Good products aren’t written under insanely tight time constraints.
Good products have some form of written specification.
Good products have an architectural model

I’m so glad I decided to take my time with Joggle and try and do it in a way that seemed sensible instead of just knocking out code that would _work_, but wouldn’t be very _good_. This is the state of the current product we’re making at work, now I have to bend it to my will to produce websites. “Impossible!” I hear you cry! And you may be right.

In other news, a Java blogging tool based around the design concepts of Bloxsom, blojosm and I’d forgotten all about the “Most popular friend of friends” page.

I changed my template on the 7th of March, anytime you want to publish it, Blogger, feel free…….
13th March: And about time too.

The Wildgrape NewsDesk is good. Very good, and it looks better than Syndirella, too. But here’s why I won’t be making the switch:

  • It comes automatically set up to a host of feeds I neither want nor care about. This should be an option, not a default.
  • The feeds it does use mostly come from, even when there are perfectly good ‘official’ versions (e.g the feeds for Boing Boing, BBC news, and others).
  • I’ve got very used to Syndirella’s address bar – it’s very useful when you want to send someone the link, or copy the URL into another window.
  • Its “update all” function is single threaded (according to the NewsDesk site). Syndirella does this multi-threaded, and certainly seems faster for it.
  • There’s no user feedback when adding a feed, removing a channel etc. so that it looks like NewsDesk has crashed. It would be polite to display a “downloading feed” or “removing channel” message whilst NewsDesk is still working.
  • It appears to have some memory issues. I subscribe to about 25 RSS feeds, Syndirella uses up about 11 MB RAM to refresh them all, NewsDesk uses a consistent 35 MB.
  • It’s slightly too harsh in parsing invalid XML, for example, at the moment even Zeldman doesn’t parse correctly (using a screenscraper, feed here), and when something’s gone wrong, it lets you know by making the icon for the feed red, but doesn’t tell you why (timeout, invalid feed, whatever)
  • As soon as you select a feed item, it marks it as read. Instantly. A timed response is much better behaviour. Added to which, unread articles aren’t as obvious as in Syndirella. NewsDesk notes them by placing an icon next to them, which disappears as soon as you click the article title. Syndirella displays unread article titles in bold, and drops the bold once the title has been selected for a few seconds.
  • NewsDesk doesn’t let you view the source for a feed. Not an immediately obvious requirement (why would you need to see the source?), but it’s kind of nice, and can be useful.
  • The biggest bug-a-boo of all – no RSS auto-discovery. This is the biggest selling point of any aggregator now, if you’ve got autodetection and a usable interface (standard three panes will do), I’m willing to overlook a lot of other bugs so long as you provide this.

But I do like the way it breaks feeds up into Channels, and has good import/export abilities (although Syndirella’s “import feeds from FeedReader” is a stroke of genius – simple, yet effective), and the way you can validate feeds just by right clicking. Also, the alphabetical ordering is a godsend, something that Syndirella could really learn from.

I’ve yet to give NewsMonster a try, but by all accounts it seems like a step too far for me; I just want a plain RSS reader (I’ve never used the “web feeds” that Syndirella lets you use), not a whole browser addon/mega-app, I want to be able to actually read Mark Pilgrims’s site (read the controversy), and anyway, I’m getting kind of bored of XUL chromes – platform specific UI, please.

My company has a new website. None of the usual web developers were involved. As a result, it doesn’t validate, is poorly-formed, uses tables for layout, has unclosed attribute values, and uses the <font> tag.

It just doesn’t get much worse. It doesn’t use animated gifs, but it _does_ have a <marquee> tag on the front page. And the navigation on the front page is different to the rest of the site.

The guy who wrote it owns Jakob Neilsen’s Designing Web Usability. His code isn’t even valid HTML 4 Transitional, and even Frontpage can produce that these days.

So depressing.