
12 Jul
2017
12 Jul
'17
12:10 a.m.
On 12/07/2017 11:52, Peter Dimov wrote:
Having said that, this would basically render it impossible to have value + exception *without* an error_code.
Why would one ever want that?
People using expected<T, exception_ptr>, for one. Perhaps those people are out of scope for Outcome, though. But as long as those people are allowed to pretend that error_codes don't exist (ie. it picks an appropriate non-default error code for any exception) then they probably won't care.