IE7
By glazou on Sunday 31 July 2005, 16:32 - Microsoftisms - Permalink
I have read almost all what was written/published at first glance about IE7, in english, french and even swedish. It just proves once again that people should think seven times before blogging bullshits. The I-ll-be-the-first-one-to-blog-about-it race is dangerous. I've also carefully read the list of improvements provided by Chris Wilson. The whole thing deserves a few very simple comments here:
- Extended CSS support in IE7, even if many people expected better than what's listed, is excellent news.
- IE7 adds support for the child selector. Wow. Fi-nal-ly! The '>' one. The FOAICB (father-of-all-ie-css-bugs; the mother being the box model). I heard a few years ago a Microsoft "representative" tell me and three others with me that adding support for '>' was not simple because of the original design of the layout engine... I guess they just missed a byte in their representation of a chain of selectors to code the combinator. Since the '+' combinator is also supported now, I guess somebody finally found enough time to add that byte, the event-based reentry and the stuff around. This raises a few questions:
- So it was not so hard/impossible to fix, right? Seriously, between gentlemen, is there a single good reason why we don't have at least '>' implemented in IE since 1998?...
- Is support for '+' and attribute selectors fully dynamic, as I hope it is?
- What's going to happen to the tons of web pages using hacks to "filter" IE6 based on the lack of support for combinators or rules or whatever? Since IE6 is not going to die any time soon (remember there are ol'Windows users?), we're going to end up with rendering conflicts between IE6 and IE7... Probably unacceptable from a customer's perspective.
- Could Microsoft (hear Chris Wilson) publish a Selectors Profile instead of just saying "CSS 2.1 Selector support (child, adjacent, attribute, first-child etc.)"? That would help a lot.
- Any plans to improve the CSS Object Model?
- Adding support for CSS3's "::" is important and should be done immediately, since it's quite simple to implement.
- I definitely agree with Chris about the Acid2 test. Some bits of it are just here to show a great respect of standards but do not necessarily improve drastically web authors' and visitors' experience. Remember the target of a browser is you, me, the girl next door, your mum, and even your salesperson (well, maybe not your salesperson, too dumb after all); the target of new language features in that browser is the Web author, and that's an entirely different beast...
Comments
Alors oui euh... bon d'accord j'ai sauté sur l'occasion de dire des méchanteries quand IE7 bêta 1 est sorti, je me trouvais très maline et là pouf, je me sens stupide (un peu). Alors ok, je veux bien faire amende honorable mais il me reste une roue de secours pour la mauvaise foi: IE6 risque de rester très longtemps majoritaire étant donné la mauvaise prise en charge de IE7 (qui n'est toujours pas sorti) sur les OS de Windows, et IE 6 n'est pas patché. Et toc tralalère :p
"What's going to happen to the tons of web pages using hacks to "filter" IE6 based on the lack of support for combinators or rules or whatever? "
I assume (such) developpers will have to rewrite parts of their styles sheets, once again.
Hardly the first time they (we) will have to hack-around-the-box-model and play with the css parser bugs...
A friend of mine used to serve different stylesheets based on the UA string and some obscure javascript tests... not a very good idea IMHO; still, relying on parsers bugs/lack of support may not be better than that...
Attribute selectors are definitely good news.
I also wonder if the support for "display" will be improved.
Css support sure is important.
Still, I personnaly think we are making too much of a deal about it, thus we forget the other IE problems (not directly css related).
- will we still have to strip off the xml prologue from xhtml docs to trigger IE "strict" rendering mode?
- what will IE7 do with a application/xhtml+xml content-type http header?
- more generally, will the http compliance be better?
- will IE7 handle self closing script tags (<script type="text/javascript"... />)?
- will IE7 finally tell the user that he needs to install a plugin when you don't use the infamous classid in your object?
- will they back-out the most stupid patch ever (IMHO), when they dropped support for user:pass@server/ urls?
"... [rendering conflicts between IE6 and IE7 is unacceptable from a customer's perspective] ...".
This is absolutely true; so, how will they manage to achieve 100% compatible behavior, and still enhance that (faulty) very behavior?
I can't help but think "IE7 as a goal" (and even some specific long awaited bug fixs) is plain useless to us.
We will end-up with just yet another half-assed UA that we will have to deal with, with specific bugs, and we won't get rid of IE6 before years.
I sincerely hope IE7 is just a mean on the road of active browser development, and that they won't go back to sleep once IE7 is out.
By the way, I wonder what we should await from IE in Vista...
Will it be just IE7, or a totally new "browser"?
Regards,
- Olivier G
You make some good points. Since this "beta" doesn't seem like much more than an alpha I'm going to take a wait-and-see approach to it myself. Frankly, I'd love it if they make the thing work properly; it would only make my job easier.
Oh, and one (potential) little nitpick: In American English, "bullshit" is never pluralized -- it's a collective noun. I can't comment on how the Brits cuss, so if you're doing it right in en-GB, just tell me I'm full of bullshit. Or bullshits
According to the IE7 blog, they are still fixing specific CSS bugs instead of looking at their box/float model implementations. It's total news to me that adjacent and child operators are supported, do we also get 'position: fixed' finally?