On 5/8/20 10:22 AM, Eli Zaretskii wrote:
>> They all seem to display fine in eg
>> https://www.gnu.org/software/emacs/manual/pdf/emacs-xtra.pdf
> Sebastian seems to say that is not the case?
The original problem report was about the Texinfo input “Bahá'í” (with U+0027
APOSTROPHE), which Texinfo renders as “Bahá’í” (with U+2019 RIGHT SINGLE
QUOTATION MARK) in info and as something similar-looking in TeX.

The quick patch was to change the Texinfo input to “Bah@'{a}@t{'}@'{i}”; the @'
calls were to suppress false-alarm diagnostics and I don't know why the @t{'}
was put in.

It now appears that there's no need to worry about the false-alarm diagnostics
(the output is fine, as Glenn noted). Also, the @t{'} is not necessary in that
it puts one wrong character into the output (namely, U+0027) instead of a
different wrong character (namely, U+2019).

The preferred Unicode spelling these days is “Baháʼí” (with U+02BC MODIFIED
LETTER APOSTROPHE). However, although Texinfo can handle that character and puts
it into info files, the character doesn’t survive transliteration to TeX (it
gets lost). I don't know whether this is a bug in Texinfo or in TeX, but anyway
we need to work around it if we're going to use the correct spelling.

I suggest that we use the preferred spelling in info files, and fall back on the
previous behavior of using apostrophe (which TeX renders like the right single
quotation mark) when outputting to TeX. I’m attaching a proposed patch, which I
emailed to Eli earlier before seeing this bug report. This patch also fixes
other instances of similar problems that I found.