Re: A query please

On Fri, 20 Oct 2000, Tantek Celik wrote:

> Presumably you refer specifically to v6 right?

Currently, Netscape 6 is only in the pre-prelease 3 stage, so to talk of
it as an immeadiate solution isn't right. I have seen it, and it looks
good, but basing judgement on pre-release software isn't a good idea.

So I was not referring to v6, rather Netscape Communicator v4.72,
specifically, running under Redhat Linux 6.2. For more info about the
platform I run, just ask, I'd be happy to oblige. 

> > And after that, test the
> > code on ie. IE allows for much more little errors, like not closing
> > tables, etc, etc, whereas netscape doesn't.
> 
> Hmmm... I'm not sure I want to get in to a quirks competition.

Me either ;-)

> Suffice it to say that any "real world" browser in use today (or hoping to
> gain users tomorrow) _must_ allow for numerous errors in popular web
> content.

But it shouldn't be this way, and isn't this why we have such a wonderful
body like the w3c, defining standards that manufacturers, and definitely
coders should stick to.

> Regardless - I think the advice to start with one browser or another because
> of better strictness, is poor advice.
> 
> If you want to develop strict content, validate your content using the W3C
> validator[1] from day 1.  Every time you make major changes, validate again.

Yes, your point is well taken. So the onice lies upon the user to write
"standard-conforming" code, and the browsers to write
"standard-conforming" interpreters, thus producing the _same_ output from
one piece of code.
 
> > I'd like to see two
> > browsers made by two different manufacturers producing the same output
> > from the _same_ code... A pipe dream? I wonder ...?
> 
> Modern browsers (IE5+,Opera,NN6) from more than two manufacturers produce
> the same or pretty much the same output from CSS-1 styling.
> 
> Key to this compatibility was/is the CSS-1 Test Suite [2] which tests every
> type of valid (and some invalid) CSS value on every CSS property.
> 
> If you want browsers by different manufacturers to produce the same output
> from the _same_ standard code, there must be a comprehensive test suite for
> that standard.
> 
> Demand public, freely available, no-strings attached, no registration
> required - etc., test suites (like the CSS-1 Test Suite) for the W3C
> technologies you want to use across browsers.

Yes. Good idea. 

dan

Received on Saturday, 21 October 2000 07:07:12 UTC