5 ms·

Opera 10 Browser UA string format changed to accomodate bad applications

32 points by aj 17y ago

aj 17y ago
Unfortunately, there is no simple way out of this. ALL the applications across the Internets will have to be changed.. Legacy code comes back to bite us in the rear again!
zaph0d 17y ago
Browser sniffing (when you actually want to find support for a particular feature or bug) is the most idiotic thing ever.
pierrefar 17y ago
I think they should just stick to /10.00 and let bad code fail. This kind of backward compatibility will only create more crud on the internet. Opera's market-share is low enough so as not to cause too much fuss (changing IE would cause too much backlash) but high enough to prompt authors of browser sniffers to act.
- aj 17y ago
  I actually thought of the same. The trouble is that most enterprise application vendors will not be so willing to change their application. And this in turn will reduce the rate of adoption of the new browser in an environment already very slow in adoption rates
- boundlessdreamz 17y ago
  I don't think it is high enough to prompt authors of browser siniffers. More likely result is that, sniffers will stay the same and opera users will suffer.
- jasonkester 17y ago
  The absolute last thing you want to do with your 1.5%-market-share-having browser is allow it to show people broken web pages. They're correct here that nobody is going to change their site to accommodate a dying browser. The best thing they can do is what they've done. If you're Firefox with 25% share and a strong developer following, then MAYBE you could get away with that. Opera, treading water as fast as it possibly can, would simply die if it didn't render every existing page on the web correctly.
  - shiranaihito 17y ago
    I sure hope Opera is not dying or going anywhere. Most of its users are probably very faithful, myself included.
    - jasonkester 17y ago
      I always liked Opera's philosophy of "do what IE does" rather than Firefox's policy of interpreting the more ambiguous parts of the standards in a way that conflicts with IE's pre-existing way of rendering. It meant that pages just plain worked first try on Opera no matter which browser you targeted initially. That definitely wins me over. That being said, Opera is the only modern web browser that we no-longer test for on Twiddla. For a while we were putting in little Opera-specific hacks to work around its painfully slow rendering engine (which ironically is the reason it's such a pleasant browser to develop against), but eventually it became too much effort to be worth it considering the 1.35% of our traffic that it comprises. So yeah, I'm with you in hoping it doesn't go away. Still, I'll stand by my statement that breaking the user experience to make a point would be a fatal mistake for them.
      - axod 17y ago
        >> "I always liked Opera's philosophy of "do what IE does" rather than Firefox's policy of interpreting the more ambiguous parts of the standards in a way that conflicts with IE's pre-existing way of rendering." So Opera duplicates IE's idiotic bug ridden excuse for rendering, whilst Firefox does things properly. I'd rather firefox/webkit TBH.
        jasonkester 17y ago
        Precisely. IE is and always has been the standard. It's only in the last few year that Firefox has captured enough market share that the W3C "standards" have any real bearing on the situation. Of course, had the W3C simply used the box model in place in IE and Netscape at the time it wrote its CSS specification, rather than inventing its own, nobody would be calling the IE way of doing things idiotic. It would simply be the logical way of rendering content in boxes.
niyazpk 17y ago
This is the take away from the article: "Browser sniffing — unless you’re writing a web stats application — is always a bad idea. It’s a misguided attempt to send different content to different user agents. This is never scalable — you can’t change every website you’ve ever made every time a new browser version comes out. It is also not future-proof, as highlighted by this article."
- access_denied 17y ago
  http://dowebsitesneedtolookexactlythesameineverybrowser.com/ http://dowebsitesneedtolookexactlythesameineverybrowser.com/
  - TweedHeads 17y ago
    text-shadow: 3px 3px 4px #a0937f; cool
JVeinbergs 17y ago
IE already has a solution. Compatibility mode.
- mseebach 17y ago
  How do you propose that would work? The problem is that scripts detect Opera as version 1, not 10, and then failing - "if digit 1 of version < 4 then fail". To the browser, the looks like (and IS) the expected behaviour, so there no not-horribly-broken automatic fallback. In a way this solution is compatibility mode. "Broken" detection scripts see version nine (which is good enough), scripts that know about Opera 10 sees version 10.
- cake 17y ago
  If you mean compatibility view in IE8 :http://blogs.msdn.com/ie/archive/2008/08/27/introducing-compatibility-view.aspx http://blogs.msdn.com/ie/archive/2008/08/27/introducing-comp... then I'll have to agree that it would be better to handle the problem the same way (a button which would send a different UA string). Think about the people who have to deal with the UA string for legitimate reasons (analytics...) they will have to deal with a special case (if (UA == "Opera/9.80...")). It's seems a bit like a ugly hack to me.
- didroe 17y ago
  IE compatibility mode is broken. It has two "modes", one of which changes the user agent version and one that doesn't (but still changes the functionality of the browser). Guess where the sniffer scripts get the version from. The only way to make it work reliably is to look at the window.documentMode property which requires changing your code, thus defeating the point of the compatibility mode in the first place.

32 points by aj 17y ago

aj 17y ago

Unfortunately, there is no simple way out of this. ALL the applications across the Internets will have to be changed.. Legacy code comes back to bite us in the rear again!

zaph0d 17y ago

Browser sniffing (when you actually want to find support for a particular feature or bug) is the most idiotic thing ever.

pierrefar 17y ago

I think they should just stick to /10.00 and let bad code fail. This kind of backward compatibility will only create more crud on the internet. Opera's market-share is low enough so as not to cause too much fuss (changing IE would cause too much backlash) but high enough to prompt authors of browser sniffers to act.

aj 17y ago
I actually thought of the same. The trouble is that most enterprise application vendors will not be so willing to change their application. And this in turn will reduce the rate of adoption of the new browser in an environment already very slow in adoption rates
boundlessdreamz 17y ago
I don't think it is high enough to prompt authors of browser siniffers. More likely result is that, sniffers will stay the same and opera users will suffer.
jasonkester 17y ago
The absolute last thing you want to do with your 1.5%-market-share-having browser is allow it to show people broken web pages. They're correct here that nobody is going to change their site to accommodate a dying browser. The best thing they can do is what they've done. If you're Firefox with 25% share and a strong developer following, then MAYBE you could get away with that. Opera, treading water as fast as it possibly can, would simply die if it didn't render every existing page on the web correctly.
- shiranaihito 17y ago
  I sure hope Opera is not dying or going anywhere. Most of its users are probably very faithful, myself included.
  - jasonkester 17y ago
    I always liked Opera's philosophy of "do what IE does" rather than Firefox's policy of interpreting the more ambiguous parts of the standards in a way that conflicts with IE's pre-existing way of rendering. It meant that pages just plain worked first try on Opera no matter which browser you targeted initially. That definitely wins me over. That being said, Opera is the only modern web browser that we no-longer test for on Twiddla. For a while we were putting in little Opera-specific hacks to work around its painfully slow rendering engine (which ironically is the reason it's such a pleasant browser to develop against), but eventually it became too much effort to be worth it considering the 1.35% of our traffic that it comprises. So yeah, I'm with you in hoping it doesn't go away. Still, I'll stand by my statement that breaking the user experience to make a point would be a fatal mistake for them.
    - axod 17y ago
      >> "I always liked Opera's philosophy of "do what IE does" rather than Firefox's policy of interpreting the more ambiguous parts of the standards in a way that conflicts with IE's pre-existing way of rendering." So Opera duplicates IE's idiotic bug ridden excuse for rendering, whilst Firefox does things properly. I'd rather firefox/webkit TBH.
      - jasonkester 17y ago
        Precisely. IE is and always has been the standard. It's only in the last few year that Firefox has captured enough market share that the W3C "standards" have any real bearing on the situation. Of course, had the W3C simply used the box model in place in IE and Netscape at the time it wrote its CSS specification, rather than inventing its own, nobody would be calling the IE way of doing things idiotic. It would simply be the logical way of rendering content in boxes.

niyazpk 17y ago

This is the take away from the article: "Browser sniffing — unless you’re writing a web stats application — is always a bad idea. It’s a misguided attempt to send different content to different user agents. This is never scalable — you can’t change every website you’ve ever made every time a new browser version comes out. It is also not future-proof, as highlighted by this article."

access_denied 17y ago
http://dowebsitesneedtolookexactlythesameineverybrowser.com/ http://dowebsitesneedtolookexactlythesameineverybrowser.com/
- TweedHeads 17y ago
  text-shadow: 3px 3px 4px #a0937f; cool

JVeinbergs 17y ago

IE already has a solution. Compatibility mode.

mseebach 17y ago
How do you propose that would work? The problem is that scripts detect Opera as version 1, not 10, and then failing - "if digit 1 of version < 4 then fail". To the browser, the looks like (and IS) the expected behaviour, so there no not-horribly-broken automatic fallback. In a way this solution is compatibility mode. "Broken" detection scripts see version nine (which is good enough), scripts that know about Opera 10 sees version 10.
cake 17y ago
If you mean compatibility view in IE8 :http://blogs.msdn.com/ie/archive/2008/08/27/introducing-compatibility-view.aspx http://blogs.msdn.com/ie/archive/2008/08/27/introducing-comp... then I'll have to agree that it would be better to handle the problem the same way (a button which would send a different UA string). Think about the people who have to deal with the UA string for legitimate reasons (analytics...) they will have to deal with a special case (if (UA == "Opera/9.80...")). It's seems a bit like a ugly hack to me.
didroe 17y ago
IE compatibility mode is broken. It has two "modes", one of which changes the user agent version and one that doesn't (but still changes the functionality of the browser). Guess where the sniffer scripts get the version from. The only way to make it work reliably is to look at the window.documentMode property which requires changing your code, thus defeating the point of the compatibility mode in the first place.