Number subindexes not detected

[Lartu]

At the moment, the LDPL compiler doesn't find any errors in codes like

data:
myNum is number
otherNum is number

procedure:
store myNum:0 in otherNum

and thus the C++ compilation crashes after the IR is generated. This should be fixed 😞

[Lartu]

@dgarroDC as you were working with sub-indexes and all that in the recent days (and thus you should remember how this all works better), is there any chance you could look into this issue?

[dgarroDC]

I can't debug this right now but after a quick look at the code, I think that this might be the problem, maybe we should also check that indexes if empty.

But anyway, can the is_num_var be simplified by just calling variable_type with the full token and verifying if the type is NUMBER? I mean, something like this:

bool is_num_var(string & token, compiler_state & state)
{
    return (variable_type(token, state) == vector<unsigned int>{1});
}

This is related to the bug we discussed here (is_num_var was returning true incorrectly), but when I fixed it (by replacing token with var_name), I removed that false positive but introduced this new false positive. In conclusion, it was okay to use token, the problem was all the other logic (dealing with indexes). We can remove all that logic and use token again, and the resulting function is much simpler.

This all also applies to the text case, of course.

[Lartu]

You are right, I do like the simpler function. We should take that approach.

[dgarroDC]

Fixed!

[Lartu]

Wonderful! Thank you very much!