Re: Background information on GET and XForms (was: GET should be encouraged...)

Tim Berners-Lee wrote:
> 
>...
> 
> In the future, this does call for a new GET-with-body or QUERY or whatever
> you like to call it, which would be defined as an operation without side
> effects
> (a function) of both the URI  and the message body.

I asked Mark Baker for this method six months ago but I am now
ambivalent about it. 

POST can be used in a caching manner. If you add GET-with-body you will
find that people who do not want to resource model will use
GET-with-body without understanding why they SHOULD resource model their
domain into standard GET. For example:

> What rammifications does this have on the XForms deprecation of GET?
> 
> I concur that QUERY is what XForms needs

Probably 1 in every 100 side-effect-free XForms would appropriately use
GET-with-body but already people are talking about using only
GET-with-body for XForms, even though it doesn't even exist yet!!!

GET-with-body is going to increase the confusion about which method to
choose for a tiny subset of all problems. It is arguable whether the
benefit is worth it. 

> (I prefer QUERY to an adaptation of GET, myself).

Today we express queries as GETs in most circumstances (after all,
search engines accept queries). That's a wonderful thing. The new method
(if any) should not be called QUERY. Perhaps FUNCTION.

 Paul Prescod

Received on Monday, 28 January 2002 17:24:41 UTC