Pilcrows

To me, Tim Bray's experiments with anchors have been an interesting hack, but I doubt that I'll implement it. Maybe it's just my erratic mouse tendencies, but the latest version (with the mouseover effect) just looks like Mozilla developers are having fun with the <blink> tag's behavior.

— Gordon Weakliem at permanent link

Memes We Live By

Patrick Logan asks

What makes software developers think most other disciplines are "predictable, schedule-able, reliable, budget-able and robust"?

Personally, I think that it's because most software developers have never done serious work in other trades. It's one thing to build a doghouse, or even a garage, and come in ahead of schedule and under budget, and another thing to build a house, or a bridge. Big developers can build houses with reliable budgets because they build the same house over and over, but few of these houses will stand for a hundred years or more. And then there are the cases of homeowners suing the developer because, for example, the developer didn't account for expansive soils, and so none of the doors in the house shut properly (My wife and I have friends who went through this). And then we have contractors doing faulty repairs to bridges. The US space program is remarkable for the small number of fatalities, or even outright failures, that it's produced, but I think that engineering at this level is the exception, not the rule.

However, there are a number of other reasonable explanations, for instance, with base technology changing so frequently, the performance characteristics of our infrastructure changes so much that it's difficult to do reliable estimating and stay with current technology. For instance, look at the difference between writing ASP applications on IIS and ASP.NET; there are a number of places where the implementation differences are gigantic, and extremely meaningful. On top of that, our idea of a good implementation changes. 8 years ago, not many people considered good markup, or even valid markup, a high priority, but these days, I see lots of complaints about the quality of markup produced by tools such as ASP.NET. Experience with tag soup, accessibility, and various other concerns, has changed people's minds about this.

Somewhat peripherally, I have a friend who works as a project manager for a big developer who told me that tradespeople hate being asked to tear down something they've completed and rebuild it. Even paying doubletime doesn't help morale. I think most software developers share this attitude, but we have to balance that against the desire to improve our work environment. Still, the building trades have to deal with this; there are a large number of changes to building codes every year, and tradespeople are legally obligated to stay abreast of these.

— Gordon Weakliem at permanent link

More on IDs

I read Tim Bray's and Christopher Dent's responses to Mark's piece on atom:id. I also didn't like the argument that permalinks might not be permanent, in spite of the fact that I've broken my own permalinks at least once, I'm trying to change my ways. Chris seems to acknowlege this reality, you should strive to make permalinks permanent, as much as possible, but sometimes, it doesn't work that way. For the problem that tag: isn't a registered scheme, this was already hashed over in the debate over feed:, and it just doesn't seem to concern anyone enough to put an end to the idea.

— Gordon Weakliem at permanent link