On May 31, 2005, at 09:50, Razvan MIHAIU wrote:

>    However there must be other restrictions for web services. After 
> reading the other definitions I came up with the following list:
>
> 1. interoperable between OSes (a message written on one OS should be 
> easily readable on another); this seems to imply the use of XML; what 
> other technology can achieve this ?

If you only need to transfer a couple of simple values (a status code 
or an URL), UTF-8-encoded plain text with one value per line would 
work.

> 2. it must be an open standard (issued by Oasis or W3C); this 
> restriction should rule out any "cgi, ejb, perl/php script" as a web 
> service; you could have a cgi acting as a web service but only if it 
> works according to a certain open protocol (this leads us to the next 
> restriction)

I don't think homegrown XML vocabularies over HTTP in RESTian services 
or objects based on homegrown classes over SOAP are excluded. A RESTian 
Web service moving FooML documents back and forth fits the "any CGI" 
scenario, and I'd still consider it a RESTian Web service.

> 3. it must have an open interface (e.g. it must work according to a 
> certain protocol); as I read in another reply there are at least 3 
> protocols to match this criteria: SOAP, REST, XML-RPC

Except REST isn't a protocol.

The only commonalities I can come up with are:
1) HTTP is involved.
2) The service is not designed to be used with a browser showing the 
service URL in the location bar. (However, browsers can act as Web 
service clients using XMLHttpRequest.)

-- 
Henri Sivonen
hsivonen@i...
http://hsivonen.iki.fi/