The Internet Explorer 8 User-Agent String (Updated Edition)

The Internet Explorer 8 User-Agent String (Updated Edition) | Coded Style

The Internet Explorer 8 User-Agent String (Updated Edition) | Coded Style — Fri, 09 Jan 2009 23:19:29 GMT

PingBack from http://www.codedstyle.com/the-internet-explorer-8-user-agent-string-updated-edition/

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Arieta — Sat, 10 Jan 2009 01:05:10 GMT

I don't see anything wrong with this, though undoubtedly there will be a lot of people who will spam the comments with SVG again.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

SVG — Sat, 10 Jan 2009 02:39:23 GMT

http://www.msnbc.msn.com/id/26646919/#element(content/4/3/10)

"Tim Berners-Lee, the British-born inventor of the World Wide Web"

"Berners-Lee, director of the standard-setting World Wide Web Consortium, or W3C, said in an interview this week that Internet Explorer is falling behind other browsers in the way it handles an important graphics feature for Web pages."

"A Web image that is encoded as a scalable vector graphic, or SVG, can be resized to fit the computer screen or zoomed into without becoming blocky and losing sharpness, as happens with images encoded as the more traditional "bitmaps." Maps are one popular use of SVG."

"If you look around at browsers, you'll find that most of them support scalable vector graphics," Berners-Lee said. "I'll let you figure out which one has been slow in supporting SVG."

re: The Internet Explorer 8 User-Agent String (Updated Edition)

SVG — Sat, 10 Jan 2009 02:39:29 GMT

http://www.msnbc.msn.com/id/26646919/#element(content/4/3/10)

"Tim Berners-Lee, the British-born inventor of the World Wide Web"

"If you look around at browsers, you'll find that most of them support scalable vector graphics," Berners-Lee said. "I'll let you figure out which one has been slow in supporting SVG."

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Ben Amada — Sat, 10 Jan 2009 02:49:24 GMT

What's the reason for "Mozilla/4.0" and the "compatible" token to be in the UA string?

Why can't it just be "MSIE 8.0; Windows NT 6.0; Trident/4.0" ?

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Michael Madsen — Sat, 10 Jan 2009 03:14:28 GMT

@Ben: It goes back a looong time, to the Netscape days and everything. Vendors competed fiercely with one another, and Microsoft at some point in time added that to try to get content that was really intended for Netscape. (Netscape did similar things too, because a lot of people were only developing for a single browser, and blocking out the other one wasn't uncommon).

It has nothing to do with the Mozilla you know now, if that's what you're thinking; it was simply the codename for Netscape back in the day. Eventually, that part became the de facto standard for a user agent string - I think these days, all browsers includes the Mozilla (compatible) part. It's fleshed out a bit more on http://en.wikipedia.org/wiki/User_agent.

@OP: It's all well and good that we should keep the UA string short, but with so many things adding to it, it's hard... just take a look at mine:

Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 6.0; Trident/4.0; SLCC1; .NET CLR 2.0.50727; Media Center PC 5.0; .NET CLR 3.5.21022; .NET CLR 1.1.4322; .NET CLR 3.5.30729; .NET CLR 3.0.30618; MS-RTC LM 8; OfficeLiveConnector.1.3; OfficeLivePatch.0.0)

A whopping 249 characters, almost twice of the 128 characters reported to cause trouble on certain web servers!

All of these are either created by default in Vista, or added by Microsoft products... Removing those placed in the registry gets me down to the 64 characters shown in the post...

re: The Internet Explorer 8 User-Agent String (Updated Edition)

EricLaw [MSFT] — Sat, 10 Jan 2009 03:21:37 GMT

@Ben: Indeed, Michael is correct. You can use Fiddler (or the User-Agent picker addon) to see that some sites will not work correctly if the "Mozilla/4.0" and "compatible" are not present in the string.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Ben Amada — Sat, 10 Jan 2009 03:37:54 GMT

@Michael: I actually realize "Mozilla/4.0" in the UA string isn't related to today's Mozilla ... and it does back a long way like you said. But I don't see any reason why it needs to remain. It doesn't add any information and takes up a lot of extra bytes in bandwidth and web server log storage when considering the billions (trillions?) of IE HTTP requests made every day.

@EricLaw [MSFT]: Not all browsers have "Mozilla/4.0" in their UA string. Firefox has "Mozilla/5.0" and looking at lists of common UA strings shows that some browsers (e.g. Opera) don't have any reference to "Mozilla". Wouldn't websites requiring "Mozilla" in the UA string not work correctly in these other browsers? I personally haven't heard of websites not working correctly without "Mozilla/4.0" in the UA string. Since the IE team appears to be recently focused more on how things "should be" rather than "used to be", I think removing "Mozilla/4.0" and "compatible" in IE9 should be strongly considered. I realize it's too late for IE8, but with advance notice (via this blog) of the removal of these meaningless terms, I'm sure the 0.000000001% of websites that would break without "Mozilla/4.0" in the UA string could make the necessary changes so their sites don't break.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

EricLaw [MSFT] — Sat, 10 Jan 2009 04:23:29 GMT

@Ben: I wish it were that easy. Unfortunately, the reality is that most sites detect the user-agent string via a library, and very often the library file relies on exact string matches. In many cases, the developers don't realize that they're checking the user-agent string at all, or if they do, they have no idea how it actually works under the covers.

We learned this painful lesson back in the IE7 beta timeframe when we introduced the "b" in the user-agent string after the browser version number. This broke thousands of sites.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

billybob — Sat, 10 Jan 2009 04:50:39 GMT

I am quite shocked to see you still supporting customization of the UA string. What real world purpose does this serve other than acting as a very good place to store unique tracking IDs?

Even 'normal' users like Michael Madsen are very easy to track because of the uniqueness of their browser UA string, even without a specific IDs.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Ben Amada — Sat, 10 Jan 2009 06:15:41 GMT

@EricLaw [MSFT]: I agree there are libraries, web stat analyzers and other programs containing UA string parsing code that the users aren't aware of. But I'd guess there would be much less negative impact removing "Mozilla/4.0" and "compatible" than the problems "7.0b" caused since it's somewhat common for libraries to extract the browser version number and the libraries are expecting a numeric value -- which 7.0b isn't and this old IE blog [1] explains that. Who's doing anything with "Mozilla/4.0" except basically discarding it? Like you said, making this change wouldn't be easy, but deciding to make it and announcing the change as early as possible would result in the best chance of success.

[1] http://blogs.msdn.com/ie/archive/2006/02/03/524256.aspx

re: The Internet Explorer 8 User-Agent String (Updated Edition)

SylvainG [MSFT] — Sat, 10 Jan 2009 07:04:29 GMT

@Ben: When the backward-compatibility of tens of thousands of intranet and internet web sites is involved, a guess may not be enough.

For quite a while, Mozilla/5.0 was indicative of Gecko and Gecko-like engines and was used to discrimate IE/Netscape clients from the new Mozilla ones. Konqueror's KHTML started using Mozilla/5.0 and that is how WebKit and therefore Safari and Chrome inherited it.

Applications delivering different code to browsers based on this token might therefore fail for any number of reasons.

But most importantly, removing it from the UA string altogether is out of the question imho for one simple reason : it would break a large number of client and server libraries that extract version number, OS etc using regular expressions that *assume* the presence of the Mozilla/x.y pattern.

And that would be a lot of grief!

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Ben Amada — Sat, 10 Jan 2009 08:50:12 GMT

@SylvainG [MSFT]: Nobody can say how many libraries or sites are currently dependent on Mozilla/4.0 being present in the UA string. Some regular expressions may assume it's present and others may not. Of course the best test would be to remove "Mozilla/4.0" and "compatible" in a beta 1 release of IE9 just to see how much (or how little) havoc may result. Especially since browsers like Opera don't even have Mozilla in its UA string, there may be hope that the number of libraries/sites that need to adjust their regular expressions/parsers to not assume Mozilla/x.y exists would be a small enough number to go forward with leaving "Mozilla/4.0" and "compatible" out of IE9 RTM. Particularly if this change was announced very early on, I think giving people a full year (or so) to make any necessary changes would be adequately sufficient. Again, you guys might be right and this change may not be feasible. But a test/experiment in IE9 beta 1 just to gauge the degree of dependency on Mozilla/x.y would be a logical first step.

IE8 unterstützt (mehr) Webstandards standardmäßig: Danke Opera! Danke EU! und -okay- Danke MS!

patch-info.de — Sat, 10 Jan 2009 11:30:48 GMT

Microsoft hat sich vor der Veröffentlichung einer Beta-Version des kommenden Internet Explorer 8 (IE8) nun doch (einigermaßen überraschend) dazu entschlossen, die ggü. IE6 und IE7 standardkompatiblere Darstellung von Webinhalten standardmäßig zu verwende

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Maximilian Haru Raditya — Sat, 10 Jan 2009 14:55:41 GMT

@Ben: I agree with you. I believe it's not impossible for MS to do that. As their breaking changes stories told (for this product or other products, in the past and up until present), if they are willing to do it, they will do it.

In my opinion, they won't do it because either they have no will to do so or, if they are willing, they might think it will involve unnecessary cost and efforts, less prioritized action, etc. Both business and technology perspectives are involved in the decision.

For me, breaking changes reason doesn't actually make sense. Sounds more like diplomatic :D. No offense.

So, I think this is an example of trade-off between compatibility and innovation (for change). And I can hardly see revolution other than evolution. But that's the way this world works most of the time.

Cheers!

re: The Internet Explorer 8 User-Agent String (Updated Edition)

KS — Sat, 10 Jan 2009 15:18:21 GMT

The customization of user-agent by each and every program is indeed a problem and should be stopped. There should be an easy way in the IE GUI to a) block all writing to it b) wipe given strings from it.

As a simple example: recently a few people reported that they were getting redirected to the wap/mobile version of Ebay and Sixt when they just wanted to access the main website. As it turned out those sites must be running some buggy software (Sixt runs Typo 3, don't know what Ebay uses. Maybe it's not the CMS, but some common library underneath) that reacts on the string "update" in the user-agent (coming from Creative AutoUpdate v1.10.10) and redirects to the wap version.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Seb Warger — Sat, 10 Jan 2009 18:18:27 GMT

Just remove totally the user agent string, it is useless or error-prone at best.

La chaîne d'agent utilisateur d'Internet Explorer 8 : mise à jour

Blog - Britoweb — Sat, 10 Jan 2009 20:52:59 GMT

J'ai déjà eu l'occasion de traiter de la chaîne d'agent utilisateur d'IE 8 dans un précédent billet. Or, le blog de MSDN vient de publier un billet rapportant que cette chaîne est revue.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Gyrobo — Sat, 10 Jan 2009 22:47:51 GMT

I believe I asked this last year, but can anyone in the IE team link to any site or javascript library for which the removal of the Mozilla token would cause breakage?

Any single site. One site. One use case. That's all I ask.

Just a response that doesn't blindly invoke "backwards compatibility" as carte blanche.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

mike — Sun, 11 Jan 2009 01:40:05 GMT

@Gyrobo - I'm sure Ted will jump in shortly to point out a site that long since died that still uses this archaic method to determine how to serve up content.

I'm with you. If your production code is currently sniffing for "Mozilla/4.0" then *YOU* have a problem to fix, the IE Team does not need to bottle feed your bizarre condition.

Oh, and just for Arieta - When will IE support SVG natively like all the other good standards based browsers? Is 8 years not enough time to read the spec?

re: The Internet Explorer 8 User-Agent String (Updated Edition)

GaryK — Sun, 11 Jan 2009 03:00:57 GMT

Why does MSIE 8 need a MSIE 7 UA? Throw in the confusion created regarding NT 6.0 and 6.1 for Vista, Server 2008 R2 and Windows 7 tokens in MSIE 8 (plus the one NT 6.1 in MSIE 7 and the NT 7.0 for Windows 7 in MSIE 8) UA that are in the wild, and you've got an awful mess.

I hate to say it but at this rate MSIE UAs soon might not make any sense at all. And that will, unfortunately, lend lots of credence to those who say UAs should be irrelevant. They shouldn't and at this point in time can't be irrelevant. But with all this confusion it's hard to provide accurate UA definitions for MSIE 8.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Erik Arvidsson — Sun, 11 Jan 2009 03:05:37 GMT

Any chance of getting rid of those .NET strings in the user agent? They add a lot of overhead to every single request.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Tino Zijdel — Sun, 11 Jan 2009 03:07:54 GMT

It's indeed a bit disappointing to see that Microsoft keeps hiding behind non(public)-proven backwards-compatibility-claims. Especially in areas where that stand causes interoperability issues imo Microsoft has at least a moral obligation to investigate and share those results with the public.

By lack of any such public released investigation results we (the public) can only either conclude that any such premises are based on guesses or we should trust Microsoft on their nice blue eyes...

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Dan — Sun, 11 Jan 2009 04:24:03 GMT

Gee whiz, that took all of 2 minutes.

First, set the header string to "MSIE 8.0; Windows NT 5.1; Trident/4.0"

Visit WellsFargo bank's website and you'll almost immediately see

http://www.wellsfargo.com/browser/denied_browser

TurboTax's website also complains.

Amazon's "Search inside this book" complains.

Adobe's site forces users to manually download their products.

Yahoo's homepage asks the user to get a different browser.

etc.

There are likely thousands of others.

It's funny who quickly some folks assume that they somehow magically know more than the MS geeks who spend their entire lives thinking about web browsers.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Gérard Talbot — Sun, 11 Jan 2009 05:31:51 GMT

Hello all,

I visited

http://www.wellsfargo.com/help/wfonline/browsertest

and got replied

Your browser is: IE 7.0

Congratulations!

when in compatibility view and then

Your browser is: IE 8.0

Please upgrade your browser!

when in standards view. This proves for the nth million times that user agent string detection is not reliable, not forward-compatible, difficult (long, tedious) to manage.

For hundreds of reasons, MSDN articles should actively and explicitly promote object/method support detection (more reliable, easier to maintain, forward-compatible), then conditional comments and then, only in rare/exceptional cases, user agent string detection. That is the best recommendable way to do cross-browser coding. MSDN never did that despite IE Blog posts from as early as 2004:

blogs.msdn.com/ie/archive/2004/09/09/227542.aspx#227640

blogs.msdn.com/ie/archive/2005/04/27/412813.aspx#412856

blogs.msdn.com/ie/archive/2005/04/27/412813.aspx#412893

blogs.msdn.com/ie/archive/2005/04/27/412813.aspx#412898

blogs.msdn.com/ie/archive/2005/09/01/459541.aspx#459938

blogs.msdn.com/ie/archive/2008/10/30/hot-off-the-press-codefocus-on-ie8.aspx#9028654

Extreme care/sensitivity for not breaking the web should not weaken or hamper or prevent good, recommendable MSDN tutorials, tips & tricks writings.

Regards, Gérard

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Gyrobo — Sun, 11 Jan 2009 05:47:36 GMT

I also find it ironic that this discussion on why the user agent should retain its antiquated structure is preceded by an article on the IE8 Blocker Toolkit.

It's like saying, "We're giving you the option of not being affected by the changes we _didn't_ make."

Incidentally, I _did_ try Yahoo's homepage with the user agent in Dan's comment. Hearken, friends, for lo! he is correct.

HOWEVER-

Mozilla/4.0 (compatible; MSIE 9.0; Windows NT 6.0; Trident/4.0)

Will fail just as well. So I guess Microsoft should think twice before starting IE9.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Tino Zijdel — Sun, 11 Jan 2009 15:47:13 GMT

> It's funny who quickly some folks assume that they somehow magically know more than the MS geeks who spend their entire lives thinking about web browsers.

It's not about assuming to know more, but about determining and quantifying the underlying cause of the problems. Yes, a different UA-string will always cause problems with sites that keep on using (faulty) UA-sniffing techniques, but the question here is in how many cases this is caused by the unrecognized version number, and in how many cases only the removal of Mozilla/4.0 will actually cause problems (and other changes in the UA-string will not).

Opera is actually also running into problems moving to a 2-digit major version ( http://my.opera.com/hallvors/blog/2008/12/19/10-is-the-one ) which only showes that almost *any* change to the UA-string will cause problems with some sites.

Keeping old markers, introducing new markers and 'hacking' existing markers will only complicate issues in the future and do nothing to solve the real problem. The only way to overcome this is to standardise (between all browser-vendors), stick to this and evangalise sites that keep on breaking.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Roman — Sun, 11 Jan 2009 16:58:46 GMT

I think that the fact that some sites don't work if you set your UA string to something random like "MSIE 8.0; Windows NT 5.1; Trident/4.0" is rather irrelevant. Of course, they don't, because the real IE has a different string. However, if the IE team was to announce they remove the "Mozilla/4.0" nonsense, I'm sure they would adapt. I don't think it's that of a big deal, to change some recognition logic.

That imaginary UA value above that Dan used is very very wrong, however. If we try to parse it using the syntax for the User-Agent header, we will get:

- the "MSIE" product (no version);

- the "8.0" product (no version);

- the semicolon... which is not allowed. Whoops!

Changing it to use the correct syntax (and semantics) will yield:

MSIE/8.0 Windows_NT/5.1 Trident/4.0

Is that harder to parse than the current Mozilla/4.0 mess? I think you'll agree that it's not.

Various third party apps and addons could then add their own tokens:

MSIE/8.0 Windows_NT/5.1 Trident/4.0 MathPlayer/2.10d FDM (What's it doing here?) .NET_CLR/2.0.50727 .NET_CLR/3.0.04506.648 .NET_CLR/3.5.21022 .NET_CLR/1.1.4322

Note that the "What's it doing here" part is a valid comment. 8=]

If IE was to switch to this (IMHO) saner format, the world would adapt. Then the other browsers would probably follow suit, ending this sad "let's stuff everything in a comment and prefix it with Mozilla/4.0 just in case" story.

However, Microsoft is not likely to do it. It's too much pain for not enough gain. For a while every detection routine will have to support both the old and the new formats, which will undoubtedly lead to some aggravation.

So I guess if MS will change their UA string, it will be only after they implement SVG (wink wink).

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Richard Fink — Sun, 11 Jan 2009 17:52:26 GMT

@Tino and All...

"The only way to overcome this is to standardise (between all browser-vendors), stick to this and evangalise sites that keep on breaking."

The problem of cruft that has built up over time in the UA string simply isn't big enough to attempt this kind of effort. (It would fail, in any event. IMHO)

Sometimes all proposed "solutions" are worse than the disease.

Let's not even bother to overcome this one, shall we?

re: The Internet Explorer 8 User-Agent String (Updated Edition)

Ben Amada — Sun, 11 Jan 2009 23:42:55 GMT

Regarding removing Mozilla/4.0 from the IE UA string, there may be more pain than gain upfront. But, over the subsequent months and years, the benefits of having a more intuitive, compact UA string will easily surpass the initial effort.

Also, I think any problems caused by removing Mozilla/4.0 and the 'compatible' token at large websites or within large libraries is not a major concern since the large sites/libraries are of course going to be sure they adapt.

Interesting Finds: 2009.01.12

gOODiDEA.NET — Mon, 12 Jan 2009 11:39:06 GMT

.NET Asynchronous Fire and Forget With Lambdas There is never a collection of Value Objects Reading all

The Trident/4.0 User-Agent String デビュー

barlog — Mon, 12 Jan 2009 11:45:55 GMT

IE 8 の User Agent に Trident/4.0 という String が加わり、アナウンスがあるかな？と思っていたところ。 IE Blog でア...

Interesting Finds: 2009.01.12

gOODiDEA — Mon, 12 Jan 2009 11:54:07 GMT

.NETAsynchronousFireandForgetWithLambdasThereisneveracollectionofValueObjects...

Riconoscere Internet Explorer 8.0

di .NET e di altre Amenit — Tue, 13 Jan 2009 11:56:00 GMT

Riconoscere Internet Explorer 8.0

Internet Explorer Browser User-Agent String Problem

deedee.brainstream.net — Thu, 15 Jan 2009 17:57:32 GMT

Internet Explorer Browser User-Agent String Problem

re: The Internet Explorer 8 User-Agent String (Updated Edition)

autoverzekering — Tue, 20 Jan 2009 01:34:45 GMT

Richard, you are absolutely right. But compared to previous versions, IE8 must be better.

re: The Internet Explorer 8 User-Agent String (Updated Edition)

blah — Tue, 20 Jan 2009 05:59:24 GMT

I'm not seeing the argument for the removal just some OCD fan boys complaining about trivial things because they've ran out of other complaints. It's like if you asked IE to be written in managed code. Except that that might even make some sense for improving security and extensibility in long haul.

Internet Explorer 8 : sulla compatibilità

Pietro Brambati Blog — Sat, 14 Feb 2009 11:41:52 GMT

IE8 al momento è stato rilasciato in versione RC1. Una delle cose che preferisco del nuovo IE8 è il fatto

Just The Facts: Recap of Compatibility View

IEBlog — Tue, 17 Feb 2009 01:18:00 GMT

We’ve said a lot about our approach to website compatibility in general and the Compatibility View feature

Internet Explorer Browser User-Agent String Problem

deedee.brainstream.net — Tue, 24 Feb 2009 05:56:02 GMT

Internet Explorer Browser User-Agent String Problem

deedee.brainstream.net — Tue, 24 Feb 2009 18:41:35 GMT

Internet Explorer Browser User-Agent String Problem

Javascript 由 IE8 User-Agent 更新想到的

LeonSky — Wed, 25 Mar 2009 08:41:40 GMT

IE开发团队更改了IE8的User-agent，

更改的部分信息如下：

IE8onWindowsVista(CompatibilityView)

Mozilla/4.0(c...

How to detect IE8 using JavaScript (Client side)

Giorgio Sardo Blog — Wed, 15 Apr 2009 04:21:13 GMT

A question I’ve been asked recently is  “How can I detect IE8 using JavaScript?”. There are several

Internet Explorer 8 의 User-Agent 문자열 (업데이트)

IE8 팀 블로그 — Fri, 17 Apr 2009 08:11:13 GMT

2008 년 2 월에 말했던 (영어) 것처럼, Internet Explorer 8 은 웹 서버와 상호작용할 때, 업데이트된 User-Agent 문자열을 발송합니다. User-Agent

IT’s For-IE-day: Week 3

Jim O'Neil's Blog — Fri, 17 Apr 2009 22:34:56 GMT

Last week we took a look at Compatibility View as a way to retain the older rendering functionality just