This should have been a first class component of Boost for many, many
years now. If we do this - this whole mess goes away.
[snip]
stuff in Detail module that must be available to users, so the name Detail
is obfuscating here.
again all addressed by the above solution.
And better yet, we now have the opportunity to resplace this component with a purportedly more correct solution which is header only. This would make things that much easier. We haven't moved on this because because we're trying to get all the bugs fixed for this release. I would hope we can address this when we have the current monkey of our backs.
In short we need to:
a) fix this. b) get the release done c) make utf8_codecvt facet a first class boost component d) consider the proposed replacement
Robert Ramey
There already is header only utf8 codecvt facet there. So (c) and (d) are done. http://lists.boost.org/Archives/boost/2015/10/226245.php It would be released in boost 1.60 but I suggest that all develop branches that use old and broken utf8 facet to switch to the new one right now. Rationale of having it in Boost.Locale (a) It was already there (also wasn't header only) (b) Facets are part of C++ standard localization library (c) It is partial case of generic codecvt of any-encoding to wchar_t/char16_t/char32_t I don't really understand why this message regarding utf8 codecvt in Boost.Locale was ignored by Boost community. I hope it didn't went directly to spam. Artyom Beilis