State of the Browsers

There’s some browser news to discuss, and I thought I’d bundle it all in one entry. Maybe I’ll even do this more often; it seems a good feature for this blog. But I’m not promising anything!

This weekend I started testing some new browsers, and meanwhile I’ve updated the HTML and CSS tables. There were no surprises. I’m continuing with the Events tables, but they’re so large and sometimes so complicated that I’m not sure when I’ll finish.

In this installment we’ll take a look at IE8RC1 and some reactions to it, Safari 3.2, Chrome’s lack of a “Check for updates automatically” feature and Opera’s antitrust complaint.

In case you’ve been brain-dead for a while, IE8 Release Candidate 1 has been released. Plenty of bugs have been solved, but a few are left, and finding them is the reason this RC1 has been thrown into the wild.

Plenty of people are bug reporting, and that’s of course good. Unfortunately, there are some people who take certain bugs in IE8RC1 to mean that the browser shouldn’t be released at all (and they even get airtime on Ajaxian).

I couldn’t disagree more; in fact I feel this is the ancient “Microsoft is Evil” meme all over again, and I’m sick of it.

Is IE8 perfect. No, it isn’t. But neither is any other browser. Is IE8 better than IE7? You bet! To me, that’s enough reason to release it.

I don’t doubt that the bugs mentioned in the Ajaxian article exist, and that they’re extremely annoying when they hold up your project, but none of them strikes me as reason enough to defer IE8’s release. In fact, the Element.prototype.cloneNode.call() bug strikes me as decidedly non-real-world. Sure, it should be solved, but it doesn’t have a very high priority.

Can’t we just forget about the holier-than-thou attitude? Please? It doesn’t solve anything. Besides, a similar case can be made against other browsers.

Recently, when working on my as-yet-unpublished political pages, I found a very annoying Firefox float bug. An open/close link that’s floated inside a block in the main content suddenly reacts to the floated elements in the right bar, and as a result it can be miles away from the block it’s supposed to open and close.

Maybe I’ll delve deeper into this bug, maybe I’ll accept it and move on. What I will not do, however, is suggest that Firefox 3 “shouldn’t have been released” with this bug; that’s just plain nonsense.

All browsers have bugs. Report them, by all means. Insist on speedy solutions, if you like. But don’t write ideology-driven drivel like this. It doesn’t solve anything.

IE8 final?

By the way, surprisingly often people ask me when IE8 final will be released. I must disappoint them: I don’t have the faintest idea. In fact I believe even the IE team itself doesn’t yet know.

The plan was to release IE8 when it’s ready, and how quickly it’s considered to be ready depends on the number of bugs found in RC1. The goal is full CSS 2.1 support, and once that goal is met we’ll see a final version. Not before.

Safari 3.2

One browser I decided not to study is Safari 3.2. As far as I can tell this is a pure security update that does not have any new CSS or JavaScript features. Apple is working on Safari 4, and I assume that’s the branch that will contain new goodies.

It was hard to find information about 3.2, though, so maybe I’m wrong. If you know of a list of 3.2 features, please leave a comment.

Chrome

When I started my testing, it turned out that Chrome 1.0 had been automatically installed on both of my systems, so the new Tables now mention Chrome 1.0 only.

Now I hate automatic updating. I’m taking considerable effort to keep older versions of browsers around on my systems, because I often want to re-check a bug in order to compare it with a new version. Chrome makes that impossible for me; in fact I have no control over the process at all.

I don’t doubt for an instant that automatic updating is a blessing for end users, but for browser compatibility testers it’s hell.

So I wondered if a future Chrome version could contain a checkbox for “Check for updates automatically.” Obviously it should be checked by default, but I really really want to be able to protect my older Chrome versions by unchecking it.

Browser Wars update

Two weeks ago I wrote a post about some unexpected consequences of Opera’s current antitrust complaint against Microsoft. If IE is unbundled from Windows, which browser should be installed instead? My conclusion was that any new computer would contain either zero or five browsers.

Opera employee Haavard responded that the hardware vendors (which are apparently called OEMs; funny name) would add one browser of their choice to new computers. Several commenters here on QuirksMode saw the same solution.

I’m not convinced this is the right thing to do. When it comes to influencing hardware vendors, Microsoft still has a considerable edge over all other browser vendors, because the hardware people need several Microsoft products (such as Windows) to make their machines worthwhile. So I doubt this is going to change anything in the short run.

Let’s for a moment suppose that Microsoft genuinely repents of its evil ways and does not pressure hardware vendors to include IE. Despite this, some hardware vendors will choose IE, and whether that’s out of habit or because they truly think it’s the best browser available is unimportant.

Now we still have some new computers that come with IE. How are we going to distinguish between IE being installed because Microsoft puts pressure on the hardware vendors and IE being installed because the vendors genuinely think it’s the best browser?

In the logic that Opera’s complaint is setting up, a hardware vendor that genuinely believes in IE would be accused of “bowing down” to the Redmond monopolist, even if this accusation is baseless. And that could be the beginning of a lot of ideology-driven nastiness.

Comments

1 Posted by Maciej Aebkowski on 3 February 2009 | Permalink

Hey! Since Chrome auto-updates, there`s no need to check bugs in older versions. Every user has the latest allready, hasn`t he?

If this works this way it would be marvelous for webdev! Not worrying a bit about older versions (imagine not having to worry about IE5x or even IE6)

2 Posted by Mike Griffith on 3 February 2009 | Permalink

I agree with Maciej re Chrome auto-updating. It's one less set of browser versions you have to check, since you can assume everyone is on the latest and greatest.

3 Posted by Alejandro Moreno on 3 February 2009 | Permalink

You guys are missing the point of bug-reporting. It really helps if you can say,

"This bug was definietly not in 0.4. It was introduced in 1.0."

or,

"0.4 had this bug, 1.0 fixed it, and now 2.0 has it again."

4 Posted by Marty Alchin on 3 February 2009 | Permalink

Regarding Chrome's auto-updating, there's still the problem of things like corporate intranets, where browsers are used daily, but perhaps without a proper internet connection by which to autodownload updates.

Granted, we aren't likely to see many corporations standardize on Chrome (though Google comes to mind) in the near future, but it's not outlandish to consider. In such situations, if the corporate rules allow only intranet traffic, the browser version will depend solely on the responsiveness of IT personnel.

Of course, you could also argue that in such situations, the internal web developers wouldn't have access to PPK's excellent resources anyway (being an internet resource, after all). But, thankfully, programmers (especially web developers) are generally exempt from such restrictions, and having a reliable resource of past browser behaviors will always be incredibly valuable.

As a side note, remember that you also need to test your sites in Chrome *immediately* upon the release of a new version, since your audience will get that update right away. No more grace period to let the early adopters point out bugs while the general populace is blissfully unaware.

5 Posted by Waylan on 3 February 2009 | Permalink

While most Chrome users will have the latest, there are exceptions. For example, the firewall I'm behind at work has never been consistent. When Chrome was first released, I downloaded and installed without incident. However, since then the firewall has been tightened up and any attempts to download the update (either automatically or manually) is blocked by the firewall. Therefore, I am stuck using an older version of Chrome.

Don't even get me started on some of the stupid restrictions the sys admin has set up... But the point is, there are cases when someone still has an older version of Chrome, yet has general access to html/css documents.

6 Posted by blurble on 3 February 2009 | Permalink

If you want Chrome without Google Update, just install Chromium or Iron. Chromium if you want bleeding edge, Iron if you want stable and with more privacy.

7 Posted by Eric Ferraiuolo on 4 February 2009 | Permalink

With the release of IE8RC1 I've updated my browser testing setup on my MacBook, and wrote about it:

http://925html.com/techniques/browser-testing-ie-on-a-mac/

I finally feel comfortable about taking on IE8 for my projects, RC1 seems to render things much closer than what I was seeing in IE8B2.

8 Posted by misterhaan on 4 February 2009 | Permalink

speaking of the chrome installer being light on options, i am mostly certain it didn’t give me the option to install to my location of choice and instead just installed wherever it wanted to.

if it didn’t automatically update itself i’m not sure how you’d have multiple versions installed when you can’t pick a different directory.

9 Posted by Hoàng Đức Hiếu on 4 February 2009 | Permalink

Maybe one can copy Chrome's exe & DLLs to a different folder for each version.

But idk how to detect when a new version is installed.

10 Posted by Steven on 4 February 2009 | Permalink

OEMs would by and large go with other browsers. Apple and Google alone would want to get in on that goodness.

Microsoft would not pressure OEMs (in theory), that's exactly what got them sued for antitrust in the first place, back when Netscape was the OEM's browser of choice.

Then again, OEMs are free to bundle browsers right now, and they don't.

11 Posted by Rick on 4 February 2009 | Permalink

I think if the European ruling stands, that OS manufacturers will bundle zero browsers.

Instead, they will include a minimal FTP to fetch the installers for whichever browser the user decides to use. Several shortcuts on the default desktop would do the trick.

12 Posted by Kit Grose on 4 February 2009 | Permalink

The main thing that causes the web development community to whine about what does/doesn't make the IE release is that IE releases are few and far-between, and in its position in enterprise, changes are even less common.

With Windows 7 looking reasonably likely to get deployed on a scale much greater than Vista (especially in enterprise), it's understandable that the web development community is struggling with all its might to make sure the browser it contains is more-or-less feature compatible with current browsers by competing vendors.

The alternative, of course, is for IE releases to be decoupled from Windows releases even more, and so running IE6 or 7 looks especially old quicker (IE10 or 11 will make 6 or 7 look ancient).

Sadly I don't see that happening any time soon.

13 Posted by Joe Walker on 4 February 2009 | Permalink

I think that what we need is undoable silent autoupdate - http://directwebremoting.org/blog/joe/2009/02/04/undoable_silent_autoupdate.html

14 Posted by ppk on 4 February 2009 | Permalink

Interesting idea, Joe, but I think this solves only part of the problem. Your example of a speaker opening his laptop on-stage only to find his rendering engine has been replaced would not be solved by it.

But maybe a choice between automatic-undoable and manual would be the best update policy.

15 Posted by riper on 4 February 2009 | Permalink

I think chrome auto update is great because it means you just have to update your bugs list for the last version and forgot previous one...

16 Posted by Terren on 4 February 2009 | Permalink

Thank you so much for the sane take on IE and on the browser wars in general. I'm sick of all the religiousness.

And thanks of course for your event and compatibility tables, they are a great resource!

17 Posted by Andrew Dupont on 6 February 2009 | Permalink

The "Element.prototype.cloneNode.call" bug is real-world if you're a contributor to a JavaScript framework and are working on IE8 compatibility. That, as I understand it, is how kangax came across the bug in the first place. He wasn't just typing weird stuff for kicks.

I admit it's an uncommon need, but bugs that cause the browser to CRASH are quite severe and should be blockers for a final release. No web site should be able to crash my browser simply with HTML, JavaScript, or CSS.

18 Posted by ppk on 6 February 2009 | Permalink

OK Andrew, if you say so. I still think it's an odd line of JavaScript, but one never knows what you need to defeat browser incompatibilities.

19 Posted by Ankeet on 9 February 2009 | Permalink

I still think people are interpreting the EU ruling incorrectly. The idea behind it was that IE shouldn't be a functional requirement for Windows. KDE still comes with Konqueror, and Gnome comes with Firefox(?). Besides, what about people who make computers themselves? They would get an option at install?

IANAL, but I think the ruling means that Microsoft can still include IE, they just have to disentangle it from OS functions.

20 Posted by Ced on 9 February 2009 | Permalink

The thing is with IE8, it's certainly better than IE7, but less than what I expected. Now I don't say they should not release it in its current state (Witch should, beside major bug, release candidate are.. well.. release candidate). But I feel they could have done more.

21 Posted by Helder Magalhães on 9 February 2009 | Permalink

> Pretty please, Chrome team?

You can disable auto-updates by disabling the "Google Updater Service". ;-)

Also, my guess is that most Chrome users and content developers just aren't aware that Chrome doesn't update in some proxy-based environments [1], used in most (?) corporate networks. There goes the auto-update feature blessing!... :-D

[1] http://code.google.com/p/chromium/issues/detail?id=14

22 Posted by kangax on 9 February 2009 | Permalink

"... All browsers have bugs. Report them, by all means. Insist on speedy solutions, if you like. But don’t write ideology-driven drivel like this. It doesn’t solve anything ..."

It does solve something. I tried submitting a bug numerous times (as I mentioned in "cloneNode bug" article) without any success. Only by writing the issue out loud, getting it on Ajaxian, that I received some kind of a response from MSFT guy (who said that IE team is aware of these bugs, yet mentioned nothing about broken bug tracker)

Don't get me wrong, I'd love to submit any bugs I find, but throwing meaningless errors on "submit bug" page is not very nice.

23 Posted by Eduardo Jordan on 9 February 2009 | Permalink

"KDE still comes with Konqueror, and Gnome comes with Firefox(?)."
Ankeet, KDE and Gnome are not OS. You should compare the IE-Windows issue against Konqueror-Firefox-etc installed on *nix, MacOS, etc.

24 Posted by ppk on 9 February 2009 | Permalink

Helder: where do I find the Google updater service?

Kangax: I didn't know you tried to report this bug (and frankly, until Andrew's comment I didn't consider this a serious bug).

25 Posted by tack on 9 February 2009 | Permalink

Auto-update: It might be useful to enumerate the software update servers and block them/redirect them outbound via hosts or gateway. Then browsers can try all they want and fail. As far as IE unbundling I envisioned that instead of a browser installed by default part of the 'unboxing' process would be an app that prompts you to choose which browser to install.

26 Posted by Jens Grochtdreis on 9 February 2009 | Permalink

For testing Chrome I recommend the standalone versions (http://stadt-bremerhaven.de/2009/02/04/portable-google-chrome-201600/). I prefer those standalones for Firefox, Thunderbird, Filezilla, Pidgin, OpenOffice and Chrome over the normal versions.

27 Posted by Jens Grochtdreis on 9 February 2009 | Permalink

To your Opera comment: I do wait for someone filing a complaint against Apple for bundling their system with Safari.
Nobody stays in the way of a freshly installed good browser. It's not Microsofts fault, that many people don't think about other browsers. For them it is just the Internet. They don't think about browsers. That's the way it should be. I cannot blame them.

28 Posted by Nick on 9 February 2009 | Permalink

The bigger point behind the Ajaxian post was that IE should think about a couple more Release Candidates. All those bugs we could whine about aside most of us will probably agree that IE8 will break a lot of the web in its current shape. Let's just hope RC1 isn't the final one.

29 Posted by Ben Cachel on 10 February 2009 | Permalink

Peter, what is it with Chrome and javascript for 'add to favorites', any such code doesn't work with this browser, is it related to the thumbnails on homepage feature implemented by Chrome? Thanks. PS: Any browser news and discussions at your blog has always been useful...

30 Posted by Ankeet on 10 February 2009 | Permalink

@Eduardo: my point still stands; you can install *nix and run apps without a web browser at all.

I think the point here is not removing IE from Windows completely, but rather getting rid of IE's tight Windows-only integration. We'd still see it installed by default, just like any browser on any other OS; it would just not have those features that make developers opt for IE-only sites. For example, the site http://www.ijji.com uses ActiveX controls to interact with their games, verifying the integrity of the files and downloading updates. From what I can see, according to the EC ruling, this functionality would no longer be, well, functional.
Many users do and will continue to opt to use Internet Explorer- the damage has already been done, and the EC never said anything about not including IE, or requiring other browser options at installation/out of the box.
IE just wouldn't be integral to the OS. That's my interpretation of their oversimplified concern "that the ubiquity of Internet Explorer creates artificial incentives for content providers and software developers to design websites or software primarily for Internet Explorer".

31 Posted by Rob Hayward on 10 February 2009 | Permalink

Really great post dude but I have to disagree to an extent, releasing browsers with known coding bugs is just wrong, render bugs maybe yea but not actual functionality that fails and Microsoft have had their fun releasing broken browsers, is time they, of all vendors, held back till it was right.

32 Posted by cthulu on 18 February 2009 | Permalink

As far as I'm concerned, MS can bundle into their OS whatever they want, people are free to choose a different OS or an additional browser. I know that's not true for corporate environments and I know that web designers get paid to provide cross-browser compatibility, but still I want to make this point:

If people surfing the web would stumble upon statements like 'this site is valid xhtml/css2.1 and if it looks all broken, your browser sucks.', they maybe would try an other browser (or get their employer to install an other), thereby moving some workload from webdesigners/-developers to the browser vendors.

33 Posted by Keith on 23 February 2009 | Permalink

I must say I disagree with your opinion that IE 8 should be released because it's better than IE 7. MS should take the time to do it right before releasing it to the wild and possibly breaking the web. Everyone needs patience or we're going to end up with yet another IE that's mostly better than the last version but is riddled with new bugs.

34 Posted by mustafa hajjar on 23 February 2009 | Permalink

yes, i agree with Keith, Microsoft should make thier browser better this time, grrrrrrr as all of thier old issues of ie were really full of mistakes.

35 Posted by Crissa on 27 February 2009 | Permalink

Oy, was doing tables and the CSS guide recommended using Em in their examples...

...And of course, neither Firefox nor Safari pay attention to Em widths for tables. Seems weirdly inconsistent.

Also, percentages in columns past the fifth don't take any precedence, so if you need, say, a seventh column to be extra-wide? Oy.

It just seems that despite new functionality, it's never the function I want.