Alex Robinson wrote: > Microsoft have said "Developers, you must use conditional comments, > you must use X-UA, these are the only sanctioned ways". > > Consequently it seems self-evident that Microsoft cannot possibly > ship IE8 with conditional comments not honouring the X-UA string > (though as we've seen, that means that for IE=5 and IE=7, the version > of IE that IE8 should match in a conditional comment is 7) > > But think about it - the conditional comment matching comes from the > vector in the registry and to get multiple copies of IE running > alongside one another using the appropriate CCs for each takes > hacking of the registry > > http://www.positioniseverything.net/articles/multiIE.html
I truly believe that IE8 using this XUA string and using any conditional comment is wrong. None of this would matter if IE8 was a full stand alone version "WITHOUT support for CCs." My brief testing was mainly concerned around IE8 in IE7 mode and supporting the crazy world of "backward compatibility." Now I have confirmed in my special test that any developer wishing to support "backward compatibility" should never play around with toggling. :-) > I'm giving Microsoft the benefit of the doubt that this *will* be > fixed, reasoning that in the face of the storm over the issue of > compulsoriness of the X-UA header, their announcements about > interoperability and convenient events to launch such things, that > they just had to get what there was out there. And currently the > support for dynamically tweaking the vector string is broken. > > Now if my conjecture is wrong, and Microsoft really do intend to ship > an IE8 that treats conditional comments as IE8 whatever mode it's in > then, well, I can't even begin to imagine the fallout for them and > for developers. You are so correct here. I hope that Microsoft decides to change there mind again. Well it won't affect me directly since I will never use X-UA-Compatible in any mode or conditional comments on any of my pages. I will only use any import or selector hacks necessary to deliver the correct style rules for any generation of IE8 beta and especially since they work consistently. :-) > >It seems like that X-UA "IE=8" and [if IE 8] or generally IE8 mode alway wins > >out. Have you tried [if lt IE 8]? > > I have now > > http://www.fu2k.org/alex/css/cssjunk/ie8/xuas > > IE8 shows up as being gt than 7 and not lt 8, regardless of X-UA > target or mode. That would be correct, I was meaning how does IE8 respond when in IE7 mode, is lt 8 used correctly? Remember to test any page with X-UA=IE7 by first closing all instances of IE8 down and reloading it (see below). > >> Second: The emulate-IE7-button should overrule the X-UA "IE=8". > > I'm not bothering with the whole emulate IE7 thing, given the things > you have to remember (and that I then forget that I'm in IE7 > emulation...) > > I run a second virtual machine with real IE7. > > Because it seems highly likely that IE7 emulation in IE8 is really > only going to be equivalent to IE8 targeted with X-UA IE=7 - I > wouldn't trust IE8 to tell me how the real IE7 behaves. Since I don't have a virtual PC I am only using IE7 mode to test the layout or parsing differences between IE7 and IE8. IE8 in IE7 mode emulates the true IE7 perfect (to my memory) in all my test cases and examples on my site. What I mean is that IE7 emulation mode and the real IE7 will shows the layout effects caused by hasLayout so toggling any test concerning hasLayout like the Guillotine bug, Peekaboo bug, collapsing margin bugs or test with floats all show correct behavior depending on which (toggle) mode your in. IE8 any mode is correct in the different support that IE7 and IE8 have for selectors (even hacks). So I see no harm in using the toggle for testing this way but when you enter the world of meta mode madness then all test go crazy. :-) > > What we need is a valid IE8 filter hack and indeed I have found one. > > > ><http://css-class.com/test/ie8hack-valid.htm> > > Yes, I see why this is important right now, if you are using the > emulate IE7 button ;) I think all of our test have now confirmed that testing XUA=IE(whatever) is totally broken when using CC or when you being toggling. I wrote in error before. > This all changes when you do either. > > 1. Toggling into IE8 Standard mode. this is CORRECT. > 2. Selecting the XUA=IE8 test. this is INCORRECT. The test cases only change once you begin toggling. So this happens after a clean load of IE8: Pre-toggle all test pages: Normal - red, red, orange, orange and blue XUA=IE7 - blue, blue, green, green and red XUA=IE8 - red, red, orange, orange and blue After toggle (IE8 standard mode) all test pages: Normal - red, red, orange, orange and blue XUA=IE7 - red, red, orange, orange and blue XUA=IE8 - red, red, orange, orange and blue After toggle (IE7 Strict mode) all test pages: Normal - blue, blue, green, green and red XUA=IE7 - blue, blue, green, green and red XUA=IE8 - blue, blue, green, green and red Can anyone please confirm that the real IE7 will show blue, blue, green, green and red on all test pages? <http://css-class.com/test/ie8hack-valid.htm> This means that testing with XUA=IE7 can only really be correct if no one decides to toggle. After toggling the registry or something gets corrupted and only a clean load of IE8 will fix this. I didn't read anything about this on the IE blog. Alan <http://css-class.com/test/bugs/ie/ie-bugs.htm? ______________________________________________________________________ css-discuss [EMAIL PROTECTED] http://www.css-discuss.org/mailman/listinfo/css-d List wiki/FAQ -- http://css-discuss.incutio.com/ List policies -- http://css-discuss.org/policies.html Supported by evolt.org -- http://www.evolt.org/help_support_evolt/
