Opera is making a good move with the recently announced change to its user agent string. For years, Opera has included bits of IE's identifier, trying to trick sites into serving it IE content, and then putting itself in the horrible position of having to chase IE's terribly incomplete, buggy, and non-standard DOM (and more.)
I don't buy the suggestion that this was a move to make sure that the big players in the web statistics trade, who publish semi-regular market share figures, correctly identify Opera. While Opera does attempt to spoof the IE user agent string, it also contains the string "Opera" which means that any self-respecting measuring organization or tool can (and does) easily identify which hits were from Opera and which hits were from a genuine IE browser. The suggestion by many in the Opera community that the major stats packages identify Opera as IE because of the user agent spoofing is simply bogus. David Naylor killed that myth early this year.
So why does this change matter -- certainly not for the major stats collectors. It does matter for regular web pages who often sniff browsers to determine which content to deliver. Many of those sniffing scripts are hand-rolled and can (and do) see Opera as IE, handing it IE content (usually proprietary IE DOM stuff) which Opera has been forced to try to keep up with for the last few years. Hopefully this move means that Opera is now going to try to get the "not-IE" content which, thanks to Firefox (and Gecko,) is becoming much more common and in some cases, a much better user experience.
More browsers requesting standards based content on the web is a good thing for all browsers that focus on strong support for the standards, and it's a good thing for the future of the web. Firefox, Safari (which identifies as "like Gecko",) Opera, and KHTML should all be saying loudly (and proudly) that they support the standards and they want standards based content.
Good move, Opera!
Posted by asa at August 1, 2005 09:37 AM | TrackBackOne of the anonymous comments over at OperaWatch pointed out that Opera now has better techniques than blanket UA spoofing to handle the problem (broken browser detection on websites) that caused them to use it in the first place.
The ua.ini file allows per-site UA spoofing, and can be updated from a central source. That way Opera can ID as itself everywhere, but spoof IE or Mozilla on a site that insists on dumb browser sniffing.
More importantly, Browser JavaScript is a centrally-maintained set of scripts that can apply to specific sites. It's sort of like a set of auto-updated Greasemonkey scripts. (IIRC User JavaScript -- a more direct analog to Greasemonkey -- was added in Opera 8.0 and the centrally-updated file was added in 8.01.) This makes it possible for Opera to push out fixes for sites with broken browser detection and for sites that use browser-specific features that can be adjusted to use standards instead.
Those two features, along with the efforts of standards evangelists at Mozilla, Opera, WaSP, etc., solve the problem more effectively (and more accurately) than spoofing another browser 's identity by default.
Posted by: Kelson on August 1, 2005 11:29 AMYou are wrong Asa... Many stats packages still fail to identify Opera. Whether they are "major" or not, I don't know, but it is a fact.
Posted by: Rick on August 1, 2005 11:38 AMRick: Sources, please?
In particular, I'd like to see:
1. Names of some of these broken stats packages.
2. Any that publish their stats widely, either with their own press releases or on pages that get picked up by other groups' press releases.
Any stats package that recognizes Opera when it sees "Opera/8.02 (Windows NT 5.1; U; en)" but does not recognize it when it sees "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; en) Opera 8.02" has a bug in its code which should be reported to the vendor.
Posted by: Kelson on August 1, 2005 11:57 AMRick, I cited a specific set of tests against a specific set of stats packages, http://naylog.blogspot.com/2005/03/browser-recognition-of-statistics.html If you're going to say I'm wrong, how about citing something that actually counters my evidence. "You're wrong" doesn't really get it when I've got the data and you don't.
- A
Posted by: Asa Dotzler on August 1, 2005 11:58 AMI hope it works out for Opera. I really do.
Posted by: yfan on August 1, 2005 12:27 PMIE also identifies itself as "like Gecko".
Posted by: Greg K Nicholson on August 1, 2005 01:42 PMGreg, no it does not.
IE 7 beta 1 identifies as "Mozilla/4.0 (compatible; MSIE 7.0b; Windows NT 5.1)"
IE 6 identifies as "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1)"
Nowhere in the IE user agent does it say "like Gecko".
Safari identifies as "Mozilla/5.0 (Macintosh; U; PPC Mac OS X;) AppleWebKit/412 (KHTML, like Gecko) Safari/412"
- A
Posted by: Asa Dotzler on August 1, 2005 01:59 PMNow someone just needs to convince the Firefox devs to change navigator.appName from "Netscape" to "Firefox" ;-)
Posted by: Andrew Gregory on August 1, 2005 08:13 PMSomethings missing....where are the claims of opera bashing?
Posted by: Doug Wright on August 2, 2005 06:17 AMDoug,
as you requested, here comes one: Only Asa could see this as something that is good thanks to Firefox. ("Opera is now going to try to get the "not-IE" content which, thanks to Firefox (and Gecko,)")
I don't think it was so bad. LoL
Asa does generally come across as a snob sometimes, but this case wasn't too bad.
Posted by: Nunya on August 4, 2005 01:04 PM