NTSerializer only encodes in ASCII

[tcmitchell]

The NTSerializer only encodes in ASCII and issues a warning if the encoding is anything other than "ascii". See nt.py line 22 and nt.py lines 27-28. See also #1376

The N-Triples Grammar says:

An N-Triples document is a Unicode [UNICODE] character string encoded in UTF-8.

This may be a change from earlier (pre-2014) versions of N-Triples. See the first bullet in Section 3 Changes from RDF Test Cases format. But this change is 7 years old.

Example 3 in Section 2.3 RDF Literals shows unicode characters in an example N-Triples document.

There is a mention of ASCII encoding in Section 6.1 Other Media Types but that is for passing N-Triples documents as text/plain. Section 6.1 goes on to say that "In ... these cases the document is not an N-Triples document" (emphasis mine).

[white-gecko]

This shall be solved with #1222

[tcmitchell]

Aha! More source diving revealed NT11Serializer, which is not limited to ASCII. Why isn't this the default N-Triples serializer? Why stick with the ASCII-only, and ancient, NTSerializer?

See #1376 (comment)