Strings are one character short with version 3.2.5 through 4.0.0 about utfcpp

[nemtrif]

See #111 Basically a regression was introduced when fixing another issue.

Your options are:

1. upgrade to 4.0.3
2. Instead of using utf8::unchecked::utf16to8(), use utf8::utf16to8(). It is safer anyway and just a bit slower.

[afh]

Thanks for the quick response, @nemtrif, very much appreciated 🙏
I'll talk with the other devs about (dis)advantages to using utf8::utf16to8() or maybe we can find a way to switch to it only for the affected releases of utfcpp.

I'd like to suggest to add a notice or warning to each the affected releases on its GitHub release page. Maybe something along the lines of:

This release introduces a breaking regression, please use at least 4.0.1. For details see #111.

[nemtrif]

No problem. My assumption they are using unchecked namespace to avoid exceptions. If so, you should better upgrade to 4.0.3. Or, just stay on a pre-3.2.5 release. 3.2.4 is OK.

[afh]

Ledger will update to utfcpp 4.0.3 (ledger/ledger#2315 is pending review), yet some distributions, like Debian, prefer programs to link to libutfcpp rather than embed it. In the case where a distro ships libutfcpp 3.2.5 a sensible path forward is to apply a patch replacing uses of utf8::unchecked::utf16to8() with utf8::utf16to8() before compiling the program for the distribution package.

Your comments have been very helpful! 💯🙏