[lug] How can I test a codepage module?

D. Stimits stimits at attbi.com
Fri Aug 23 12:16:22 MDT 2002

rm at fabula.de wrote:
> On Fri, Aug 23, 2002 at 09:22:26AM -0400, Michael Hirsch wrote:
>>On Thu, 2002-08-22 at 21:10, D. Stimits wrote:
>>>If you logged in to a regular console, and used a text-based web browser 
>>>(e.g., lynx), and went to a web page using that character set, maybe it 
>>>would show up correctly. I'm not sure if using an X11-based browser 
>>>would depend on the same support, maybe it would use other fonts.
> I don't think this will actually test codepages, D. Aren't you mixing
> up codepages and locales here? Linux doesn't need any codepage support
> per se - locales are handled entirely on the application level (glib
> takes care of that, or at least it should :-)
> Codepages are only needed for (broken?) filesystems such as the
> FAT and VFAT filesystems of DOS.

Yes, that was just a guess on my part. However, I think a locale 
requires a code page to display correctly if the character is not from 
the local code page. Code page and native language support work 
together...I do not think the character set of a native language can 
work completely correct if the code page for it does not also work. All 
of this is under filesystem support, and many parts of filesystem go 
beyond the physical partition (/dev/stdout and friends are "files"). I 
have no idea if it really extends this far, but browsers in console mode 
should in theory respond to character set in <META http-equiv> 
statements, and if the character set is not available, it wouldn't 
display it correctly. Code page is not directly related, but I would bet 
that kernels with the right code page cannot work unless character set 
works first...a broken character set should exclude the code page from 
working, although a broken code page might not break the character set. 
[or so goes the idea; none of which necessarily applies in a graphical 

D. Stimits, stimits AT attbi.com

>>Thanks, that should work.  I found another way.  For some reason smbfs
>>wants these code pages when it mounts a windows share.  I could generate
>>error messages by mounting share.  After installing the modules the
>>messages went away.
> That's a good test - and the reason you need codepages in your kernel ...
>   Ralf Mattes

More information about the LUG mailing list