[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Always use UTF-8 when running base-config?

[Tomohiro KUBOTA]
> I imagine you are thinking about using bogl-bterm also for the
> second stage.  Note that the UTF-8 support of Linux kernel itself
> is very limited and cannot used for "universal" installer.

Yes, something like that.  I am aware that the kernel UTF-8 support
only handle 512 different glyphs, and that this is not enough for
several languages.

> However, the Linux console's UTF-8 support can be used for a certain
> range of locales.  I don't know the exact range, but I imagine it
> should include Latin, Cyrillic, and Greek characters.  It cannot
> used for Asian languages.

My idea is to use the same terminal wrapper for all languages, to make
sure it is well tested, and to make sure we only need to include one
terminal wrapper for all languages, instead of several for different

> Bogl-bterm can display not only UTF-8 but also various encodings
> including ISO-8859-*, KOI8-*, EUC-*, and so on, when proper locale
> is installed.  However, it is very slow.  I don't know people can
> tolerate for it.

Is it equally slow when not using VMWare?  I've never tested it on
real HW.

> Jfbterm also supports various encodings and very fast.  I think
> jfbterm can be a candidate not for the second stage but also for the
> first stage.

How much would we need to include to get jfbterm working in the first
stage?  I noticed that it supports UTF-8 by using iconv.  If I'm not
mistaken, iconv uses large tables to convert charsets.  I hope we
don't need to include these on the floppy.

> The disadvantage of UTF-8 is that the second-stage softwares are not
> tested very well for UTF-8.  Especially, there are not very many
> developers who can test them with east Asian doublewidth characters.
> I don't know it is a good idea to switch to UTF-8 when we are going
> to freeze for the next release.

Me neither.  We will have to test it.

> BTW, how about using jfbterm instead of bogl-bterm for the first
> stage?  Jfbterm is much faster than bogl-bterm.  It supports UTF-8
> (without combining character support -- is it needed?).

Make a jfbterm-udeb, so we can test it. :)

Reply to: