Browser testing in 2007

What's your list?

[encyclo]

This question comes around regularly and the answers change over time as the browser landscape slowly evolves. So, what visual browsers (excluding text browsers such as Lynx) are you currently using for testing layouts?

Here's mine:

Firefox 1.0, 1.5 and 2.0 (covers all Gecko variants)

IE 6.0 and 7.0

Konqueror 3.5

Opera 9.x

Safari (latest)

IE5: I don't offer any specific support or testing for IE5.x any more, simply because on the sites I have which are tested and compatible with IE5.x the visitor numbers using that browser are non-existent.

Safari: I don't own a Mac, so most testing is in Konqueror with final verification on a borrowed machine. There's rarely any difference.

Netscape: I don't check in anything by Netscape, as the later versions are the same as Firefox and the earlier ones are dead now.

Reference: Browser breakdown, February 2007 [webmasterworld.com]

If you are building a new site with no specific (unusual) requirements, what's your list in 2007, and why?

[Robin_reala]

Well, we’ve recently revised our browser support checklist so this is up-to-date as of last week:

PC:

• Firefox 1.5, 2.0
• IE 5.5, 6, 7

Mac:

• Firefox 1.5, 2.0
• Safari 1.3, 2.0

We’re still getting enough IE5.5 to not be able to drop it yet (~2.5%), although it’ll definitely be the next one to go. Browsers that didn’t make the cut were Safari 1.0 (buggy and miniscule usage percentages, also we upgraded our 10.2 test machine), and Netscape 7.2 / Firefox 1.0 (we don’t get enough users to warrent dealing with the few annoying bugs they have).

Regarding Opera, it usually ‘Just Works’ and we don’t really have enough people using it to justify an explicit testing schedule. I keep an eye on it though, and usually give our sites a once over in it.

To be fair, ignoring IE it really feels like the promise of web standards has paid off.

[steve]

PC: IE6 & 7, FF 1.5 & 2.0

MAC: only accounts for 2.3% of visitors, so we don't bother (perhaps we would if we had one laying around!)

What I am really wrestling with is the minimum screen size to support. I'd love to move from 800 x 600px to 1024 x 768px, but 13% of my visitors still use it.

[encyclo]

MAC: only accounts for 2.3% of visitors, so we don't bother

That's a pretty high percentage (and dependent on your audience, a high number in absolute visitor numbers), in my opinion I would look to cater for that audience.

Testing for Safari is the hardest to achieve unless you're a Mac user, as the only way is to buy or borrow a Mac. However Konqueror is virtually identical (Safari is based on Konqueror's khtml core), and testing is easy with a live-CD of a KDE-based Linux distro. Perhaps it's just my particular audience demographic, but Konqueror/Safari support is an absolute must for me.

For Firefox 1.0, I tend to test with the equivalent Mozilla version instead, I don't mind minor display bugs but it's got to remain fully-functional.

One important consideration when analyising browser share for a particular site: if your site doesn't work in a particular browser, then naturally your visitor numbers using that browser are severely diminished - it's a self-fulfilling prophesy as those users just navigate away.

Is anyone doing testing with multiple OS versions, for example testing IE6/Win2K versus IE6 with XP SP2?

[Robin_reala]

Is anyone doing testing with multiple OS versions

Nope, that’s not something we worry about.

[bill]

Konqueror is virtually identical (Safari is based on Konqueror's khtml core), and testing is easy with a live-CD of a KDE-based Linux distro.

Now that's something I hadn't thought of...I could put up a Kubuntu Virtual PC or VMware appliance and test for Safari that way.

[tedster]

The last time I caught a difference that depended on the O/S was IE6 on Win2K versus IE6 on WinME - and it was quite a minor thing. I stopped testing for different O/S about 2002.

[SuzyUK]

• Firefox 1.5 and 2.0
  
• IE 5.x, 6.0 and 7.0 (I still need to troubleshoot old hacks and conditionals in 5.x ;))
  
• Opera 8.x, 9.x

No Mac either but Safari Support is so similar to FF it hasn't really been a problem

Is anyone doing testing with multiple OS versions, for example testing IE6/Win2K versus IE6 with XP SP2?

No, never been a problem BUT Vista is reported as varying from XP so :o

[encyclo]

BUT Vista is reported as varying from XP

I've seen the reports, and it's a concern - IE7 is buggy enough that I think it may require some more detailed testing if you are building a very complex or important site.

There are some differences in behaviour (but seemingly not layout) between IE6 XP SP2 and other IE6 variants, mostly due to the extra security features. In theory MS only supports SP2 and XP, not older OS versions (but that doesn't stop them remaining in use). IE6 SP2 is "stricter" (for example the pop-up blocker) so usually if it works there, it will work in earlier versions.

One other cross-platform issue is testing fonts: for example vanilla Linux installations don't have the usual Microsoft core fonts (although most users install them), so you can see what effect your secondary font choices have and tweak as necessary.

Robin, are there any particular isues that made you choose to test Firefox on both Windows and Mac platforms?

[Robin_reala]

Not really, apart from the fact that a lot of our clients are in the publishing sector and therefore doing UAT on their Macs. The only problems we've really come up against (apart from a couple of very minor and isolated bugs) are font issues - nothing major but sometimes stuff can be slightly out. Safari's noticeably better with fonts due to its use of ATSUI [devworld.apple.com], so hopefully in future rendering issues will go away as Firefox 3.0 will adopt this.

[simonuk]

I wish I could start letting some decade old browsers out to pasture but my current job still requires at least IE5 and above.

[Robin_reala]

Do you still get significant numbers of people using them? I found that I had far more leverage in dropping browsers when I had both accurate usage numbers from our current sites, and an estimation of the time spent on supporting each browser I wanted to drop as a percentage of the overall project.