New website (inspired by the "Website layout" thread)

Amar Takhar verm at
Fri Feb 6 13:54:44 PST 2004

On 2004-02-06 22:40 +0100, Jeroen Ruigrok wrote:
> -On [20040206 20:12], Amar Takhar (verm at xxxxxxxxxxxxx) wrote:
> [1st-4th generation browsers]
> >Yes, speed is also a huge issue, 'nobody who matters' is an avenue Microsoft
> >followed, look at the horrible mess they created.
> I fail to see the relevance of this to the original topic.  Yes, it
> might very well be Microsoft managed to do some non-standard things with
> IE.  But the fact is that _less_ than even 1% of the user agents on the
> web are 4.x or lower.  That means that 99% are v5 and higher and have the
> relevant technical standards onboard.

So, you're saying, when you can have backwards compatability, with the exact
same site, there is absolutly no point in it?

For what reason do you suggest that backwards compatability to the very, very
first browsers be dropped.

There is absolutly no merrit in dropping this kind of support, wether it be 1%
or .05%, when you can do it, do it.

> >If it's a technical site, be technical, nobody is comming to the site for the
> >latest music video, they're comming to the site for a technical reason.
> And this is a reason to have v1-v4 still supported?  You want to create
> a maintenance nightmare trying to support v4 and lower UA's
> (User-Agents) just because it is a technical site?

It's not a maint nightmare, please provide examples of where it is a nightmare
-- i'd provide why it's not, but there simply arn't any.  If you have a good
understanding of HTML it is extremely easy.

> >> You can of course also use static pages for this, which would have made
> >> more sense given your stance towards static pages.  If you find the
> >> referring URL information useful I wonder how much it adds for a small
> >> website.
> >
> >Of course you can, however using static pages does not give you the
> >abilitity to show the referring URL, which is why custome error pages
> >are better.
> It might be very handy, but the best system is still just basic log
> files analysis of error messages.  These include referral information
> (provided you have enabled such in your webserver configuration).  This
> take is less error-prone than depending on the user of the site, whom
> in, basing on my experience, 80%-90% of the cases do not bother
> notifying the webmaster.

Yes, that is a short-term, when you have a few thousand web pages, and over 100k
or 1,000,000 visitors per day, that 10-20% who _do_ infact respond, become
extremely useful.

For an example, please direct yourself to the FreeBSD WWW mailing lists.

Or, infact, the mailing lists of mplayer, xfree86, to name quite a few who get
plenty of responses.

Eventually you're going to get _dead_ URLs, by having the referring URL logged
in a notification, you can sometimes email the maintainer and let them know that
they have an old URL.

> [site map]
> >Start at the beginning and  it gets done, start later and you're adding 4,000
> >links
> If you have 4000 links you already have a very flawed website design.  :)
> >You still havn't said WHY you can't use it, this entire email has been short
> >snippets of dis-agreeing.
> You obviously chose to read over the statements where I said "Agreed.".
> Please do not be so very selective and then say I have only be
> disagreeing with you.
> I said that some changes are worthwhile to do, some are not.  At least
> in my opinion.

No, I havn't skipped over those, what I meant was, the ones you disagreed with,
you chose not to provide any examples of why you disagreed, only to say that you

> >Please give me some points, and examples of _why_ SGML does not work over
> SGML/HTML do not promote well-formedness.  XML does.  Plus the learning
> curve of XML is less steep than that of SGML.

HTML is not well formed?  SGML is the system to handle the site, HTML is the
actual code, I don't understand by what you mean by it not promoting

Please refer to one of my previous emails on this thread which explains in some
detail on how some people can get confused when they see SGML.

SGML is used to create entities, and handle included files, nothing more, the
rest of the site is in 100% HTML.

> Furthermore, XML is easier to extend than SGML is.  The XML
> specification is understandable for mere mortals.  Same goes for XSLT.
> Ever read the SGML specification?  And the DSSSL specification?  It
> might be pearls for information management, but they suck hard to get
> your mind around.  This way (using XML/XSLT) you lower the contribution
> hurdle.

Ok, I can partially agree with you here, XSL is a bit easier to understand, only
beacuse lisp makes me _livid_.

However, If you need to create xml systems, and xslt parsers, you can very,
*very* easily fold these back into the site, for examples you can look at how
this is done on, or

Making the entire site use XML/XSL seems rather pointless, when you can have 95%
of the site in HTML, and the small 5% in XML/XSL where it is handy.

That way, most people out there can easily follow, submit and modify the HTML,
those who are more advanced in nature can fight with XML/XSL head on.  The
underlying SGML parsing system using the HTML DTD and DSSSL backend will *ever*
ever, ever have to be touched throughout the life of the project.


More information about the Submit mailing list