The problem turned out to be something odd with how IE handles styles which apply to both an ID and a class without any other selector. For example, if you have this stylesheet:
But I got it all working. So the neat project I'm working on will survive while still using nice clean semantic markup. (Or at least as clean as our app framework will allow.)
Meanwhile, my cable Internet still isn't working. Too much noise on the line, and the techs at the cable company are being uncooperative. Go figure. I'm just going to cancel it and go with Speakeasy after all.
It is really a blight. I've been screwing around with "ajax" stuff lately, and I swear that 'getting it working in IE' takes as long as actual development.
Back when I did DHTML for a living, just getting something working with different versions of IE was a nightmare.
Ironically, the Ajax part of what I'm working on was the easiest thing to get working.
Incidentally, maybe you can help me out with something... let's say I have a bunch of CSS rules which all reference the same image (on the background property). How can I make sure IE doesn't load a unique copy for every single image? On my computer it loads only one copy (like it should) but on some other systems it loads one for every element, making IE suddenly take up several hundred MB of RAM. (It's a lot of references to a fairly large image.) I can't figure out any rhyme or reason to what's causing this to run fine on some systems but to mess up horribly on others. The only difference between mine and the others is that my system doesn't have VS.NET installed yet while theirs all do. (Though I'm getting VS.NET next tuesday so maybe that'll help to narrow things down, as well as get my system to where I can do a much faster debug cycle than 'hey could you reload the page and tell me what happens?' and so on.)
I've done everything I can regarding cache/proxy headers and so on, so I'm pretty sure it's not that...
I've only ever had the opposite "IE caching things it shouldn't" issue.
When I did heavy DHTML (no CSS), our images were tiny, so we probably wouldn't have noticed.
When things act flaky on different Windows systems, it's likely because of the level of updates on the various boxes. Microsoft just loves to update things without telling the user.
It could be something obvious, like NT vs. 9X, or it could be something less obvious like "hasn't run update in a while" or worse, "hasn't installed Office".
About six years ago, I ran into a situation where something worked on our systems and failed at the customer site despite matching service packs, etc. Turns out that Office was installing a different version of a DLL, so it failed Office had been installed.
This was before the days of Windows update, though.
I have a feeling it just has something to do with its caching behavior. If I can isolate it in a way that makes it so that we can give directions to customers (which aren't just "use Firefox instead, you dweeb" or including a warning about possible system stability when using the app on IE in that particular mode or whatever), then I think it'll be fine.
Meanwhile, this is a non-critical feature, but it's a REALLY COOL feature which would be really nice if we could deploy it universally, especially in version 1. Oh well.
At least I fixed an issue which was making IE actually crash when hiding the sidebar.
Comments
div#foo.bar { background: red; }
div#foo.baz {background: blue; }
and this HTML code:
<div id="foo" class="baz">stuff</div>
the div wouldn't be styled at all, regardless of whether there was a #foo.bar on the page. (which there isn't.)
IE also handles width=100% strangely. (It uses the width of the window, not the width of the container.) KHTML and Mozilla both do this correctly...
Back when I did DHTML for a living, just getting something working with different versions of IE was a nightmare.
Incidentally, maybe you can help me out with something... let's say I have a bunch of CSS rules which all reference the same image (on the background property). How can I make sure IE doesn't load a unique copy for every single image? On my computer it loads only one copy (like it should) but on some other systems it loads one for every element, making IE suddenly take up several hundred MB of RAM. (It's a lot of references to a fairly large image.) I can't figure out any rhyme or reason to what's causing this to run fine on some systems but to mess up horribly on others. The only difference between mine and the others is that my system doesn't have VS.NET installed yet while theirs all do. (Though I'm getting VS.NET next tuesday so maybe that'll help to narrow things down, as well as get my system to where I can do a much faster debug cycle than 'hey could you reload the page and tell me what happens?' and so on.)
I've done everything I can regarding cache/proxy headers and so on, so I'm pretty sure it's not that...
When I did heavy DHTML (no CSS), our images were tiny, so we probably wouldn't have noticed.
When things act flaky on different Windows systems, it's likely because of the level of updates on the various boxes. Microsoft just loves to update things without telling the user.
It could be something obvious, like NT vs. 9X, or it could be something less obvious like "hasn't run update in a while" or worse, "hasn't installed Office".
About six years ago, I ran into a situation where something worked on our systems and failed at the customer site despite matching service packs, etc. Turns out that Office was installing a different version of a DLL, so it failed Office had been installed.
This was before the days of Windows update, though.
Meanwhile, this is a non-critical feature, but it's a REALLY COOL feature which would be really nice if we could deploy it universally, especially in version 1. Oh well.
At least I fixed an issue which was making IE actually crash when hiding the sidebar.
God I hate IE.
Oh, and if you want to mess with a silly toy, IM me some time.