Paradox: Is Opera reliance on standards killing its usefulness?

[urlwolf]

I posted this in the opera forums, but would like to get your views too...

According to most studies, Opera is the best in following standards.
But the web is not written for Opera.

In my experience if the site is heavily AJAXified, with dropdowns for all the options, ect
Opera misses all that functionality 100%.

Honestly, I have to fire up another browser at least once a day.
And sometimes, the site is half-functional and you don't even know if you are missing something by visiting it with Opera.

The paradox is that using the most standards-compliant browser will get you the least compatibility with websites (!).
At the end of the day, what you need is to get as close to 100% of websites rendered as the webmaster intended.
This is not the case for opera. I know it's due to webmasters not testing against opera... but still. The end result is that some sites (randomly) will do something wrong and I won't even know which ones when it's not very evident.

What can be done? I don't want to switch browsers, but I'm always tempted.
That and the horrible desaster that M2 is in the current beta is making me think again about which browser to fire up!

[Lashiec]

My view is that the problem with Opera is that it does not handle too well non-standard websites. As Firefox (and supposedly IE), Opera has two rendering modes: Strict and Quirks, both self-explanatory. During the development of Firefox, the developers ensured Gecko could handle non-standards compliant websites as good as Internet Explorer, which as you see paid off, because Firefox compatibility for those sites is better than Opera (not perfect, though). It does not help there are developers blocking Opera because it does not handle their websites (not always true).

For what I read, Opera 9.5 handles AJAX sites better than Opera 9.2, doesn't it? I'm sure nontroppo will tell you more about it.

[nontroppo]

Short Answer: The biggest problem is not about Opera's consistency with standards. It is that web developers don't test in it.

Long Answer: The issue is the following:

The damage from the first browser war was a fractures into "built for netscape" and built for IE". Mozilla's quirks come from the netscape camp, actually it has poorer IE quirk handling than Opera. Opera tried to handle *both* sets of quirks, because browser sniffing has been the dominant way to quickly fix issues:

Code

if name=IE
  blah
else name=netscape
 other blah
endif

Opera must handle one quirk or the other and must try to reverse-engineer exacly what they are doing. With layout this can be tedious as the combinatorial possibilities are large.

Opera designed a masquerade, it would pretend to be another browser, and set about coding round the quirks. Those quirks are horrid to code round, and occur even in "standards" mode. Opera has tried very hard to handle quirks gracefully. First it emulates major quirks of IE and Netscape as best it can. Secondly in recent versions it has browser.js which fixes sites on-the-fly (hooking custom scripts into the page loading process) and auto-updates those fixes. Finally site preferences dynamically change its masquerade behaviour per-site, again auto-updating.

In my experience, Opera is a lot more compatible now than it was a couple of years ago (9.5 is a great step forward here too). *BUT* browser sniffing is back with a vengence with Web 2.0 (Google being the worst offender), and so is coding for IE or NetscapeFirefox. The biggest problem is not Opera's compatibility. It is that web developers don't test in it.

Developers hack something and it works in Gecko. Then they hack something that works in IE. Then they ship the product. If they strictly code to standards then it is most likely to also work in Opera and Safari, but not guaranteed. This is especially the case when Gecko has got the spec wrong, as has happend causing clear Opera incompatibilities recently.

Opera have made some positive steps: Opera now powers Adobe CS suite (Dreamweaver, Device central) which will slowly get developers aware of rendering and functional bugs in it. And they are furiously coding new web developer tools to make Opera a better browser to develop in. The idea is to get parity with Firebug. They have opened http://dev.opera.com to provide developer docs that have an opera-aware bent. And they have a small team who "Open-the-Web" - find bugs in web sites and contact the owners with possible solutions. But the web is big and Opera's resources are small.

I know this sounds very soft on Opera from my part, but I really believe Opera has and is trying its best to be compatible, but they are running against a huge Sea of complexity. Things are getting better and the fact that the mobile-aware web is growing will benefit Opera, as will the release of the dev tools to get people developing in it.

[Ampa]

From my (albeit) limited experience as a designer...

I build my pages and test in Opera first, then FireFox, followed by IE7 and finally IE6.

In the vast majority of cases: If it works in Opera, then it will work in FireFox and surprisingly well in IE7 (though a few CSS tweaks are sometimes required). IE6 invariably takes a bit more work.

My point: that by designing to be standards compliant, the work load for compatability with the other borwsers is minimal.

[jgpaiva]

Unfortunately ampa, not all developers are aware of that

[Lashiec]

So, nontroppo, KHTML/WebKit based browsers suffer from the same problem as Opera?

[nontroppo]

Lashiec: yes, although as the dominant browser on most Macs, developers are more likely to test with it (well, at least now that Safari is Pc compatible, PC-based devs can). And as Macs are much more prevelant in design circles that also increases the chances of web designers testing on Safari. For example the Ruby on Rails bunch are largely Mac based and I suspect lots of them use Safari...

If Presto had been chosen instead of KHTML by Apple things would be very different, but Apple would never have chosen a closed-source component.