Friday, November 28, 2008

URLs Should Be Forever

I know link rot is almost as old as the web. Still cool urls don't change! Now the web will always have new users, and people who just don't get it, so link rot is here to stay. However I just wrote some code for Who Reads What to update the image URLs because Amazon decided to change them for no known reason. Amazon is neither new to the web and has long been a great example of a company who get's the web. So I'm at a loss as to why they decided to change a bunch of image URLs. sigh. Code is in place and the WRW is looking more like its self again.

- Peace.

1 comment:

dalin said...

Yeah, in an ideal world there would be no dead links. We deal with this quite often when we migrate a client from some archaic and poorly designed system (ex. Kintera) that has URLs like:

http://example.com/site/c.mwL5KkN0LvH/b.1406015/k.867B
/Content_Search/apps/s/search.asp

or

http://example.com/atf/cf/{39FAE17B-8AA5-4B26-934C-710DD031608
8}/flashobject.swf

There's obviously no easy way to reverse engineer what defines those URLs.

On the new site we'll setup URLs like:

http://example.com/search

or

http://example.com/files/flashobject.swf

We'll tell the client: this is the cost of creating an old->new map of URLs on the new site (both developer time and performance cost (having a billion rewrite rules noticeably slows things down)). And usually the client says "screw it". Sometimes we'll setup a nice 404 page that auto-searches for what the user might have been trying to get to, but this has performance implications as well (it means doing intensive searches in response to every bot searching for cgi-bin/some-known-exploit ).

Needless to say, the real world often makes compromises.