,mode,15]);
 started=1;
 v;
 "

EDIT: Removing some of the code makes the bug go away. So, I cannot find the minimum example other than this which triggers the bug.

[dtschump]

Well, I'd suggest that you check your copy() functions.
Using copy is prone to error as you can scramble the values in the internal math parser memory.
Just for instance:

eval "
  a = 0;
  b = pi;
  copy(a,[1,2]);
  print(b);
"

will print b = 2....

[Reptorian1125]

Thanks, looking a closer look at it resolved it. Now off to finish a new combinatorics command.