To get UTF-8 to work when sending drafted messages in Wanderlust with Emacs 23.1 I had to apply this patch to FLIM.

I also checked out the wl-2_15_6-vtl branch in the wanderlust repository and the semi-1_14-vtl branch in the semi repository. (See repos for info about the repositories.)

Someone once said something along the lines that to those in the Western world who don't speak Japanese there is an alternate universe of FOSS [1] development in Japan [#china], and every so often things from that universe just appear in the Western FOSS universe as mysterious foreign objects. Ruby seems to have made a solid connection between the universes, but many others less approachable.

Wanderlust development and the development of the emacs packages that it depends on (APEL, FLIM, and SEMI) feel like that to me. Some of them have web pages, but they're usually quite out of date. It can be difficult to their CVS repositories, and even more difficult to find the browse-able web interfaces to their CVS repositories. Every time I want to look at the code, I have to figure this all out again, hence this post.

There was a post on the Wanderlust English-language mailing list from Katsumi Yamaoka <yamaoka@jpl.org> on Wednesday, 27 May 2009 that gave the location of the APEL, FLIM, and SEMI cvs repositories, and mentioned that development in the FLIM and SEMI repositories was taking place under the flim_1_14 and semi_1_14 branches, respectively. The post also said that the place for discussion about APEL is the APEL mailing lists:

apel-en@m17n.org

apel-ja@m17n.org

To subscribe, send an empty e-mail to apel-en-ctl@m17n.org and/or apel-ja-ctl@m17n.org.

The place to discuss FLIM and SEMI is the EMACS-MIME mailing list:

emacs-mime-en@m17n.org

emacs-mime-ja@m17n.org

To subscribe, send an e-mail containing “subscribe YOUR NAME” in its contents to emacs-mime-en-ctl@m17n.org and/or emacs-mime-ja-ctl@m17n.org.

So, here are the CVS commands I just used to check the various packages out:

cvs -d :pserver:anonymous@cvs.m17n.org:/cvs/root checkout wanderlust
cvs -z9 -d :pserver:anonymous@cvs.m17n.org:/cvs/root checkout apel
cvs -z9 -d :pserver:anonymous@cvs.m17n.org:/cvs/root checkout -r flim-1_14 flim
cvs -z9 -d :pserver:anonymous@cvs.m17n.org:/cvs/root checkout -r semi-1_14 semi

Here are the browse-able CVS web interfaces:

• http://cvs.m17n.org/viewcvs/root/wanderlust/
• http://cvs.m17n.org/viewcvs/root/apel/
• http://cvs.m17n.org/viewcvs/root/flim/
• http://cvs.m17n.org/viewcvs/root/semi/

If you happen to be using Windows Emacs and Cygwin Python there's an annoying interaction where “M-x run-python” hangs. The easiest way to fix it is to mount the directory where you have your emacs installation into the Cygwin file name structure. For instance, I have my emacs installation in C:\emacs\emacs-22.2, and in a cygwin shell I did mount -b ‘C:\emacs' /emacs, and now running the Cygwin python works.

The emacs function run-python adds the emacs data-directory to the PYTHONPATH in the emacs process-environment before running python; unfortunately, since this is the MS Windows emacs, its data-directory starts with a drive letter and a colon. When the Cygwin python initializes sys.path it splits PYTHONPATH at the colons, which means sys.path ends up with the drive letter as one component (usually interpreted as a relative path) and everything after the colon as another component. It that's a valid Cygwin pathname for the Emacs data directory (which is what the above mount command did), things work (accidentally).

It turns out that if you execute the command xterm-mouse-mode (or evaluate (xterm-mouse-mode 1) in your initialization file) when running Emacs under Screen it allows “non-modified single clicks” to work. Normal mouse functionality is still available by holding the Shift key while clicking. I use the PuTTY ssh client for remote access to various servers, and this works well Emacs in Screen under PuTTY, too.

I found a very useful Emacs minor mode today: linum.el, which puts line numbers at the beginning of all the lines in a buffer.

Ok, suppose you want to evaluate a particular bit of code after emacs loads a particular emacs-lisp file, but you want to pass values of local variables into that code. The function eval-after-load makes you quote the expression and doesn't allow passing values into the expression. How about this?

(require 'cl)
(defmacro* eval-after-load* (file varlist &rest body)
  "Like `eval-after-load', but bind variables according to VARLIST in
the current environment of the `eval-after-load' expression, not the
environment when BODY is evaluated.  This allows easy passing of values
into BODY.
Each element of VARLIST is a symbol (which is bound to the current value
of that symbol) or a list (SYMBOL VALUEFORM) (which binds SYMBOL to the
value of VALUEFORM in the environment of the `eval-after-load' expression."
  `(eval-after-load ,file
     '(let ,(loop for v in varlist
                  collect (if (symbolp v)
                              `(,v ,(eval v))
                            `(,(car v) ,(eval (cadr v))))
                  into new-varlist
                  finally return new-varlist) ,@body)))
(put 'eval-after-load* 'lisp-indent-function
     (1+ (get 'eval-after-load 'lisp-indent-function)))

Here's a contrived example which demonstrates when things happen.

(let ((f (make-temp-file "tkb-madness" nil ".el"))
      (x 1))
  (unwind-protect
      (progn
        (save-excursion
          (let ((buf (find-file f)))
            (insert (format "(y-or-n-p \"In the file '%s'! \")" f))
            (save-buffer)
            (kill-buffer buf)))
        (y-or-n-p "This happens before the eval-after-load*")
        (eval-after-load* f
            (x
             (y (y-or-n-p "This happens when the eval-after-load* is executed?"))
             (z 2))
          (y-or-n-p (format "x: %d y: %S z: %d" x y z))
          (y-or-n-p "This happends during the delayed expressions"))
        (y-or-n-p "This happens after the eval-after-load* expression")
        (load f))
    (when (file-exists-p f) (delete-file f))))

You should see something like:

Wrote /tmp/tkb-madness88647vuE.el
This happens before the eval-after-load*(y or n)
This happens when the eval-after-load* is executed?(y or n)
This happens after the eval-after-load* expression(y or n)
Loading /tmp/tkb-madness88647vuE.el (source)...
In the file '/tmp/tkb-madness88647vuE.el'! (y or n)
x: 1 y: t z: 2(y or n)
This happends during the delayed expressions(y or n)
Loading /tmp/tkb-madness88647vuE.el (source)...done

(y-or-n-p is used instead of message so you see each message when it happens.)

Does the eval-after-load* macro make sense?

Ever use the emacs command describe-char? It's even more fun with proper unicode lookup data!

;; First, we'll bind it to a key.
(global-set-key "\C-cD" #'describe-char)

;; Now we'll download it if necessary.
(let ((udf-url "http://www.unicode.org/Public/UNIDATA/UnicodeData.txt")
      (udf-dest "~/tmp/UnicodeData.txt"))
  (if (file-readable-p udf-dest)
      ;; Let describe-char know it exists.
      (setq describe-char-unicodedata-file udf-dest)
    ;; It doesn't exist, and we need to download it!
    (when (y-or-n-p (format "You need to download %s ! Do it? " udf-url))
      ;; Really weird: wget -O 'file' complains that file doesn't exist.
      (let* ((cmd (format "cd ~/tmp/ && wget -O %s --progress=dot '%s'"
                          udf-dest udf-url))
             (buf (get-buffer-create (format " *wget '%s'*" udf-url)))
             (proc (start-process-shell-command "wget-unicode-Data"
                                                buf cmd)))
        (display-buffer buf)
        (set-process-sentinel
         proc
         `(lambda (proc event)
            (unless (string-match "^finished" event)
              (error "unexpected status '%s' getting '%s'" ,udf-url event))
            (setq describe-char-unicodedata-file ,udf-dest)
            (message "Try describe-char now! ☣☥☸▧◉✘✽☮⅙▧⚅☑☢☹☺♠♥♦♣♨♻⚔")))
        (message "Downloading... check describe-char later")
        nil))))

Once this is run and it tells you to try describe-char you can position your cursor over one of the Unicode characters in the message (“C-h e” to display the “*Messages*” buffer) and press “C-cD” and look for the “Name:” line. You'll see something like this:

      character: ♻ (299515, #o1110773, #x491fb, U+267B)
        charset: mule-unicode-2500-33ff (Unicode characters of the range U+2500..U+33FF.)
     code point: #x23 #x7B
         syntax: w    which means: word
    buffer code: #x9C #xF2 #xA3 #xFB
      file code: #xE2 #x99 #xBB (encoded by coding system mule-utf-8)
        display: terminal code #xE2 #x99 #xBB
   Unicode data:
           Name: BLACK UNIVERSAL RECYCLING SYMBOL
       Category: other symbol
Combining class: Spacing
  Bidi category: Other Neutrals