I mean the same image for different Unicode codes and not vice-versa.
And we mean the opposite, indeed. I linked the wikipedia page about Han Unification which explains this very clearly, and give a table with many examples of characters that use the same Unicode codepoint, but are expected to look different in different languages.
If it helps you to understand, let’s take the letter A as an example:
Both of these are the letter A and the same unicode codepoint. But they are rendered with different fonts, and a completely different glyph. Well, it happens that this is not a problem for latin languages, we are used to both forms and will read text using either just fine. However, this isn’t as much the case for the “CJK” characters: while the character is indeed the same, it is drawn very differently in different countries.
Also the Code2000 font you mentionned is not under an open source license, is several versions of Unicode behind, does not provide such language specific variants, and was not updated for 10 years. So it wouldn’t be a suitable replacement.
This is why the Noto font is so large: for some characters it has to include not just one, but sometimes 4 or 5 variants of the character. But if we want to provide support to all of our users worldwide, this is required.
And this is just for CJK, there are other languages where the mapping between characters and their on screen rendering is a lot more complex. Sometimes the characters and the rendering are not in the same order. Sometimes they combine together (such as ´ + e = é). In the latin alphabet you can also think of ligatures (some fonts include a specific rendering for “ffi” which is not just putting the images for f, f, and i together, for example). Again, latin users are used to typewriter text now and we have greatly simplified our writing and lowered our typography expectations when working with computers, but with other alphabets this sometimes isn’t the case.
So, no, your font are not indexed by Unicode codepoints only. They are indexed by codepoint AND language. Which is what I’m saying several times now: you can’t get away with just a fontname and some unicode string and expect the right thing to happen. You need to specify a language.
