Mark wrote:
> a while ago I had a strange problem with mysql and java servlets. The
> servlet would not pull up accented characters and umlaut-characters
> correctly when they came from the database. I finally figured out that
> I had to set the MySQL-JDBC driver to use ISO88591 rather than UTF8
> (which was the default). The default came from a system property, and
> so setting "LANG=en_US.iso88591" in my tomcat startup-script fixed the
> problem as well.
And presumably the database was storing stuff in ISO8859-1. So that's
not too surprising... (I don't know if there's any better way inside
MySQL to handle this).
> Now I had some other problem with displaying man pages. One server
> shows it fine, a newer FC3 server screws up doublequotes and other
> characters.
Was this manpages from packages supplied by the Fedora project or
elsewhere? If they come from Fedora-supplied RPMs, then they need
bugzilla entries.
> So here is my question:
> /etc/sysconfig/i18n says:
> LANG="en_US.UTF-8"
> SUPPORTED="en_US.UTF-8:en_US:en"
> SYSFONT="latarcyrheb-sun16"
>
> What will happen if I change this to LANG=en_US.iso88591 ??? Will this
> break tons of other stuff on my server?
It's what you might expect. You'll start working with stuff that expects
ISO8859-1, but you'll stop working with stuff that expects UTF-8. The
good news is that practically all software that handles UTF-8 does so by
using the standard libraries which can also handle ISO8859.
It's the documents, and some of the communication systems, where you
might have problems. You might well have some documents encoded in
UTF-8, and ssh (for example) doesn't have any way for the client or
server to work out what character set the other expects. So if you SSH
from a ISO8859-1 client to a UTF-8 server (or vice versa), you might
have problems.
In my experience, practically everything in Fedora is happy with
ISO8859-1 at the moment (although I haven't tried too much with it:
mainly server-type stuff).
What would concern me, though, is that you're already having to store
umlaut characters and accents. What else are you likely to have to
store? Is there any chance you'll have to store Euro signs? More unusual
accents? Greek or Russian or Chinese names?
As you've discovered, handling multiple character sets is no fun. If you
run into ISO8859-1's limitations hard enough, you may find you have to
go back to UTF-8.
Hope this helps,
James.
--
E-mail address: james | It is difficult to produce a television documentary
@westexe.demon.co.uk | that is both incisive and probing when every twelve
| minutes one is interrupted by twelve dancing rabbits
| singing about toilet paper. -- R. Serling
