CSE and IE8 "standards mode"

[MikeGale]

IE8 will shortly be released. Since they announced it I have been concerned about the IE8 standards mode idea.

Having given it a small amount of testing what I see is:

1) A setup approach that seems right, at least for a thinking user.

2) Manual ability to set a site as needing IE7 mode (= NIE7M for short) on a site by site... basis.

3) Setting NIE7M from the source / web server.

4) Some sort of automatic switch that throws a page and probably a whole site into NIE7M.

This last one (4) concerns me. What threw a site into NIE7M. Maybe there is some tool that tells me but I'm not currently aware of it.

Has anybody done some work on this / Is anybody here able to throw some light on this?

It strikes me that CSE is very well placed to make this identification (should there not be another tool out there)

Assuming CSE doesn't do so already (I haven't checked), analysis that says X might throw IE8 into NIE7M would be very useful. Preferably in 9.02 but later if not possible.

The critical time for the market is now as some people are currently concerned and are, I expect, ready to do something. (That could fade if it turns out that most users just give up and switch on NIE7M permanently, or are not worried when the auto switch kicks in and marks a site.)

NOTE: I've only just started running IE8 (rc1) in a virtual machine so I haven't researched this much yet!

[MikeGale]

With more work than I expected I have been told that IE8 does not have a way to tell you why it flipped into NIE7M.

With that under my belt I started testing some of my own sites. Not complete tests just a sampling. I found two pages on one site that tripped NIE7M in IE8.

I set up a way to test IE8 from a virtual PC image. After several test I found the issue.

It's a feature that has been in IE7, FF etc. for some time. The marquee tag, which isn't official XHTML 1 Transitional but works. They've taken it out of IE8 "standards mode".

The user will get a transient message that the browser is flipping out. (To me the message implies, this site a load of utter rubbish, which is unfortunate to say the least.)

To me that looks like a backward step, but was always the danger with this move for IE8.

Having taken a slightly longer look at X/HTML 5 in the last few minutes I'm unconvinced that this will provide an answer any time soon. So I guess we're stuck.

Maybe that IE8 test is just a strict implementation of your chosen "standard". (So custom modifications to suit what browsers actually do may need to be changed or removed.) The end result might be a less capable web for some time. (I haven't quantified that.)

[thacker abcdefg]

Mike--

IE8 has implemented a meta tag that can explicitly direct IE8 into standards mode or direct it to the IE7 rendering engine, compatibility mode.

The following will explicitly set IE8 into standards mode:

meta http-equiv="X-UA-Compatible" content="IE=8"

My view is that while Microsoft has stated that IE8 will render in standards mode by default, there is some ambiguity as how Microsoft collects information on Web content that will be listed in the 'compatibility' mode blacklist.

Short and simple, if you have a production site that should be rendered in standards mode, use the previously listed meta tag.

If you have a production site that is not standards compliant and requires IE7 rendering use the following:

meta http-equiv="X-UA-Compatible" content="IE=7"

When either tag is used, the compatibility button adjacent to the address bar in IE8 will not appear.

For testing servers, I do not use the meta tag so that manual switching is available between standards and compatibility modes.

For those who are confident that production content is standards compliant, no meta tag is necessary. However, use of it will preclude potential problems of the site from getting accidentally blacklisted as an IE7 compatibility site and getting forced into such by Microsoft.

Conversely, if the IE7 tag is used, the developer has the ability to change the tag when corrections to the content are made to bring it into standards compliance. If a Web site is blacklisted [any sub-domain compatibility issues will force a blacklist of the entire TLD and all sub-domains], contact must be initiated with Microsoft to have the site removed from the blacklist. Use of either tag leaves the control in the 'hand' of the developer rather than Microsoft.

This meta tag is a Microsoft proprietary tag that is recognized on by IE8.

All of this nonsense is available for review at the IE blog: blogs.msdn.com/ie/

[MikeGale]

Oh yes I'm entirely aware of that special tag. (And hope never to need to use it.)

I wasn't going to report back on this until I had feedback from Microsoft but I will give an update now that you comment.

I found an issue with some pages I wrote which use marquee.

These pages validate clear in CSE. (XHTML, CSS, JS all the bits.)

A guy on the IE8 release team referred me to an XHTML 1 Transitional test page of his that implements marquee and does not trigger the issue. I then tested some other pages, I added marquee to, and they didn't trigger the issue either.

That alerted me to the fact that this is not as simple as my initial conclusion. There's some sort of interaction.

I had a shot at narrowing down the problem (essentially bracketing fire to get close to the root of it). When I hit Test14 I'd had enough. Test14.htm is very simple no real text, images removed, one include (css) left... So I've turned it over to the IE8 release team to go the last mile.

I'll report back when I know what they've determined.

So a quick summary:

1) IE8 implements marquee.

2) IE8 does not have an exception reporting interface (that I'm aware of) to tell you exactly what makes it flip out. This makes narrowing issues down harder.

3) IE8 is not release software yet. (I'm now cutting them some slack.)

4) There is a way to voluntarily go into IE7 = NIE7M = flip out mode. This should, in my view, not be needed if you have a validating web site.

I've always been apprehensive of the "standards" decision for IE8. I won't form a final conclusion till the real release version comes out. Ultimately it can be a good thing if it drives an improvement of the awful markup that is this web of ours. (I was reading yesterday about a survey that concluded that 93% of pages had badly flawed markup. Might have been a Google survey. I haven't checked the survey and how it was done but I'd guess it's lenient in it's judgements.)

People who use CSE have less to worry about than the great mass of web sites.

My impression is that the IE8 release team work very hard and are smart, knowledgeable people. They are understandably sceptical of reports they get but once you get through you can sense just how hard they work. (If you saw some of the moronic stuff they have to deal with...)

[Albert Wiersch]

This is very interesting discussion. If anybody has a specific suggestion for a message for CSE HTML Validator to generate regarding this (because you may find it useful or think someone else might), then please let me know.

I will review the validator message for the marquee tag and see if it needs to be updated.

[MikeGale]

Hi Albert,

I should get feedback from MS.

I haven't got that yet so it might be a hard issue to identify or resolve.

If/when I get feedback I'll report back here.

[MikeGale]

According to http://webdesign.about.com/od/htmltags/ ... arquee.htm

The tag is supported by something like

Code: Select all

Netscape 7+
Mozilla 1+
Firefox 1+
Internet Explorer 2+
Opera 6+
Safari 1+
WebTV / MSNTV

The article is not new (or recently updated) but it seems right to me.

The article is specific about XHTML. My own interest is with XHTML 1 Transitional. I presume HTML does it and restrictive XHTML's might not but I haven't tested.

To this I would add that Chrome 1+ has support (from my own testing).

[Albert Wiersch]

According to http://webdesign.about.com/od/htmltags/ ... arquee.htm

The tag is supported by something like...

Thanks. I think I found that same information and have already made these changes:

marquee tag:
[5] The "marquee" element is not an official HTML or XHTML element but is supported by most modern browsers like Firefox 1+, Internet Explorer 2+, Opera 6+, and Safari 1+. Also, be aware that using this element may reduce accessibility.

blink tag:
[5] The "blink" element is not an official HTML or XHTML element but is supported by some modern browsers like Firefox 1+, Netscape 2+, and Opera. However, Internet Explorer (including version 7) does not support the "blink" element. NOTE: CSS officially supports "text-decoration: blink" which can be used to achieve the same effect (but still does not work in Internet Explorer). Also, be aware that using this element may reduce accessibility.

I did not mention Chrome... should I?

[MikeGale]

On Chrome.

In my view there is a chance that it becomes popular. (I don't have stats to hand but a 1% shift per month is probably the highest market share change I'd expect to see.)

These guys are serious. There is Gears, Chrome is fuelling some of the JavaScript performance improvement we see at present (well done I say, it's about time), it automatically updates pretty often...

Your call but I'd probably put it in if I was in your shoes (without checking I wouldn't be surprised if it is more popular than Opera already).

PS. Thanks for those changes. They look good to me.

[MikeGale]

I mentioned a survey of web syntax quality. I found a different version of it today, here it is:

It's part of an interview:

xhtml.com (asking the question)

Why not put an end to "tag soup" by requiring user-agents to only accept markup written to specification?

Ian (Hickson, the editor of the X/HTML 5 specification and seemingly the driving force behind it, in reply)

There are literally dozens if not hundreds of billions of documents already on the Web. A study of a sample of several billion of those documents with a test implementation of the HTML 5 Parser specification that I did at Google put a very conservative estimate of the fraction of those pages with markup errors at more than 78%. When I tweaked it a bit to look at a few more errors, the number was 93%. And those are only core syntax errors — it didn't count misuse of HTML, like putting a p element inside an ol element.

If we required browsers to refuse those documents, then you couldn't browse over 90% of the Web.

But consider — if one browser showed error messages on half the Web, and another browser showed no errors and instead showed the Web roughly as the author intended. Which browser would the average person use?

If we want to make HTML 5 successful, we have to make sure the browser vendors pay attention to it. Any requirements that make their market share go down relative to browsers who aren't following the spec will immediately be ignored.

He (Hickson) makes good points.

[Albert Wiersch]

He (Hickson) makes good points.

Yep, it's about real-world issues.

I also went ahead and added Chrome to the marquee message. Not a big issue so I just went ahead and did it.