Mark Anderson wrote:
>...
> Matthew, you have way too much time on your hands. :) (What are your
> real thoughts on this, anyway?)
If you can't guess, then I haven't been doing a very good job over the
past year ...
Ceteris paribus, from a usability point of view the equation is fairly
clear: the browser which can view the most pages wins.
(If you don't like the idea of a particular browser `winning', then
fine. Just add a further layer of abstraction. The set of standards --
Internet Explorer standards, or W3C standards -- which more pages are
written to, will end up as the /de facto/ standard. Browsers which do
not recognize the /de facto/ standard will fail. Ok? Ok.)
There are two ways you can get in the situation of viewing the most
pages. Firstly, you can convince Web authors to use code which only
works in your browser, and not in other browsers. Sound familiar? It
should. Secondly, you can treat badly-written pages more kindly than
they deserve -- even coping with such fundamental things as missing
</table> tags. Sound familiar? It should.
Tightening the screws on invalid HTML, or URLs, or whatever, is
something you can only afford to do when you are in a commanding
position <http://htmlhelp.com/tools/validator/reasons.html>. When
Mozilla is at 20 percent market share, and dropping, we cannot afford to
be anal about what we accept and what we do not.
So, if <http:/foo/bar> is completely meaningless as far as the RFCs are
concerned, then we can do whatever we like with it. And if one
particular interpretation of that URL produces the desired result in the
majority of cases (as it must do, judging by 4.x's and Internet
Explorer's behavior with the same URL), then we just have to swallow
that bitter pill and do the same thing.
For the same reason, if someone provides code for Mozilla to support
<layer>, that code should be accepted into the tree.
--
Matthew `mpt' Thomas, usability weenie
(Fading away into the distance)
