[pg]

Sort of yes, sort of no. It's a rapid prototyping technique. Essentially you fix it case by case, by taking individual bits of code that use this technique and replacing them with the uglier and less flexible but more efficient alternative of a hard-coded url.

[bkmartin]

Paul, I really do not mean any disrespect here because you are truly a class act and first rate player in the start up world. You are also a great hacker that loves to push the limits. You've created an amazing community here that I have been able to learn a ton from.

I have to ask, and I'll probably get down voted to hell because I'm naive or something, but what is so elegant about a coding technique that breaks under normal usage conditions? If I put out a customer facing piece of code, especially after 4 years, wouldn't it make sense to use an "uglier and less flexible but more efficient alternative" that doesn't break?

I understand your previous explanations of why this happens and of rapid prototyping etc. But at what point does the architecture actually get changed to eliminate this bug?

[pg]

It doesn't make sense to call any specific amount of traffic "normal conditions."

What's good about this technique, and about rapid prototyping in general, is that you can write an initial version quickly in very little code, then gradually make it more efficient as the demands on the app increase.

The rate of expired links says more about how busy I personally have been lately than about the desirability of storing state in closures.

[bkmartin]

I'm not referring to any specific amount of traffic. I'm referring to how users expect a website to work. If the user sees a link, especially a More or Login link then the user expects it to do just what it says. When those don't work I would call that a bug. I'm in agreement that this technique can be useful for rapid prototyping, but I also think this site is probably the most active and mass used prototype I've ever seen. ;)

My goal for a web site or web app is to have 0 expired links. Sometimes stuff you link to outside your site will go dead, and it must be fixed or removed or whatnot. But for your own internal stuff... I don't know... something doesn't feel right about an architecture that allows that systematically. How much time could you save if you didn't even have to worry about fixing any expired links? Any idea on what the ROI on your time would be?

Anyway, just thinking out loud. Thanks again for the site though. I do indeed enjoy it very much regardless.