Hacker News Reader

Comment by m4lvin | Hacker News Reader

I am still surprised that the two-character icons which Zoom uses for accounts without profile pictures become "A�" for anyone called "Alice Ørland" or similar.

It is even more surprising that Ö, Š, etc. become �, but 文军 shows up fine.

View on HN

Not so weird when you consider the context. Zoom is an North American company with most (if not all) of it's product/engineering development happening in China. Just reading public information, it seems they have no development teams in the North of Europe nor Eastern Europe, so it's only natural that some character sets ends up better supported than others.

It's likely that all the characters are being transferred and stored in the same charset (probably a Unicode one) regardless of what character is entered. The replacement glyph (the question mark) might be caused by their use of a font with incomplete Unicode coverage, or (more likely) by a buggy "take the first character of each word in their name" algorithm, for example using the first byte and then special-casing Chinese under the mistaken impression that nothing else needs more than one byte.

From someone who doesn't know a lot in that area of software, thanks for the (possible) correction, learned something new today!

I agree, probably mishandling of combining characters

Yeah, that would work as an excuse back in ANSI days.

we have UTF8 as a standard nowadays.

Unicode doesn't magically make this stuff go away, much less any specific encoding of it. A glyph can consist of multiple codepoints, and then those can sometimes be standardized to other codepoints. For instance, Unicode has codepoint U+00C4 for an A with a diaeresis (aka umlaut). But it also has codepoints U+0041 U+0308 for an A with a combining diaeresis, which should then map to the combined U+00C4 for font rendering.

and both cases should be handled properly, as it is in Unicode standard.

Specifically for that example, Zoom is written and maintained in China, so I would expect the developers made special code for segmenting chinese names but not non-English latin ones.

Are these characters strictly speaking "Latin"? Would they be categorised as "Nordic" or "Cyrillic"?

Standard "Latin" characters to me are those that are supported on a Latin keyboard, and provided for by the latin codepage.

They are all Latin according to Unicode 1.0

https://en.m.wikipedia.org/wiki/Latin_script_in_Unicode

They're Latin characters with diacritics. They're not usually treated as separate letters.

Reply on news.ycombinator.com