1 Jun
2017
1 Jun
'17
1:53 a.m.
Mere moments ago, quoth I:
Simplifying expected
into result<T> and outcome<T> makes sense, and the design of error_code_extended seems reasonable, including the content and ring buffer history.
Another related point that I forgot to mention: The default ring buffer size seems too small, especially for multi-threaded use. The docs specify that it's possible to enlarge it but I worry this could be problematic if not consistently sized between multiple translation units.