master 9d0385d: Fix problem with 8bit content-transfer-encoding in nndoc mbox files

master 9d0385d: Fix problem with 8bit content-transfer-encoding in nndoc mbox files

[Eli Zaretskii]

> --- a/lisp/gnus/nndoc.el
> +++ b/lisp/gnus/nndoc.el
> @@ -352,6 +352,7 @@ from the document.")
>  	    nndoc-group-alist)
>        (setq nndoc-dissection-alist nil)
>        (with-current-buffer nndoc-current-buffer
> +	(set-buffer-multibyte nil)
>  	(erase-buffer)
>  	(condition-case error
>  	    (if (and (stringp nndoc-address)
> 

I think you want to swap the order: first erase-buffer, and then
set-buffer-multibyte.  That's because the latter in a non-empty buffer
can be expensive, and tis actually waste of cycles if you are going to
erase the buffer.

Re: master 9d0385d: Fix problem with 8bit content-transfer-encoding in nndoc mbox files

[Stefan Monnier]

> I think you want to swap the order: first erase-buffer, and then
> set-buffer-multibyte.  That's because the latter in a non-empty buffer
> can be expensive, and tis actually waste of cycles if you are going to
> erase the buffer.

Indeed, IMNSHO `set-buffer-multibyte` should signal an error when the
buffer is not empty ;-)


        Stefan

Re: master 9d0385d: Fix problem with 8bit content-transfer-encoding in nndoc mbox files

[Lars Ingebrigtsen]

Eli Zaretskii <eliz@gnu.org> writes:

> I think you want to swap the order: first erase-buffer, and then
> set-buffer-multibyte.  That's because the latter in a non-empty buffer
> can be expensive, and tis actually waste of cycles if you are going to
> erase the buffer.

Yup, I've now done this.  Looking at the code, it looks like a possibly
offensive case of defensive programming -- that buffer will be empty
when the `erase-buffer' is called, I think.

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no