compatibility?

poll

[bull]

how many browsers/os's do you use to test your pages whether they render properly?
(here: MSIE 5, 6 (PC) 5.01 (MAC) Opera 7.11, NS 4.7, 7.1)

[grahamstewart]

Depends on the target audience but usually just IE5/5.5/6, Opera 7 and Netscape 7.

More importantly (in my opinion) I always use w3c validated HTML and CSS and I handle ALL the styling and layout in CSS. This gives me maximum compatability with other (modern) browsers and allows me to support text only browsers and accessibility tools such as screen readers.

[hartlandcat]

Some people would say I test in too much...

Internet Explorer 6.0 (Windows XP)
Internet Explorer 5.2 (Mac OS X)
Internet Explorer 5.0 (Windows NT 4)
Internet Explorer 4.0 (Windows 98)
Internet Explorer 3.0 (Windows 95)
Netscape 7.1 (Windows XP)
Netscape 7.0 (Mac OS X)
Netscape 6.2 (Windows 98)
Netscape 6.0 (Windows XP)
Netscape 4.8 (Windows XP)
Netscape 4.0 (Windows 98)
Netscape 3.0 (Windows XP)
Opera 7.2 (Windows XP)
Opera 6.0 (Windows XP)
Opera 4.0 (Windows XP)
Mozilla 1.2 (SuSE Linux 8.2)
Safari 1.0 (Mac OS X)
Konqueror 3.1 (SuSE Linux 8.2)
Galeon 1.1 (SuSE Linux 8.2)
Camino 0.7 (Mac OS X)
OmniWeb 4.2 (Mac OS X)
iCab 2.9 (Mac OS X)

You don't really need to test in all these browsers... I'm just obsessed. I realise the original question was testing to see if they render properly... well, my site doesn't look quite so pretty in Netscape 3.0 and the like as it does in more recent browsers, but that doesn't matter. I test to see if my sites are usable, not if they render without the odd glitch here and there.

My policy is that my site has to display seamlessly in Trident and Gecko, almost seamlessly (if not seamlessly) in Presto, KHTML and Tasman, and still be functional in everything else.

[edited by: hartlandcat at 12:28 am (utc) on Jan. 8, 2004]

[iamlost]

I generally test on the following assumptions:

1. that "non-IE" browsers are used by choice by knowledgeable users who will update to latest version. At worst will be one version back.

2. that "IE" users will "update" only if installing new OS with new IE version. Therefore must test back several versions.

3. that if site is likely to be accessed by government or educational institutions must test for NN4.x and IE4 and may have to test for IE3 and NN3 as they never update. Depends on client.

The following are my current test browsers. The bold browsers are my minimum test. The actual site compatability testing is subject to client requirements.

Browsers With Reasonable CSS Support

• Firebird 0.7
  
• Mozilla 1
  
• Safari 1
  
• Opera 6, 7
  
• Opera 7 utilizing "small screen rendering" option to test for PDAs
  
• IE5.0
  
• IE5.5win
  
• IE6
  
• IE5.5mac
  
• Konqueror 3.1.4
  
• Netscape 6, 7.02
  

Browsers With Half-A$$ed CSS Support

• IE4
  
• NN4.7
  

What's CSS?

• IE3
  
• NN3
  
• Lynx
  

I hope to add screen readers (both audio and braille) later this year. It is yet another way to differentiate myself from the bosses neighbour's kid.

I thought I could get away with following good accessibility practices and checking with a text browser. I was visiting a blind neighbour and discussing her web browsing problems and (patting myself on the back) asked her to check out a couple I had done last year and thought were accessibility gems. To put it mildly they were better than most but not what I could call gems. Ah hubris.

[hartlandcat]

IE 3.0 actually does support some CSS... just not very much. So yes, it's generally best to consider it a non-CSS browser. I've never had anyone visit my site using IE 3.0 (apart from me when I tested it), but I had a regular visitor that was still using Netscape 3.0, although it seems as though they've upgraded now. My site only gets around 150 visitors each day, so if there's someone visiting regularly using an unusual browser, my stats will become skewed. Similarly, I was surprised to find that so many people from Denmark were visiting my site... then I realised that it was one person visiting 5-6 times each day.

Some people ask why I worry about non-IE browsers since 90% of my visitors use it. From the way I look at it, if 10% of people that visit my site don't use IE, then that's over 100 people each week that would be saying "who the hell designed THIS?" That's a lot of people, considering the small interest area base of my site. And obviously I wouldn't want to have to open IE just to visit my own site.

[grahamstewart]

Some people would say I test in too much...

You test too much!

I mean I am all for cross-browser compatability, access for all and all that jazz, but I mean... IE3?

A quick web search reveals IE3 went final in August 1996! Thats almost eight years ago, a lifetime in computer terms (several in fact)

[tedster]

This kind of decision really is a trade-off between development time, target audience, desired effect and so on.

For instance, consider this scenario:

I can use a particular effect that works well in recent browsers and I've tested it enough to know that it will increase sales by 20% over the "plain" page. However, it does not degrade gracefully at all

If the site stats show that I will lose 5% of my visitors with older browsers, but gain 20% more sales from the other 95%, then that may be a reasonable trade-off -- especially if there is not sufficient budget to allow the development of alternative pages for old browsers.

Not a purist approach at all, but eminently practical.

I approach nearly all my work like this, rather than using a hard and fast rule. One site in parcticular is courting an audience that thends to be 12% to 15% Macintosh. You can be danged sure I test those pages in a variety of Mac browsers.

Another site is a nearly captive audience who can be counted on to upgrade to any browser or plug-in we want, becuase they cannot get the content any other way. So we set out the requirements and that's that for testing.

[txbakers]

Some people would say I test in too much...

You can test all day long if you don't have to produce income.