Re: complexity

On Mon, 19 Apr 2004, Tantek [ISO-8859-1] Çelik wrote:
> On 4/19/04 6:19 PM, "Ian Hickson" <ian@hixie.ch> wrote:
>
>>> It is perfectly reasonable* to build a declarative-only browser engine
>>> whose aim would be to support interactive documents with declarative
>>> technologies like XHTML and CSS while foregoing all the complexities
>>> (and security and accessibility problems etc.) of DOM and scripting
>>> support.
>>
>> It's amusing that you immediately assumed I was talking about
>> scripting. I didn't mention non-declarative technologies at all.
>
> You mentioned an "application platform", which implies/necessitates
> procedural support and scripting follows logically from that on the Web.

Not at all. While I personally think there is little wrong with scripting
and am all in favour of Web-based applications using scripting, there are
several examples of specifications intended to be the basis of
applications that make a significant effort to _avoid_ scripting, such as,
for instance, XForms.


> Also, "DHTML" is often referred to as an "application platform" in the
> context of the web, and that too relies on scripting.

It's one. It's not the only one, merely the most used.


>> And an XHTML+CSS UA would still be a lot of work; more, I would
>> postulate, than a single person could accomplish while still keeping
>> abreast of new developments in those technologies.
>
> I didn't say it would be easy. I could just see that a single very smart
> dedicated person might be able to build this.

I doubt it. At least not one that would be useful for CSS exiting CR.


>> (Not to mention SVG, XForms, XML Events, XPath, etc, that would be part
>> of a W3C, non-scripting-based,
>
> Why? I've yet to see evidence that there is a need for any of those in
> particular on the *Web*.

I didn't say there was. I was pointing out that if you wanted to do more
than have a document viewer, then you would need more than XHTML+CSS. And
if you want more than a document viewer while still avoiding any scripting
and still doing things per W3C specs, the alternative is currently the
XForms route.


> There have been offhanded mentions of doing a simple "CSS shapes"
> proposal that handles 90% of what web authors might use SVG for.

I was talking about real specifications. If we want to include mythical
ones too, then sure; that just reduces the likelyhood that the UA will be
done by one person even further.


> You yourself have proposed a much simpler HTML Forms extension.

...which requires scripting to do most of what XForms does without.


> Are XML Events necessary in Web documents?  XPath for that matter?

Both of these are pre-requisites for XForms.


>> application platform environment.)
>
> You're stuck on that term.  IMHO a semantic XHTML document styled with a
> bit of CSS for a mostly static and somewhat interactive presentation is
> *not* an application.  It's a document.  There's a lot of value in
> documents.

Of course there are. But most people using the Web now are using it as a
lot more than just a library, as sites like eBay, Amazon, and the like
will happily attest. Try using a site like voidwars.com without scripting.


> Even TBL makes the distinction on his personal web page [1].

I don't really understand what that has to do with anything. (I agree with
him, but so what?)


>> Actually, based on both my personal experience and the volume of bug
>> reports in both Mozilla and Opera bug systems reporting sites that fail
>> due to broken DOM support, it would appear that a large number of sites
>> do not, in fact, work without scripting.
>
> You misunderstand.  Of course not all the visual effects as intended
> with animated drop-down menus etc. are available, and thus bug reports
> are made. Bug reports fail to communicate what matters.

I'm not talking about menus and the like, I'm talking about things like "I
cannot buy a plane ticket on this Web page" or "I can't make a deposit on
my account". Which I see a _lot_.


> The point is that most such sites work just fine in terms of viewing the
> content, clicking links, navigating etc.

Most document-orientated sites maybe. Our users appear to want a lot more
than that these days.


The context of this discussion was that W3C specs were so complicated that
one person couldn't write a Web browser alone. My argument is that that is
to be expected given how much people are requiring of UAs now. Your
argument appears to be that if all you want is a document viewer, you
could probably do that with just one person.

While I think it's unlikely, I don't really deny it. I just think that
misses the point -- people don't want just a document viewer now. They
want the Web to provide them with a fully-fledged application environment
in much the same way that operating systems have, in the past, provided
one for locally installed software. (Your employer seems to agree,
otherwise why the fuss over Avalon-over-HTTP?)

And if the UA is going to be providing this, then forcably it will be a
bigger project than is possible for a single person to undertake, just as
all the application environments before it have been.

-- 
Ian Hickson                                      )\._.,--....,'``.    fL
U+1047E                                         /,   _.. \   _\  ;`._ ,.
http://index.hixie.ch/                         `._.-(,_..'--(,_..'`-.;.'

Received on Tuesday, 20 April 2004 09:11:27 UTC