encyclo - 5:21 pm on Jul 20, 2005 (gmt 0)

If the bugs we use to fix other bugs are eliminated, but the bugs we've been using them to fix remain, what then?

This is the biggest problem brought about by IE7: Microsoft said for years that IE6 was the end of the line until Longhorn, so developers felt confident that their hacks would last the test of time. With IE7's release rapidly approaching, it is becomng urgent to get back to all those CSS designs which have been stable for a few years (I've got several untouched since 2002) and refactor the CSS.

My tactic on CSS-P sites is to check the stats and abandon IE5.x support completely unless absolutely required. So far, I've not seen the need. That means I can clear out all the box model hacks and such and concentrate on just IE6. After that, I've been moving all the IE6-specific code to a separate stylesheet brought in with a conditional comment. Once IE7 is out, I will add a separate stylesheet for that if necessary, otherwise it'll get the full CSS treatment currently sent to Gecko/Opera/KHTML.

However, this sucks on several levels: firstly, the idea of CSS and forward compatibility has proved to be rather illusory (perhaps due to inexperience?). Secondly, the idea of "one stylesheet for all" is now illusory too - we're back to the bad old days when we were writing different code for NN4/IE4. IE conditional comments function perfectly for what we want, but the whole idea was to avoid the need for such browser-specific code as user agents converged to a standards-based uniformity.