[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

RE: Why XML for Messaging?

• To: "'Didier PH Martin'" <martind@n...>,"'Bullard, Claude L \(Len\)'" <len.bullard@i...>,"'XML Developers List'" <xml-dev@l...>
• Subject: RE: Why XML for Messaging?
• From: "Michael Kay" <mike@s...>
• Date: Sun, 5 Jun 2005 10:08:57 +0100
• In-reply-to: <0IHL009NZDPULD@V...>
• Thread-index: AcVoYTRjuwMUVnivQFaxDiBx/OcbcwAyjksgABRHwDAADDAUIA==

Mainly (b): the fact that you couldn't get access to any application from
any workstation unless you first configured that workstation to run that
particular application; the horrendous cost of deploying an upgrade to the
application, and so on.

But also the fact that the typical client-server application architecture of
the 1980s kept a large amount of session state in memory on the server and
thus limited the number of concurrent users to 100 or so (if you were
lucky). It didn't have to be that way, but it usually was.

Michael Kay
http://www.saxonica.com/ 



> -----Original Message-----
> From: Didier PH Martin [mailto:martind@n...] 
> Sent: 05 June 2005 04:30
> To: 'Michael Kay'; 'Bullard, Claude L (Len)'; 'XML Developers List'
> Subject: RE:  Why XML for Messaging?
> 
> Hello Michael,
> 
> And there was me, thinking that the industry had regained its 
> sanity after
> the crazy and unsuccessful experiment called "client-server 
> computing"!
> 
> (The reality, of course, is that there's room for both...)
> 
> Didier:
> Can you expand Michael. What did you found insane in the client server
> "experiment" ? 
> 
> a) the rich user experience? I guess this wasn't that, doesn't it?
> b) the cost of ownership and more particularly the cost of 
> installing the
> software on every station. 
> c) the transformation of a data model into another one encoded in apps
> without any tracking tool (the knowledge of how an object is 
> represented)?
> In every platform it seems that we have this problem still 
> unresolved by our
> industry and barely in the radar of our academic community. 
> Curious no? I
> can ask to an object what are your properties and most 
> probably what are
> their types (reflectivity) but I still cannot ask an object: 
> how can you be
> represented? In the Web architecture I can ask a resource 
> what is the mime
> type, does a mime type is equivalent to a representation? If 
> yes, we may
> have the beginning of an answer in the REST architecture. It 
> implies that
> the client need to know in advance all the representations? 
> Oooops a bug
> there, client are not born...sorry made with that innate knowledge.
> d) other? Something I need to know...
> 
> Personally I found insane the cost of ownership. Or maybe I 
> should say not
> insane but surely expensive. The latter is now resolved in different
> platforms: java, .net and AJAX. You post on a server and the code is
> deployed on all stations without installation. So b) is resolved.
> 
> So, Michael, so what was insane? I am curious to know. Or I didn't
> understand the meaning of what you said (different possible
> interpretations). In that case, be more explicit on what was insane.
> 
> Cheers
> Didier PH Martin
> 
> 
> 
> 

• References:
  • RE: Why XML for Messaging?
    • From: Didier PH Martin <martind@n...>

• Prev by Date: RE: schema-aware XML parser (java heap error)
• Next by Date: Re: Another Microsoft XML patent
• Previous by thread: RE: Why XML for Messaging?
• Next by thread: RE: Why XML for Messaging?
• Index(es):
  • Date
  • Thread