Page 1 of 1

kernel level UTF8 support?

Posted: Sun Aug 30, 2015 2:59 pm
by mercury
I look around the source code, it all in ansi.
I'm wondering if it possible make the visopsys kernel use UTF8.

And fontutil seems does't support UTF8 or double-byte characters.

I want use Source Han Sans in visopsys with simplified and traditional Chinese, Korean, and Japanese characters.
Also need a IME for typing characters.

Last year I made a patch for Menuet OS(32bit) to display double-byte characters.
It's really trouble to write ASM. At last I did it. But Menuet OS doesn't work fine for me.(Open a minesweeper needs 30seconds+)
KolibriOS is good, but I don't know how to patch. (it changed a lot from Menuet OS, and more complex)

BTW the registration question is funny. But what is "Mars Bar", it can be eat?!?!

So, I'm wondering if it possible make the visopsys kernel use UTF8.

Re: kernel level UTF8 support?

Posted: Mon Sep 07, 2015 9:56 am
by andymc
Hi Mercury, welcome.

The upcoming 0.8 release has better character set support, but it's not UTF-8.

Internally in the kernel, and in the new keymap and fontutil programs, glyphs and keyboard mappings will be stored using Unicode, but these are converted to 8-bit charsets according to the language chosen (ISO-8859-15 for western European, ISO-8859-5 for Russian, ISO-8859-9 for Turkish, etc).

Unfortunately, that probably doesn't help you with Chinese and the rest. I haven't really looked into UTF-8 yet, is it difficult to make that switch?

A Mars bar is a chocolate bar, yeah! ;-)

Re: kernel level UTF8 support?

Posted: Sat Sep 12, 2015 2:33 pm
by mercury
UTF-8 characters have different length.
Look here https://en.wikipedia.org/wiki/UTF-8

And need a program to read pcf, ttc or other format font file. Some font file encode in many different code page.
libiconv can convert Unicode, UTF-8 and many other code page. Is that possible port to visopsys?

Re: kernel level UTF8 support?

Posted: Wed Sep 16, 2015 10:15 pm
by andymc
mercury wrote:UTF-8 characters have different length.
Look here https://en.wikipedia.org/wiki/UTF-8
Yes, I'll need to think about what we do with multibyte characters.
mercury wrote:And need a program to read pcf, ttc or other format font file. Some font file encode in many different code page.
libiconv can convert Unicode, UTF-8 and many other code page. Is that possible port to visopsys?
I created Visopsys's VBF format to serve a particular need for bitmapped glyphs, but sparsely mapped, and minimal file size.

The new format coming in 0.8 is slightly different, but the idea is the same. The font can contain any number of glyphs, corresponding to any Unicode (32-bit) values, in any order.

So, until we have a proper vector fonts engine, this format suits our needs pretty well, I think?

Re: kernel level UTF8 support?

Posted: Wed Sep 16, 2015 10:35 pm
by andymc
But, right now, the Visopsys font code kind of assumes a .vbf file contains an 8-bit character set. We'd have to modify that part.