Question

Trying to learn boost spirit and the example given in the docs have me a little confused.

Referring to this code:

http://www.boost.org/doc/libs/1_46_1/libs/spirit/example/qi/roman.cpp

Particularly this segment of grammar:

        start = eps             [_val = 0] >>
            (
                +lit('M')       [_val += 1000]
                ||  hundreds    [_val += _1]
                ||  tens        [_val += _1]
                ||  ones        [_val += _1]
            )

Could someone explain to me why it is +lit(‘M’) and not *lit(‘M’). Because after all can’t there be zero or more M’s versus one or more M’s?

Answer 1

Both +lit('M') and *lit('M') are correct. But the former is more readable than the latter (semantically), in my opinion, as the former says add 1000 to _val if there is one match, and do it repeatedly. On the other hand, the latter is difficult to read, as one could read it as add 1000 to _val even for zero-match which is wrong. 1000 is not added to _val for zero-times match, yet the parser *lit('M') seems to match for zero-match as well (seems kind of confusing).

So +lit('M') is preferable.

---

Alright. I read your comment. CCLLIX is not a valid roman number. What do you think its value is? 309? If that is so, then what value would be for CCCIX? It’s too 309, and its correct. Yours is wrong. Hence the parser stops when you use *lit('M'). Note also that the parser would also stop even if you use +lit('M') for this wrong input.