For lack of content...

For lack of content...

[Michael Welsh Duggan]

There are some posts which I read which contain characters in
non-default coding-systems, which don't include a proper Content-Type
header.  As a result, these messages end up in my buffer as the raw
bytes of the coding method.  There are a couple of possible solutions
that I can think of:

1) A decoding alist of some sort, such as an alist of regexps
   (matching a newsgroup name) to a list of coding-systems.  If the
   visited newsgroup matches that regexp, then it should try to decode
   the body using the mehods in the associated list,m in order.

2) A washing method, for which the user can supply a coding-system
   with which to decode the article.  I actually wrote a prototype of
   such myself, but ran into the problem that decode-coding-region
   trashes text-properties.

If something already exists to cover this, I apologize for not having
found it yet.

-- 
Michael Duggan
(md5i@cs.cmu.edu)

.

Re: For lack of content...

[Lars Magne Ingebrigtsen]

Michael Welsh Duggan <md5i@cs.cmu.edu> writes:

> There are some posts which I read which contain characters in
> non-default coding-systems, which don't include a proper Content-Type
> header.  As a result, these messages end up in my buffer as the raw
> bytes of the coding method.  There are a couple of possible solutions
> that I can think of:
> 
> 1) A decoding alist of some sort, such as an alist of regexps
>    (matching a newsgroup name) to a list of coding-systems.  If the
>    visited newsgroup matches that regexp, then it should try to decode
>    the body using the mehods in the associated list,m in order.

Or group/topic params.  I've now added this.

> 2) A washing method, for which the user can supply a coding-system
>    with which to decode the article.

And this.

>    I actually wrote a prototype of such myself, but ran into the
>    problem that decode-coding-region trashes text-properties.

Hm.  It does seem that that function trashes text props.  Why?

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen

Re: For lack of content...

[Michael Welsh Duggan]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Michael Welsh Duggan <md5i@cs.cmu.edu> writes:
> >    I actually wrote a prototype of such myself, but ran into the
> >    problem that decode-coding-region trashes text-properties.
> 
> Hm.  It does seem that that function trashes text props.  Why?

Here's a segment of the discussion engendered by my original bug
report:

| From: Kenichi Handa <handa@etl.go.jp>
| Subject: Re: Text Properties in decoded regions (correction)
| To: rms@gnu.org
| CC: md5i@cs.cmu.edu
| Date: Thu, 27 Aug 1998 14:35:01 +0900
| 
| Richard Stallman <rms@santafe.edu> writes:
| > I asked Handa what he thinks we should do.
| > In principle, decode-encoding-region should preserve text properties,
| > but if that is hard to do, maybe it should delete them.
| 
| It's a rather hard task to preserve text properties with
| decode-coding-region and encode-coding-region.  At least, I can't do
| that before the release of 20.4.
| 
| So, I agree that we should delete text properties of the target region
| and document it.

(BTW, is there a good way in Gnus to do forwards of other articles in
replies to other articles, or cites from multiple articles?)

-- 
Michael Duggan
(md5i@cs.cmu.edu)
.

Re: For lack of content...

[Justin Sheehy]

Michael Welsh Duggan <md5i@cs.cmu.edu> writes:

> (BTW, is there a good way in Gnus to do forwards of other articles in
> replies to other articles, or cites from multiple articles?)

I tend to just process mark them all, making sure that the one I'm
actually replying to is the first one marked.  Then, R or F comes
close enough to what I need.

-Justin

Re: For lack of content...

[Michael Welsh Duggan]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> Michael Welsh Duggan <md5i@cs.cmu.edu> writes:
> 
> > 2) A washing method, for which the user can supply a coding-system
> >    with which to decode the article.
> 
> And this.

I tried it out today.  Very nice!  Only one minor complaint, and not
necessarily one you should "fix".  When using `C-u M W c', it would be
nice to have tab completion on the entire set of coding systems (as
per (interactive "z")) instead of the small subset.  (I am much more
used to typing old-jis than iso-whatever.)  Typing old-jis works, I am
just used to the tab completion.  I won't complain if you don't change
it however, as the smaller list of matches may well be worth it.

-- 
Michael Duggan
(md5i@cs.cmu.edu)
.

Re: For lack of content...

[Lars Magne Ingebrigtsen]

Michael Welsh Duggan <md5i@cs.cmu.edu> writes:

> When using `C-u M W c', it would be
> nice to have tab completion on the entire set of coding systems (as
> per (interactive "z")) instead of the small subset.  (I am much more
> used to typing old-jis than iso-whatever.)  Typing old-jis works, I am
> just used to the tab completion.  

Yup.  Fix in Pterodactyl Gnus v0.14.

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi@gnus.org * Lars Magne Ingebrigtsen