Unicode support

[kiith-sa]

I added (limited) Unicode support (need to support diacritics in my project).

Changes:

• API:
  • string passed instead of const(char)[] for strings that will be rendered in the GUI
  • imguiInit() has an optional fontTextureSize parameter. Users of less common Unicode characters, especially CJK, may need to tweak these. (that said, imgui doesn't really support CJK anyway)

• Internals:

  • Font texture is 1024_1024 (1MiB) by default instead of 512_512 (256kiB)
  • The g_cdata array in gl3_renderer.d (always) takes up 1.25 MiB.
  • Dimgui draws all glyphs (present in the font) it can fit until it runs out of glyphs or font texture space. This is somewhat hacky as the user may need to increase texture size to draw some Unicode characters, but... no less hacky than the rest of imgui and specifying character count in the API would require less trivial error handling or slower/more complicated init (it can be done if needed, though; it just doesn't seem to be in the spirit of the library)

  The default font texture size is overkill to ensure there are no problems for ASCII, various latin characters with diacritics and Cyrillic, but e.g. CJK users would need to increase it.

  • For glyphs that are not present in the font, only one null glyph is drawn on the texture instead of drawing it all over the place separately for each glyph.

[ghost]

Hmm.. Forcing to use string means the user has to either allocate on each render or do an unsafe cast (char[] to string for example) if they want to render some frequently updated piece of text.

Since imgui is an immediate mode GUI lib it makes sense that it takes non-immutable data methinks.

[ghost]

IIRC imgui did have some unicode support somewhere, or maybe imgui is based on some fork which used to have unicode support. Maybe I'm remembering it wrong though. Anyway this looks like great work so far, thank you!

[kiith-sa]

The main reason I used string is that in D it's 'obvious'/self-documenting that string is UTF-8. But makes sense, const(char)[] is compatible both with string and char[], we just need to specify in the docs that it's UTF-8, not ASCII.

[kiith-sa]

Update:

• Changed the API back to const(char)[].
• Didn't add 'UTF-8' everywhere in docs because it seemed to be too much noise.
• Instead I added a label/value using Unicode to the demo example.
• Also, the Unicode support code triggered a bug in imguiSlider which I fixed (an entire buffer, including (invalid) trailing characters, was formatted into a displayed string)

[kiith-sa]

(ping)

[ghost]

Woops! Sorry, I was busy with the whole "adjust to living in another country" period. :P

I'll get to your pull in a few days, promise!

[ghost]

Thanks for this work. If you want I could add you as a maintainer if you want to do further work on the project.

[kiith-sa]

It's likely I'll have more additions in the future but it'll be solely on a I-need-this-feature-now/it's-too-slow basis (the latter is pretty much certain), so I'm not sure if I'm the right person to be a maintainer.