aKregator in CVS

[Frerich Raabe]

On Tue, Dec 07, 2004 at 04:35:41PM -0500 or thereabouts, George Staikos wrote:
>    Furthermore, akregator uses its own librss right now because various bug 
> fixes have been applied.  However I'm sure that there are fixes in 
> kdenetwork/librss that aren't in akregator's too.  I think these fixes need 
> to be merged somehow.

I have yet to be told about any bugs which are in kdenetwork's librss, other
than the "bug" which makes it use kio_http's cache (and there has been
endless discussions about whether one should do that or not, I don't really
consider it a clear technical bug).

On the other hand, the 'librss' which aKregator forked handles much more than
RSS - it works with Atom-style feeds as well. Since this is entirely beyond
the scope of librss (it's not called 'libsyndication' for a reason), the
way Atom support was added in a way which is very error prone and as it is
right now I am not terribly fond of the idea of adding that to kdenetwork's
librss (which is what knewsticker and dcoprss use).

I suspect kdenonbeta/akregator/src/librss/README doesn't say "Please DO NOT
report any bugs about it to Frerich, since he most probably did not introduce
the found bugs." for a reason. ;-)

>    Finally, knewsticker and akregator solve the same problem from two 
> different approaches, but I think akregator is much more useful on a 
> day-to-day basis for people who use RSS heavily.

Is tempting to compare the two given how much of the backend functionality
they share, but such comparisons are flawed, and so is yours. The two
applications have entirely different use cases: knewsticker is meant as a
monitor for ticker-like things like stocks, sport scores, headlines, stuff like
that. It also is meant as a very non-obtrusive addition to the desktop. The
newsticker works best when you run it all the time.

Of course you can run aKregator all the time as well, but it's really something
you fire up in the evening and then check all the various news sources at once.
It also is more suited for blogs, since those often transport a lot more text
in their RSS files (as the <description> for an item, usually).

I think the newsticker and aKregator are very similar in terms of what
information is provided (hence stuff like sharing the list of feeds makes
sense), but very different in the way the information is presented (and
assumed to be accessed).

> The problem I have is dcoprss.  I am -really- annoyed at having to 
> constantly kill stray processes 
> ( http://bugs.kde.org/show_bug.cgi?id=68905 ), and I'm also not convinced 
> about how well dcoprss would match with akregator's design yet (I have found 
> knewsticker to be very unreliable due to the dcop service, but akregator to 
> be rock-solid). 

KNewsTicker does not use dcoprss at all, but if you could elaborate a please
on how exactly you found it to be unreliable, I might be able to tell you the
actual reason.

> I think there should be a strong effort to merge these 
> backends or make them compatible somehow.

Yes, this is a worthwhile goal IMHO. I've been brainstorming about merging
the backend facilities (read: fetching stuff using librss, maintaining
a list of newsfeeds) into a 'libakregator' library, and then merging
knewsticker into aKregator, so that it still seems to be an independant
application, but really only is an aKregator in applet mode. The
ticker-specific settings could be added as an additional page to akregator's
configuration dialog.

For what it's worth, I just remembered that akregator already copied the
filtering code from knewsticker, so that is prolly something which could
be shared as well.

- Frerich