During the last week or so, we've been discussing some lexical_cast-related improvements.
However, when improvements starts to divert from the library they're intended to improve, and/or demand non-trivial changes, it may be the case that they constitute a new library.

The proposal I made for adding a non-throwing lexical conversion was fairly simple. Peter Dimov brought up the need for to_string and to_wstring functions, which is closely related and could be implemented in terms of that lexical conversion - but only if the conversion was fixed to handle strings correctly (and lexical_cast could of course use the to_string implementation). Some interesting ideas came up as to how the signature should look and how the function(s) should work.

The conclusion seems to be that the functionality is considered to be useful. However, I'd like some input on whether the changes should be minimized, and possibly added to lexical_cast, or if I (we) should take the input from the discussions, go back to the drawing board and incorporate the changes/improvements in a new library and ask for a formal review.

Any thoughts?
[Kevlin, as the father of lexical_cast, what do you think?]

Bjorn