The class boost::date_time::special_values_parser has a method named likely. However, I am using this class in code that uses likely/unlikely as branch-prediction macros (see How do the likely/unlikely macros in the Linux kernel work and what is their benefit?). Is there a way to still use this code while also using the macros?
Asked
Active
Viewed 144 times
0
jhourback
- 4,381
- 4
- 26
- 31
1 Answers
4
Is there a way to still use this code while also using the macros?
Yes. Function-like macros require () immediately to expand, regular functions don't. Workaround:
(boost::date_time::special_values_parser::likely)( /* ... */ );
Aykhan Hagverdili
- 28,141
- 6
- 41
- 93
-
That's a good way to call it, but unfortunately, just including the header file for `boost::date_time::special_values_parser` conflicts with the macro. – jhourback Mar 11 '21 at 18:15
-
1@jhourback include the headers *before* defining macros. – Aykhan Hagverdili Mar 11 '21 at 18:16
-
That could work, but it's tricky because the header is included by other headers, so I would have to mandate a specific order of header inclusion in other headers. (i.e., include the header that defines `likely` only after you include the header that includes `boost::date_time::special_values_parser`). – jhourback Mar 11 '21 at 18:19
-
3@jhourback defining lower case macros with no prefix is the decision that led you to a name conflict, and this solution tries to mitigate that. The actual solution is to give the macros ugly names like `MY_LIB_LIKELY`. – Aykhan Hagverdili Mar 11 '21 at 18:20