A few weeks ago, the list got a following note, but no one replies to it
From: Chia-I Wu <b90201047@...>
Date: Sun Oct 30, 2005 3:32 am
Subject: ***SPAM*** [BUG] bug in syntax/css.vim
When I set LC_CTYPE to zh_TW.Big5 and opn a HTML, vim gives me these
Error detected while processing /usr/share/vim/vim64/syntax/css.vim:
E16: Invalid range
E475: Invalid argument: cssIdentifier "#[A-Za-z?-?][A-Za-z?-?-9_@-]*"
I guess this is because the big5 charset doesn't contain some of the
characters list in the bracket.
When I set LC_CTYPE back to C or undo the change
then it's OK again.
I am running debian unstable + vim 6.4.
I'm also getting the same annoying message from vim when I set a locale
environmental variable such as LC_CTYPE to ja_JP.eucJP, ja_JP.sjis, or
something like that.
The problem lies in the regexp-like definition of cssIdentifier which
contains a grave A and a umlaut y.
Perhaps, European users might not notice what such innocuous characters
are doing on the other side of the earth, particularly when they come
accross some exotic encodings related to zh_TW or ja_JP.
In my own case where encoding=euc-jp, the combination of a grave A and
and a hyphen constitutes a Chinese character which stands for sex. :-()
On the other hand, the combination of a umlaut y and a 0 (zero) becomes
an empty space. (Hey, what did you expect?)
That's funny, anyway. But I'm now getting bored of it. So I wish
someone could fix it.
Attached is a tentative patch I use to fix the problem. I do think
it's not a good solution, and never hope it should be included into
the official vim runtime source. Rather, my intention of sending my
poor patch is to clarify what is the problem and to suggest a quick
fix for it. Based on it or not, I hope someone could find an elegant,
sophisticated solution to it in near future.