Re: minor bugs with negative article numbers (was: Patch To Fetch Messages From Google)

[Simon Josefsson <jas@extundo.com>]

On Wed, 10 Apr 2002, Dan Christensen wrote:

> This is a great feature!  While playing with it, here are some
> problems I noticed.  I hit ^ to make Gnus find a news article that had
> expired from one of my nntp groups.  Gnus used google to bring the
> article in, and the article looked fine, except that it had a few
> garbage characters at the bottom.  Then I used `B c' to copy the
> article to an nnfolder group, and there the article didn't appear
> in the summary because it was corrupted:  there is a garbage line
> in the second line of the header, and there is garbage at the end.
> I've attached the article as it appear in my nnfolder file.

This seems to be a URL bug:

(url-insert-file-contents  
"http://groups.google.com/groups?selm=1f8fd51.1sjfq9b1u3kq7yN%tells@nowhere.invalid&output=gplain")

returns a trailing ^M.

If I fetch the URL with wget, there is no ^M.

> Another minor comment:  Gnus won't let me `!' an article, but it
> will let me `d' and `E' it.  Not sure if this inconsistency is
> intentional.

I think so, the article doesn't have an article number so it is difficult
to mark.  The reason d and E work is probably that it already was marked
as read, so pressing d or E doesn't result in any action.

> And a final comment:  each time I view the article, Gnus downloads
> it again from google.  I guess avoiding this would be difficult?

The backlog doesn't handle non-integer articles.  In fact, the backlog 
doesn't seem get invoked at all when I fetch articles using nnweb.