BUG: <foo>[0:] is not working

[joszamama]

No description provided.

[andreas-zeller]

See also #88. Slices, as in [0:], may have a lower priority at this point, but we should have a more reasonable error message.

[joszamama]

I am testing this problem currently, and it does not crash, but fandango is unable to find a solution. See this example:

Grammar:

<start> ::= <int> <int> ;
<int> ::= "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" ;

str(<start>[0:]) == "6" ;

After running FANDANGO for 100 iterations, we are unable to find a solution, but there is no crash. Also, I have checked manually, and the input "66" (which in this case, should be the solution) has a fitness of 0. The problem is not in the parser itself but in the constraint handling. I will keep working on this problem.

[andreas-zeller]

The question is what <start>[N:] should be returning - but neither option can be "6". <start>[N:] could be

1. A list of all its children (since <start>[0], <start>[1], etc. would return the 1st, 2nd, ... child). Then, <start>[0:] would return a list of [<int>, <int>] whose string representation would be, say, "[6, 6]".
2. A string representation of the first characters starting at position N (since we treat symbols such as <start> as a string in other contexts). Then, <start>[0:] would return, say, "66".

I think the interpretation (1) is more in line with Python expectations (if we indeed make <start>[N] return the n-th child as a tree, which I'd recommend).