OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   Re: [xml-dev] Generality of HTTP

[ Lists Home | Date Index | Thread Index ]

1/24/2002 9:07:06 AM, Paul Prescod <paul@prescod.net> wrote:

> I have no idea what you are talking about. The Web was
> designed for the  mainstream of its proposed user base. 
> And I'm sure the same is true for
> AC power.

Gavin may be referring to the fact that early adopters of 
electricity, driven by Thomas Edison, used DC power and literally 
needed a generating plant for every large building or city block.  
Thus AC power very much disrupted the "mainstream" electric users of 
the day, albeit a very small number.

> What is the alternative? To micro-optimize in advance for
> every conceivable use case? To design for the hardest 
> problems no matter what the cost in usability or popularity?

I agree, the AC/DC analogy is not very applicable to the web.  HTTP 
is more like AC power, easily scalable to a very large scale.  
Arguing that HTTP should have been designed better up front seems 
like arguing that Westinghouse and Tesla should have thought about 
the environmental implications 100 years in the future of a 
centralized power system that makes it difficult for individuals to 
get off the grid and use renewable energy sources.  The problem is 
real today, but expecting up-front design to handle both current 
technology and the long term implications as technology changes is 
totally unrealistic. 

The Web *is* a triumph of keep it simple, 80/20, evolutionary design 
principles.  Up-front design to eliminate the corner cases we have to 
deal with today would not have led to a better Web, it would have led 
to no Web; we'd be faced today with far worse problems of 
interoperability between more highly designed but incompatible 
systems. If HTTP and HTML hadn't been dirt simple in their first 
generation, they wouldn't have spread like wildfire. Think HTML is 
kludgy and under-designed?  How would you like to be using Blackbird 
on Windows, PDF on Macs, Frame on Unix, and trying to build anything 
resembling what we have today out of all these highly-designed 

Of course, the flip side of "extreme" design is that you have to be 
willing to refactor the system when it gets *too* kludgy.   That's 
the lesson I take from the history of HTTP, HTML, and XML ... it 
wouldn't be here if the participants had taken the time to design the 
Right Thing.  The Web is a garden to be watered and weeded and 
evolved, not a machine to be maintained in accordance with the 
factory specs. 


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS