Re: 19.34 --> 20.4, 5.6.43 --> 0.80

["François Pinard" <pinard@iro.umontreal.ca>]

> D> Do I have to set a language environment?  Should I run emacs with
> D> --unibyte or not?

Lars told a few times Gnus works better when --unibyte is not used.
I tried both, and both were workable for me.  Yet, in --unibyte, I was
unable to see much of those sexy characters, beyond French ones :-).

> D> I mostly use English and don't understand much about Mule.  I did
> D> install the intlfonts package.  What's a good font(set) to use
> D> for emacs?

I define these in .bash_login:

export EMACS19FONT='-adobe-courier-medium-r-normal--*-140-*-*-m-*-iso8859-1'
export EMACS20FONT='-*-fixed-medium-r-normal-*-16-*-*-*-*-*-fontset-standard'
export XEMACSFONT='-*-fixed-medium-r-normal-*-16-*-*-*-*-*-fontset-standard'

and use this code in .emacs:

(setq EMACS19 (string-match "\\<Emacs 19" (emacs-version))
      EMACS20 (string-match "\\<Emacs 20" (emacs-version))
      XEMACS (string-match "XEmacs\\|Lucid" (emacs-version)))

;[...]

(cond (XEMACS (set-face-font 'default (getenv "XEMACSFONT")))
      (EMACS20 (set-default-font (getenv "EMACS20FONT")))
      (EMACS19 (set-default-font (getenv "EMACS19FONT"))))

If you have a longish `.emacs' file (which is my case), I suggest you
execute the above code early, because it might change the dimensions of
your Emacs frame.  You can then reposition the frame with the mouse while
the initialization proceeds.

> D> 1) I run ispell from my message-send-hook, and it gave an error
> D> message:

> D> Ispell misalignment: word `JSa' point 266; probably incompatible
> D> versions

I got the same, at a time I was not using --unibyte.  I reported the
problem to appropriate people, and had no indication that the problem
was ever clearly identified.  My intuition was that it is a Mule related
problem, but this was not confirmed.  Recently, with a recent Emacs pretest,
I removed the --unibyte, and these `ispell' problems did not come back.
However, the conditions of my testing are different, as I now use the nice
Manuel Serrano's `flyspell' package (included in the pretest).

> D> locate-library shows that this is the ispell.el that comes with
> D> 20.4.  Our ispell binary is 3.1.20 which I installed just 7 months
> D> ago.  I guess this is unlikely to be a Gnus problem.

I'm not sure `ispell.el' fully knows how to do arithmetic in buffers with
all those internal-supposedly-invisible `\201' added in them.

> D> 2) I set message-send-hook to nil and tried again.  After `C-c C-c'
> D> I got an error: "Can't encode a part with several charsets."  Isn't
> D> Gnus supposed to break the message into parts automatically?

Yes, it should.  But recently, Gnus gave me this unexpected message as well.
I reported the matter to Lars.

-- 
François Pinard                            mailto:pinard@iro.umontreal.ca
Join the free Translation Project!    http://www.iro.umontreal.ca/~pinard