Cross LFS and UTF-8

Alexander E. Patrakov patrakov at ums.usu.ru
Wed Jan 18 20:33:10 PST 2006


Jim Gifford wrote:

>
>> The main valid reason to use UTF-8, IMHO, is compatibility with 
>> RedHat. The problem is that LFS already has interoperability problems 
>> with the latest RedHat distros: e.g., you just can't share files with 
>> them over NFS and have non-ASCII filenames readable on both computers 
>> (because RedHat doesn't support anything else than UTF-8 and LFS 
>> doesn't support UTF-8). And LSB is (unfortunately) on RedHat's side.
>>
> So this is a Redhat bug not a LFS related issue. If Redhat/Fedora 
> should be compatible across the board. Has anyone opened up a trouble 
> ticked with them?

Informally speaking, that's indeed a RedHat bug. But if you submit it, 
they will certainly say that you should stop using your buggy non-distro 
that doesn't support UTF-8 locales, just "because UTF-8 is the future". 
Add UTF-8 is required for NFS version 4.

Note that the LFS intention is not to force everyone use UTF-8 locales. 
The intention is to support both variants without two sets of 
instructions. In non-UTF-8 locales, the code added by the UTF-8 patches 
is just dead code that never runs.

And in fact, both questions "how to upgrade my existing system to use 
some UTF-8 based locale without starting the build from scratch" and 
"what to do if I don't want to use UTF-8 locale" were asked. I think 
that the right way is to have one common set of build instructions (the 
current ones) and two sets of configuration instructions. This will 
completely avoid the first question, and have a simple "see the 
non-UTF-8 configuration section" answer to the second.

> See a lot of people were under the impression that utf-8 was to 
> support different locales, the example above is only to prove to 
> people that locales can be supported without utf-8.
>
You are completely right, UTF-8 is in the book not due to the need to 
support various languages. The exact reasoning must be in the book, in 
order to prevent such misunderstanding.

Locales (non-UTF-8 ones, like de_DE) were already supported by LFS and 
CLFS. But some people (e.g., those who often translate texts between 
German and Greek) wanted to have de_DE.UTF-8 instead of de_DE. I think 
that both variants should be supported in the same book (if one ignores 
BLFS issues).

-- 
Alexander E. Patrakov



More information about the cross-lfs mailing list