A bug that might occur if base_decimal or base_hexadecimal are put inside an if-branching

[FlatAssembler]

This program (you can see it live here):

address 0
if 2+2=4
	base_decimal
endif
constant ten, 10
inst ten
inst 10

assembles into:

0000a
00010

I'd expect it to assemble into:

0000a
0000a

It has to do with the fact that the assembler doesn't look into if-branchings, it simply jumps over them while assembling, because mnemonics must not be put inside the if-branchings. Given the way the assembler is structured, there doesn't seem to be a simple solution.

[FlatAssembler]

An obvious solution might be to make the assembler, when it encounters an if-statement, check whether the condition of the if-statement is true, and then, if it's true, invoke the preprocessor again with the if's second child (the body) as an argument. Putting something like this in assembler.js:

if (/^if$/i.test(node_of_depth_1.text) || /^while$/i.test(node_of_depth_1.text))
	if (node_of_depth_1.children[0].interpretAsArithmeticExpression(output_of_preprocessor.constants))
		makeCompilationContext(node_of_depth_1.children[1]); //Notice that we are not modifying the output_of_preprocessor

However, that would just replace this problem with another problem. Consider this code:

address 0
constant i, 1
if i>1
	base_decimal
endif
constant i, 2
inst 10

The problem here is that when the preprocessor sees if i>1, the i>1 will evaluate to false. But when the assembler sees that, it will evaluate to true.
So, once again, there is no obvious solution given the way the assembler is structured.