Now that we’ve had a fair bit of user feedback, I thought I’d add a bit of feedback of my own, explaining some of the design decisions and philosophy of FeedGhost.

Before I begin, a little note: I’m not trying to belittle or ignore any comments that have been sent. Also these are my thoughts and opinions on the matter; Lee’s (and yours) might well be different. Additionally, these views as expressed in FeedGhost might well change; we’ll go where customers point, and furthermore, we might well split the FeedGhost Windows application into a more complex management tool, and a simpler headline reader.

I always wanted the FeedGhost experience to be IJW – “it just works”. I’m not a big fan of a user controlling software like some complicated machine, I think that 99% of the time, the software should just silently do its job properly, and the user not even notice it’s there.

To take an example, when you delete a blog article, there’s no UI for (directly) viewing those deleted articles. Is this a problem? I don’t think so: if you’ve made a mistake, you can hit undo and it will come back, and if you want to see that article again, it’s there, hiding in the background, indexed and available if you search for it. Is this a benefit? Obviously I think so: what could have been a complex bit of UI for navigating deleted articles has become two very simple controls: an undo button, plus a search box.

In fact, here’s a radical idea, and one which we’d almost certainly put in place in a simple headline reader: I don’t think we even need the tree or the list. Here’s why:

• They’re an expression of hierarchical data: folders to feeds to articles. The web’s pretty much moved away from that structure – when was the last time you used a web directory for example? (Actually there is one last bastion of hierarchy: shopping sites. Maybe that’s because their search isn’t up to scratch). Articles can be presented via search criteria: show me some unread articles; show me articles about blah; etc.
• They’re necessarily very complex, but have nothing to do with reading or managing blogs. They’re a generic tool when maybe a tailored solution would be more appropriate. Folders? Just categorize and add it to your search. Grouping? Tell the search what order you like your feeds in.
• They give an overview of the data, but not a useful one: when you subscribe to hundreds of feeds, a list or tree isn’t showing you the big picture anyway, it’s a scrolled window onto that picture.

Imagine a feed reader with no management functions. Would it matter if feeds accumulated that weren’t updated? The application could just update them less often. How about feeds you no longer want to read? A button on the article you’re reading could remove it. Perhaps you could even mark feeds as liked or disliked, so that if you didn't finish reading all your blog articles, it wouldn't matter: you would have read your favourites.

Of course, life and applications aren't that simple, and our users do want to customize the application. Every choice we've made when coding the application has been just that, a choice, between several perfectly valid options. What should we do when a group of users desperately feel that we made the wrong choice?

• Ignore our users, and decide that we've made the right choice? We'll have made the most expensive customized application, suitable only for ourselves.
• Make the change? There's bound to be another group of users who feel strongly about the status quo.
• Add a checkbox, combobox or whatever to the options dialog? Eventually we'll have a huge, impenetrable dialog that no-one but the most hardened techies want to use.

I'm beginning to think we should add some kind of scripting capability to FeedGhost. One of the reasons I'm a Firefox user is that the browser provides the basics, then I add the extras that I want, and I'm not cluttered with stuff I don't want.

A positive review here, and with a really nice idea: using FeedGhost to serve up bite-sized chunks of news as a way of sweetening your working day.

I can imagine myself doing that: "I'll do the washing up... then treat myself to three bites. Next I'll fix that nasty bug... and another three bites."

OK, OK, I'm being melodramatic. I actually really rather like Visual Studio; my first proper job involved C++ coding in EMACS on Linux, and while I'm sure things have improved since then, the most impressive coding enhancement there was syntax highlighting. I remember the first time I started Windows development and met VB6; I felt half impressed (at the IDE) and half disgusted (at the language).

Anyway.

Sometimes VS really, really, really annoys me. In particular, there are some scenarios you just can't debug.

Firstly, why are breakpoints keyed by name? ie, if I place a breakpoint, why do I get phantom breakpoints on every file with the same name?

Secondly, why does VS just lock up, particularly when debugging Windows Forms code? And if it can detect the situation, why can't it send the error report, rather than asking me to do it? I don't know what to include in the report, but Visual Studio has all the information right there.

And finally, once all this has happened, why does Visual Studio just give up entirely and fail to do anything even remotely useful? In a managed environment I don't expect the debugger to be telling me that things have got so bad that I essentially have no 'this' pointer.

OK, rant over. It's still a dozen times better than MFC. I just hope these issues get ironed out in the service pack.

An interesting idea from Rands in Repose, so here's my entry:

Blatant advertising, or just that FeedGhost is my life? I can't decide.

The new FeedGhost.com is now up and running, and looks pretty smart, if I do say so myself. I developed it in Firefox and IE7, and always knew I'd have to make some tweaks for IE6 -- boy was I unprepared for the devastation that appeared. I'd hate to have to spend my working life acccounting for other people's shoddy browser, and I'm guessing that's what web developers do.

Thankfully, it's pretty much sorted now. IE6 users will have a slightly degraded experience -- no smooth corners, some odd layout issues, and no font resizing. Any developer in a similar situation might want to use this:

http://tredosoft.com/Multiple_IE

Given that it took me a couple of hours last night, to make the adjustments for a tiny tiny website, makes me realise this: IE6 isn't a free browser. Let's say you visit a shopping website: the extra web-dev costs have to be passed on, and they'll end up being passed on to you. Perhaps shopping sites could charge an "old browser premium", rather like small shops do with small credit card purchases. That would encourage people to make the switch.