Re: BUG: Japanese UTF-8 files > 65 Kbyte get corrupted during saving

Attached is a small lisp file "bug.el" to reproduce the problem. Execute it as follows: mfabian＠gregory:~/xemacs-bug$ LANG=C xemacs -q -batch -eval "(require 'un-define)" -l bug.el Wrote /tmp/utf-8-output mfabian＠gregory:~/xemacs-bug$ LANG=C iconv -f utf-8 -t euc-jp /tmp/utf-8-output > /dev/null iconv: illegal input sequence at position 65535 mfabian＠gregory:~/xemacs-bug$ and you see that one Japanese character in the output has been destroyed. This problem seems to occur for any big Japanese files in UTF-8. When saving them, characters will be destroyed at file positions close to N * 65535 bytes.

;;; -*- coding: utf-8 -*- ;;; xemacs -q -batch -eval "(require 'un-define)" -l bug.el (require 'un-define) (setq number-of-characters 22000) (defun reproduce-utf-8-bug () (interactive) (with-temp-buffer (set-buffer-file-coding-system 'utf-8) (let ((i number-of-characters)) (while (> i 0) (insert "あ") ; <--- insert Japanese hiragana 'a' in UTF-8 (setq i (1- i)))) (write-file "/tmp/utf-8-output"))) (reproduce-utf-8-bug)