...

Seems perfectly normal to me. You need omit[] around stuff that you don't want to appear as part of the attribute for a rule. Since your rule's attribute is only a byte, including big_dword as part of the rule without omit would mean that its part of the attribute for the rule too and that doesn't match the layout of the test struct.

Hi,

What's wrong with the following code, and how can I fix it? The semantic actions seem to be inconsistent (or maybe I'm using them wrong). No omit, but no value?

Semantic actions in Karma are supposed to _supply_ the attribute value to generate output from. Therefore your start rule needs to be written as: start = big_dword >> big_dword[_1 = ref(temp)]; Regards Hartmut --------------- http://boost-spirit.com

2011/5/1 Hartmut Kaiser Semantic actions in Karma are supposed to _supply_ the attribute value to generate output from. Therefore your start rule needs to be written as:

   start = big_dword >> big_dword[_1 = ref(temp)];

I believe it was the Qi part (use %= instead):

    start %= big_dword >> big_dword [ref(temp) = _1];

Doh! Sorry for that. I should have read your code more carefully. Regards Hartmut --------------- http://boost-spirit.com