Bill Spitzak

I believe the issue here is that they actually made directories called "F:", etc, on the Linux machine, so that paths can be used completely unchanged, assuming the cwd is...

There is absolutely a requirement that \xNN turn into the BYTE with the value 0xNN. Otherwise it is impossible to store arbitrary bytes as strings in JSON5. This makes it...

I believe it is somewhat random whether \xNN turns into the given byte value or into a pair of bytes (first one \xC2 or \xC3), depending on the JSON library....

Yes, making an array of numbers is one of the possible solutions, however that is extremely user-unfriendly when 99.99999% of the strings actually are readable text. The reason to allow...

Yes it would be a good idea to disallow \xNN for bytes that are easily encoded another way, for security reasons. This is a problem right now even if you...

That sounds like what I was complaining about. I do not want to have to write all the filenames as Base64 just because 0.00001% of them have some invalid UTF-8...

I'm now not sure I understand what you are saying. What I want is the sequence "\xNN" to turn into the byte with the value 0xNN when a quoted string...

`std::to_chars` prints the shortest possible sequence of digits that accurately represents the floating-point number being encoded. You can't do that with `printf` the best you can do is `%.18g` which...

Seems excessivly complicated. I would prefer something that translates each "disallowed" byte into a small pattern, such as "__XX" where XX is the hex code, and turn '_' into '_5F'...

The transcoding should only make changes necessary to make a valid USD identifier. It should not be making unnecessary changes such as changing valid Unicode characters that USD allows.