Yet another terminal/iterm umlaut irritation

Question

Hi all
I'm back again with umlaut issues.
However I'll begin by describing the set up.
I have GNU bash, version 3.2.17(1)-release (powerpc-apple-darwin8.9.0) installed.
I also use vim version 7.1b.1 BETA
My .bashrc contains the following
PATH=$PATH:/usr/local/bin/:/Users/duffy/'Java Arkiv'/Glassfish_Arkiv/glassfish/bin/
export LCALL=svSE.UTF-8
export LANG=sv_SE.UTF-8
export JAVA_HOME=/Library/Java/Home
export TERM=xterm-16color
alias mysql=/usr/local/mysql/bin/mysql
alias vim=/opt/local/bin/vim
alias vimdiff=/opt/local/bin/vimdiff
SHELL=/opt/local/bin/bash
my .bash_profile contains
if [ -f ~/.bashrc ]; then
source ~/.bashrc
fi
my .inputrc contains
set input-meta on
set output-meta on
set convert-meta off
set meta-flag on
My terminal settings are as follows
I login with usr/bin/login
use xterm-color
screensettings -> utf-8 encoding, wide characters count as 2
And in my .vimrc I have set the encoding to utf-8
That about does it for my set up.
Now - here goes.
Terminal command window displays umlauts correctly, and appears to display umlauts correctly when typing. If I do a ls on a directory and see a file called lets påsk.txt the file name looks fine. If i type ls på* I get No such file or directory, however if I mark the file name and do a cut and paste in terminal, i.e I cut and paset "på" and paste it into ls then I get ls på* and it works fine. In other words the keyboard input and terminal output display doesnt seem to be the same. I cant figure out why this is. On top of that, when I start vim I can type å ä ö but everytime I do it uses two spaces, so the word påsk looks like this på sk, thsi despite the fact that the terminal uses utf-8 and vim encoding is set to utf-8.
Maybe I've missed some info here, but I've pretty much put all the info in here I think is relevant.
I await words fron the grea Guru out there.

Answer 1

If i type ls på* I get No such file or directory
In HFS+, which is the MacOSX's default file system, accented characters in file names are stored in "decomposed form". For example, "å" is stored as "a" + "COMBINING RING ABOVE". The file names output by the ls command are decomposed:
ls p*.txt | od -t x1 -c
If you enter "å" from your keyboard, on the other hand, it is in pre-composed form, i.e., a single Unicode character:
echo å | od -t x1 -c
If you type "ls -w -l påsk.txt", the "å" is pre-composed and bash passes this pre-composed form to some system call (such as stat(2)). But the system call seems to internally convert it into decomposed form, so it works.
If you type "ls -w på*", on the other hand, bash gets filenames (which are decomposed) in the current directory and compares them with "på" (which is pre-composed), resulting in no match. A workaround is to use "ls -w pa*", i.e., without the accent.
I also use vim version 7.1b.1 BETA
Where did you get this? I guess it has been built without the multibyte support. In vim, type
:version<Return>
and search for a string "multi_byte". If it has "-" in front of it then multbyte support is not included in the vim.
The ones in
http://macvim.org/OSX/index.php (vim7.0) or
http://code.google.com/p/macvim/ (vim7.1)
will support multibyte.
Or you can build vim7.1 by yourself (if you have Xcode Tools installed).