Marking up invalid sites is unquestionably pointless, believing valid
sites are all you should do is also pointless, but valid mark-up
doesn't prevent anything, and will catch an awful lot of errors.
What on incredibly odd way to develop, why not author for the web?
So you only do any QA with the old browsers, sounds an odd way of
doing QA - I bet you charge more for your "maintenance" yeah?
Authoring for the web is a nice idea, but... where are the limits?
Going back to 3+ browsers? Being accessible for LYNX ? It could give a
nice feeling that your site is LYNX-compatible, but is it really
useful? I doubt
I am mostly active in field of web applications and I usually require
IE5+. This gives me enough space to use some client script, xml, css2,
tabular controls and so. Other browsers get an error message "You need
IE5+ to continue". Most of my work consists out of customer login,
intranets etc. and until now I never had someone complaining like "but
I use Netscape 2!!" or so
People with that kind of old browsers
mostly have other browsers as well; they know what they are browsing
with and know what to expect.
I believe it's not odd to develop that way. It is a matter of what is
required for what type of thing you want to do. LOT of databases with
browser interfaces set requirements. Especially when it comes to
intra/extranet usage. There is no other choice if you want to use more
complex client scripting.
Of course when developing an "average" website for the public, the
idea must be totally different and it must be compatible for more
browsers. Basically that is a matter of choice of the webmaster - one
says "IE4+ NS4+" other says "HTML3+" other says "800*600+" etc. For
public websites I use something like IE4+ NS4+ 640*480. Now we're not
gonna discuss whether or not that should be HTML3+ or so, OK ;-)
No, it's the majority, in fact you probably won't find a UA which
tells the truth... Of course they might do somewhere in the string
aswell as the lies, but then you're relying on the strength of your
measuring script, I've yet to see a good one. Also of course people
reaching your webserver has little to do with people viewing your
site, and then there's the self-fulfilling aspect of logs. Could you
explain why you believe the logs are "more and less accurate" ?
I scan logs with ANALOG which is a well known analyser that interprets
Apache logs. I believe Apache logs are more or less accurate because I
believe that for 5 years or so. Considering the general main stream
information about Apache logs, I do not see many reasons to doubt.
Now I need to go back to work guys!
was a nice discussion, thanks
Bart