3 ms·
#4 is double-counting, since that's a special case of #3. In any case, these are all concerns for a decoder, but not for an API, which is what we're discussing
by mikeash 9y ago
#4 is double-counting, since that's a special case of #3.
In any case, these are all concerns for a decoder, but not for an API, which is what we're discussing here. In fact, the original comment I replied to up there was advocating the opposite: UTF-16 internally, and UTF-8 for interchange!