|--==> "KY" == Katsumi Yamaoka <yamaoka@jpl.org> writes:

  KY> Loading of rmail is only for the `rmail-select-summary' macro,
  KY> isn't it?  If so, it can be solved by not compiling that section:

  KY> --- gnus-util.el~	2003-12-03 23:28:53 +0000
  KY> +++ gnus-util.el	2003-12-05 00:44:47 +0000
  KY> @@ -864,8 +864,9 @@
  KY>  	      (narrow-to-region (point) (point-max))
  KY>  	      (rmail-count-new-messages t)
  KY>  	      (when (rmail-summary-exists)
  KY> -		(rmail-select-summary
  KY> -		 (rmail-update-summary)))
  KY> +		(dont-compile
  KY> +		  (rmail-select-summary
  KY> +		   (rmail-update-summary))))
  KY>  	      (rmail-count-new-messages t)
  KY>  	      (rmail-show-message msg))
  KY>  	    (save-buffer)))))

  KY> How about it?

So this would be as a replacement to Jerry's patch or in addition to
it? 

And the doc string from `dont-compile' troubles me a little...

,----[ C-h f dont-compile RET ]
| `dont-compile' is a compiled Lisp macro
|   -- loaded from "/usr/local/lib/xemacs-21.4.14/lisp/bytecomp-runtime.elc"
| (dont-compile &rest BODY)
| 
| Documentation:
| Like `progn', but the body always runs interpreted (not compiled).
| If you think you need this, you're probably making a mistake somewhere.
`----

...that last line in particular.

-- 
|---<Steve Youngs>---------------<GnuPG KeyID: A94B3003>---|
|              Ashes to ashes, dust to dust.               |
|      The proof of the pudding, is under the crust.       |
|------------------------------<sryoungs@bigpond.net.au>---|