Oops, the last one is in broken mime format. I will send it
>>>> In <yosug099l2d2.fsf(a)jpl.org>
>>>> Katsumi Yamaoka <yamaoka(a)jpl.org> wrote:
Ben changed font-lock.el in the main trunk of CVS nine days ago.
It causes a serious problem to write Japanese text using Egg-its
Though I am not good at English, I should explain more details.
;; Thanks Stephen for the advice. :-)
- (cond ((not (eolp)) (setq end (1+ end)))
+ (cond ((/= end (point-at-eol)) (setq end (1+ end)))
The reason why `(not (eolp))' is wrong in this situation is:
When we write a Japanese text using Egg, we will put a hiragana
text in advance of a kanji text. Then we will convert it to a
kanji text (choose one among some candidates) and fix it. Here
is a sample picture for that work:
If such a kanji text is inserted in the end of a buffer, the
hook `after-change-functions' will receive the positions of the
beginning and end of the range of a kanji text as arguments.
So the argument `end' points to the end of a buffer, in other
words, it points to `eol'. However the cursor is in the
beginning of a kanji text at that time. It is a rough
explanation how an error is occurred.
Katsumi Yamaoka <yamaoka(a)jpl.org>