On Monday, May 02, 2011 07:12:14 PM Paul A. Bristow wrote:
Does not make much sense to me either - BUT
OTOH, the functionality exposed by convert is much needed and has to make it into Boost somehow.
Strongly agree. We are missing out while trying to find the ideal solution.
So I'd favour accepting Boost.Convert, but being prepared to fairly quickly deprecate/supersede it in favour of
* A new version of lexical_cast which does take more arguments.
I strongly disagree here. The boost.convert interface is broken by design! It should not be accepted as it is right now. I explicitly did not give a "conditional yes", because I believe that the library will be completely different after fixing it. You apparently waited so long now. Does another half a year (or let it be a year) really make a difference?