[Wylug-help] Browser Detection (and HTML loops)

[Dave Fisher]

On Mon, Nov 22, 2004 at 11:57:36AM -0000, Andrew Teal wrote:
> I'd have to ask why you wanted to do browser detection?

Because anything approaching sophisticated CSS layout runs up against
major browser differences in box model and float model implementation.

The deliberate and consistent IE misimplementations can be handled
reasonably elegantly, without resorting to nasty CSS hacks (mainly abusing
unimplemeted CSS properties to hide stuff) or producing spaggetti code
CSS, or using JavaScript.

Unfortunately, none of the big browser layout engines (IE, Gecko,
KHTML/Safari and Opera) are especially good on consistency (e.g. Opera
has used both IE and standard float models, while there is great
variation between the rules for quirks/standards modes and the
conditions which trigger them). Hell, even the standard itself changed
the box model between CSS2 and CSS2.1 !


> Put "browser detection deprecated" into Google and you get a few reasons
> why you might no longer want to do it, not all JavaScript based. I
> suspect that these, rather than any technical difficulties, are behind
> the "not recommended" comment on the JavaScript version. And these would
> mostly carry forward into the perl version.

Yes.  I think I am aware of most of the difficulties, including the
often well-intentioned, but misguided practice of user agent spoofing.

My problem is that I may not be aware of all the alternative
'solutions'.  Without such knowledge, it seems unreasonable to conclude
that server-side browser detection combined with careful CSS
modularisation is thoroughly irredeemable.

One proferred 'solution' (JavaScript object detection in the browser
DOM) seems to threaten more nasty side effects than the browser
detection disease it claims to cure.

Using the absolute minimum of common CSS or table-based layouts are also
pretty problematic.

I'm happy enough to abandon the idea of browser detection, if I can find
an alternative which enables low cost, low maintenance, standards
compliance, usability, visibility and accessibility.

Not asking much am I?  ;-)


> BTW, I was quite interested in your email on the gb course on "seo" --
> however, we've already put in quite a lot of work on the technical side
> of our website. Do you think I'd get value from the course?! See
> http://www.hud.ac.uk/hhs. The hardest job is to get up to date content
> out of possible contributors, and I suspect that creation and provision
> of relevant up to date content is the best way to enhance the ranking?

No question about it.  This is why even technically competent sites
(most aren't) tend to fail.

Dave