bug#64202: bug#66188: 29.1; Include Atom feed reader

[James Thomas via "Bug reports for GNU Emacs, the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>]

James Thomas wrote:

> Daniel Semyonov wrote:
>
>>>>>>> James Thomas writes:
>>
>>     >> If the posts appear in a chronological order, it should end up
>>     >> being correct.  I think what's happening is that the feed you're
>>     >> testing actually uses a reverse chronological order for posts.
>>     >> If not, send me the feed, I might be misunderstanding something.
>>
>>     > Hmm.. It seems to be happening that way for every server I used:
>>
>>     > git.savannah.gnu.org/cgit/emacs.git/atom/?h=master
>>
>>     > rss.slashdot.org/Slashdot/slashdotMainatom:Slashdot
>>
>>     > github.com/emacs-mirror/emacs/commits.atom
>>
>> These feeds (except for the second one which I can't access) are in
>> reverse-chronological order.
>
> Right. I was hoping that that could be made the default.
>
> But I think the patch is fine to be included as it is; from my 1-2 weeks
> of testing.
>
> Cheers,
> James

This doesn't seem to work for some feeds such as:

en.wikipedia.org/w/index.php?title=Special:RecentChanges&feed=atom
thehackernews.com/feeds/posts/default

In the first one above, is the colon character to blame?

Another thing: I've been using your branch also with my addition below
for assuming reverse-chronological as a default (Sorry! But it was
really bugging me! :-)).

diff --git a/lisp/gnus/nnatom.el b/lisp/gnus/nnatom.el
index 2fcf59c1441..9f2b9c359aa 100644
--- a/lisp/gnus/nnatom.el
+++ b/lisp/gnus/nnatom.el
@@ -64,14 +64,16 @@ nnatom--read-feed
                     (when (eq (car data) 'top)
                       (setq data (assq 'feed data)))
                     (dom-add-child-before data auth)
-                    (catch :stop ; Collect feed authors, stop at first entry.
-                      (dolist (child (cdddr data))
-                        (let ((tag (car-safe child)))
-                          (if (eq tag 'entry)
-                              (throw :stop data)
-                            (and (or (eq tag 'author)
-                                     (eq tag 'contributor))
-                                 (dom-add-child-before auth child)))))
+                    (let ((prev (cddr data)))
+                      (while-let ((next (cdr prev))
+                                  (child (car-safe next))
+                                  (tag (car-safe child))
+                                  (_ (not (eq tag 'entry))))
+                        (and (or (eq tag 'author)
+                                 (eq tag 'contributor))
+                             (dom-add-child-before auth child))
+                        (setq prev next))
+                      (setcdr prev (nreverse (cdr prev)))
                       data)))))))
 (defvoo nnatom-read-feed-function #'nnatom--read-feed
   nil nnfeed-read-feed-function)


My case for such a change:

- Reverse-chronological makes sense as a default because a feed link
  only guarantees to provide the latest updates, not an unbroken chain
  of all events since the beginning.

- nnrss.el does it that way: see the 'nreverse' in line #643.

- Performance impact should be minimal because nreverse is O(n) and
  perhaps having to sort by date afterward (rather than simply use the
  existing number) would more than compensate for that.

- Running -catchup-to-here (on a datetime-sorted Summary buffer) will
  only have to record a single article number range in the newsrc.

(My code is probably not as beautiful as many would like it to be)

Would like to know your thoughts.

Regards,
James