nnrss doesn't use If-modified-since

[Dan Jacobson]

Problems with nnrss-version"nnrss 1.0" and or gnus-version"No Gnus v0.6"
 (Debian emacs-snapshot package, version 1:20060923-1)
using /usr/share/emacs-snapshot/site-lisp/gnus/nnrss.elc (which inside
says it was compiled from
$ wc</usr/share/emacs/site-lisp/gnus/lisp/nnrss.el
 1098  3794 36774)

Error messages getting covered up.
We do M-x gnus, and from the *Messages* buffer let's take
a look at all the important messages one misses as they all overprint,
leaving us looking at only the last.

Checking new news...
Opening nndoc server on /tmp/eraq...done
nnmbox: Reading incoming mail from file...
nnmbox: Reading incoming mail (no new mail)...done
nnrss: Requesting 台掃 - 最近更改 [zh-tw]...
Contacting host: localhost:8080
nnrss:
http://localhost/mediawiki/index.php?title=%E7%89%B9%E6%AE%8A:Recentchanges&feed=rss:
Not valid XML (error XML: (Not Well-Formed) Invalid end tag (expecting
LINK) at pos 1135) and w3-parse doesn't work (void-function w3-parse-buffer)
nnrss: Requesting 台掃 - 最近更改 [zh-tw]...done
Opening nndoc server on /tmp/d...done
Opening nndoc server on /tmp/m4m...done
Opening nntp server on localhost...
Denied server
Opening nntp server on localhost...failed #never mind this
Checking new news...done

Anyway, _error_ messages could be made to linger longer.

(Actually, the way it is now is what I want, as I am offline 99% of the
time: if I am not connected to the net, then don't tell me that you of
course failed.

But wait, tcpflow(1) shows what was sent:
GET http://localhosthttp://localhost:8080/ HTTP/1.1...
localhosthttp? why of course the response will be
HTTP/1.1 404 Not Found
or perhaps
HTTP/1.1 503 WWWOFFLE Remote Host Error...
which of course is not valid XML.

Can you believe doing
$ http_proxy=http://localhost:8080/ emacs-snapshot -f gnus
causes
GET http://localhosthttp://localhost:8080/ HTTP/1.1...
or at least that is what tcpflow shows.

$ unset http_proxy; emacs-snapshot -f gnus
works fine.

$ http_proxy=http://localhost:8080/ emacs-snapshot -f gnus
works fine only after one hits an additional gnus-group-get-new-news
(g keystrok.)

Also, observing repeated hitting of g (gnus-group-get-new-news) with tcpflow
shows that the same request and the same response are exchanged, with
_no use of If-modified-since_ etc. traffic reducing features!

Another problem, I am in the nnrss buffer looking at
  R  1|11月08   ?|MediaWiki defa:台掃:Site support
     1|11月09   ?|MediaWiki defa:台灣鐵路管理局
     1|11月09   ?|MediaWiki defa:台灣高速鐵路
     1|11月09   ?|MediaWiki defa:海岸巡防署
and I try gnus-summary-save-article. Well, it proposes a totally
garbled filename to save to. Not unicode ready.
Indeed
$ head -n 1 ~/News/rss/台掃*
;; -*- cod...ng: emacs-mule; -*-
whereas
$ head -n 1 ~/.bbdb
;; -*-cod...ng: utf-8-unix;-*-
But doesn't the rss Info page say
   Note: you had better use Emacs which supports the `utf-8' coding
   system because RSS uses UTF-8 for encoding non-ASCII text by default.
   It is also used by default for non-ASCII group names.
Well News/rss/* should use unicode too then.
Furthermore, mails I sent from that buffer have
X-Draft-From: ("nnrss:HORRIBLE MANGLED former UTF-8 [zh-tw]")
I do save the article, only to notice
Message-ID: <address@hidden>
which might not be a valid Message-ID, but I didn't check.
$ munpack -r eeee
$ validate part2
*** Errors validating /tmp/part2: ***
Error at line 1, character 1:  missing document type declaration; assuming
        HTML 4.01 Transitional

Furthermore, I am looking at a message,
  From: MediaWiki default
  Subject: 海岸巡防署
  Newsgroups: 台掃 - 最近更改 [zh-tw]
  X-Sent: 1 day, 6 hours, 21 minutes, 52 seconds ago

  <p>Importing text file</p> <p><b>新頁面</b></p><div></div>
[why do we see the HTML tags?]
  
http://localhost/mediawiki/index.php?title=%E6%B5%B7%E5%B2%B8%E5%B7%A1%E9%98%B2%E7%BD%B2
  
http://localhost/mediawiki/index.php?title=%E8%A8%8E%E8%AB%96:%E6%B5%B7%E5%B2%B8%E5%B7%A1%E9%98%B2%E7%BD%B2

but those last two lines need me to run
perl -MMIME::QuotedPrint -ne '$_=encode_qp($_);s/%/=/g;print decode_qp($_)'
on them to become comfortable reading. None of the "wash" menu items
apply.

I (set-variable (quote nnrss-wash-html-in-text-plain-parts) t nil)
But nnrss-request-article: Symbol's value as variable is void: charset

Also when in the Summary buffer, there is a lot more choices available
when clicking the "options" menu, then there are when clicking it from
the Article buffer.

Also I notice snownews gets many more items than gnus and there is no
way to change it, at least from reading the rss Info page.

Here we see that line 1 above is "link", line 2 "comments" but we are
not told this above:
  $ lynx -assume_charset=utf8 -force_html -dump part2
     Importing text file

     新頁面

     [1]link

     [2]comments

  References

     1. 
http://localhost/mediawiki/index.php?title=%E6%B5%B7%E5%B2%B8%E5%B7%A1%E9%98%B2%E7%BD%B2
     2. 
http://localhost/mediawiki/index.php?title=%E8%A8%8E%E8%AB%96:%E6%B5%B7%E5%B2%B8%E5%B7%A1%E9%98%B2%E7%BD%B2

Anyway, you can test all this on e.g.,
http://zh-classical.wikipedia.org/ or
http://radioscanningtw.jidanni.org/ 's recent changes RSS feed.