In the first he points out how good we are at ignoring history, using the BeanContext API as a good example. I wonder if this will influence the nano/picocontainer work.

The second is about protocol adaptors. Here's his musings on PyProtocols:

Now if you reflect a bit, you realize that a lot of menial work in programming covers the transformation of one type into another. It's just embarassing the amount of time spent doing this kind of activity. It makes so obvious the limits of Object Oriented resusability.

My aggregator is my portal to the blogosphere. If you disembowel your feed, you remove yourself from the world. Maybe Mark could have waited until there was actual an Echo serialization format before doing this.

The newspaper view is a win. It appears to be done via XSLT, so you could put back the outlining feature of Aggie's pixel skin. It has some of the same limitations as the Aggie HTML view -- bad HTML in the feed hoses the newspaper view. If you have too many items in a listing's newspaper view, the newspaper just omits them. Scoble disappeared midway because of this. Because the newspaper view is HTML rendered via XSLT, there are some UI things I'd like to do that you can't do from the newspaper -- I'd like to mark items read/unread from the newspaper. I'd also like to put items into a bin from the newspaper - now I have to switch to the feed and click the item, and as soon as I do that, the newspaper for the channel replaces the listing newspaper. So then I have to switch back to the newspaper view and find the place where I left off. Also the group by and filter boxes don't affect the newspaper view. I do like the filtering pulldown on the newspaper view, I just wish for a little more control over the criteria.

There are a bunch of issues related to the newspaper view and tabbed browsing. The biggest issue I have with the newspaper view is that right-click, open in new window creates a new tab (good), but that tab becomes the focused tab, which is bad. I'd prefer that tab to be in the background. If I have a bunch of browser tabs open, quit, and restart, they are forgotten. Sure wish it used Firebird as the tabbed browser.

Watches are a win. I haven't set any up yet, but I will if I decide to keep FeedDemon.

News bins are not as big a win as I thought they would be. I originally thought that they could help with the workflow (scan all feeds, noting those to be saved/posted/whatever). The problem is that you can stick items into bins from the newspaper view, which is where you're doing the scrolling.

I'm coming from Aggie, which has no notion of categories. I like the idea of listings as a way to categorize feeds. However, 150 items per listing is too small. Actually, most of the limits are too small. Also, the UI for selecting and managing listings is impoverished - something tree structured with drag and drop would really help.

Periodic polling doesn't seem to be working for me -- If I select as listing after the polling interval has passed, FeedDemon gets all the feeds. Shouldn't it be doing this quietly in the background and notifying me? This may be because I don't understand how polling aggregators work -- Aggie needs me to kick things off by hand. Better polling status/progress information would be nice. It'd be nice to know roughly when the next poll would be.

There appear to be big memory leaks. The commit charge in my system drops by a few hundred meg after I exit FeedDemon.

Some other little nits. It needs more keyboard shortcuts. There's just a little bit too much clicking - switch to a listing, scan down, right click to open new tabs, do the news bin click dance, ooops newspaper has too many items -- click to mark some items read so we can finish the newspaper, now click to mark the whole listing read. FeedDemon doesn't appear to support GZip encoding. Maybe I'll have to update the Gzip list...

I know that I've just nit picked a lot of stuff, but overall, my impression of FeedDemon is pretty positive. It's fast. The newspaper views appear to overcome a problem that I was having with Aggie, where I kept seeing posts that I had already read. Watches are a win, and maybe I'll get over news bins and find a good way to use them. Considering this is a 1.0 beta, I'm impressed -- I think that I'll be able to read more items faster with FeedDemon than with Aggie. Of course, I wish that FeedDemon was open source, so I could tweak it -- I would still be happy to pay for it to make sure that Nick Bradbury continues to improve it. Other things I'd like to see - support for Blogger / MetaWeblog - or a way to link news bins to something like w.Bloggar, so you could post out of FeedDemon. Also, CommentAPI support -- but that's all up in the air with Echo now. Dare, Torsten, Luke, Simon, Joe -- this is the one to beat.