Saturday, October 11, 2003
My thin client needs a sandwich. Extra mayo, please.
I just spent about 25 minutes trying to figure out why this blog was malfunctioning. When the page would come up, it would at first be correct, and then suddenly resize itself to be too small to see many of the posts.
I knew I had caused this problem myself, as I had been goofing around with it on Friday. I spent a little while trying to back out my changes, which had the effect of making the problem intermittant, which was decidedly not an improvement.
Finally I just did a "view source" into a local copy and started ripping and tearing. I came to the conclusion that the links to the stat and directory services that I had placed in my sidebar (where the links are) were making the render of that div take longer, and for some reason IE would resize what it thought of as the entire page size when that render was complete, making the browser only show enough vertical space to fit in the links and images in the sidebar. Although I still don't have any idea why IE would do that, I just worked around it by putting all those links in the main column, at the bottom.
Almost five years now, I've struggled with HTML/browser GUIs. It's not getting better. There was a time in the world of interface design when you could depend on things to work. Everything worked differently, but it worked better. If I had seen a raw Windows or OS/2 API or Swing GUI do something like that, I could almost always find out why. Nowadays, with "thick clients" being almost hysterically out-of-fashion, we're left with HTML and a browser. Increasingly, the process of working with these things is that you just assume that there are ten million spec errors, odd spec implementations, and flat-out bugs that live under the covers, and to carry that analogy one step further, you just do your best to sleep on top of them.
The worst part is that I'm really starting to stick out in my chosen profession for not just ignoring this problem. Hasn't anyone noticed that after these last 5 years of churn, the most basic problems involved in this stateless, view-after-view model of application design remain unsolved? Go on out to TheServerSide and read a little. I love that site, but for the last bunch of years, every day there has brought the announcement of 5-15 new projects and/or products. All of them can be classified as attempting to solve one of the four or five basic underlying problems involved in writing web applications (security, session management, presentation, etc.). What does it tell us that nobody agrees on the solution to these problems? What does it tell us that the various user groups that spring up around each of the implementations of these solutions resemble nothing so much as congregations.
I acknowledge the reality that the web is a fixture now, and that this status makes it extremely difficult to affect change in the basic infrastructure. But all the same, after five years I still can't bring myself to drink the kool-aid on this one. The dream of a universal client is a wonderful one, but we're far from the promised land on this score, and web browsers, I am increasingly convinced, are just not ever going to get there.
I knew I had caused this problem myself, as I had been goofing around with it on Friday. I spent a little while trying to back out my changes, which had the effect of making the problem intermittant, which was decidedly not an improvement.
Finally I just did a "view source" into a local copy and started ripping and tearing. I came to the conclusion that the links to the stat and directory services that I had placed in my sidebar (where the links are) were making the render of that div take longer, and for some reason IE would resize what it thought of as the entire page size when that render was complete, making the browser only show enough vertical space to fit in the links and images in the sidebar. Although I still don't have any idea why IE would do that, I just worked around it by putting all those links in the main column, at the bottom.
Almost five years now, I've struggled with HTML/browser GUIs. It's not getting better. There was a time in the world of interface design when you could depend on things to work. Everything worked differently, but it worked better. If I had seen a raw Windows or OS/2 API or Swing GUI do something like that, I could almost always find out why. Nowadays, with "thick clients" being almost hysterically out-of-fashion, we're left with HTML and a browser. Increasingly, the process of working with these things is that you just assume that there are ten million spec errors, odd spec implementations, and flat-out bugs that live under the covers, and to carry that analogy one step further, you just do your best to sleep on top of them.
The worst part is that I'm really starting to stick out in my chosen profession for not just ignoring this problem. Hasn't anyone noticed that after these last 5 years of churn, the most basic problems involved in this stateless, view-after-view model of application design remain unsolved? Go on out to TheServerSide and read a little. I love that site, but for the last bunch of years, every day there has brought the announcement of 5-15 new projects and/or products. All of them can be classified as attempting to solve one of the four or five basic underlying problems involved in writing web applications (security, session management, presentation, etc.). What does it tell us that nobody agrees on the solution to these problems? What does it tell us that the various user groups that spring up around each of the implementations of these solutions resemble nothing so much as congregations.
I acknowledge the reality that the web is a fixture now, and that this status makes it extremely difficult to affect change in the basic infrastructure. But all the same, after five years I still can't bring myself to drink the kool-aid on this one. The dream of a universal client is a wonderful one, but we're far from the promised land on this score, and web browsers, I am increasingly convinced, are just not ever going to get there.
