W3C home > Mailing lists > Public > www-tag@w3.org > January 2002


From: Paul Prescod <paul@prescod.net>
Date: Wed, 30 Jan 2002 17:33:10 -0800
Message-ID: <3C589ED6.B924EFCB@prescod.net>
To: Shakeel Mahate <shakeelmahate@hotmail.com>
CC: w3c-forms@w3.org, www-tag@w3.org
Shakeel Mahate wrote:
> Dont GET and POST have different semantics?
> POST means change something on the server.  So how could you propose
> that GET + BODY or QUERY + BODY is equivalent to POST?

I don't think it is equivalent in a strict sense. Rather, you can model
a QUERY with a POST and a GET. Mark takes a stronger stance than I would
that it is important for every query to have a URI. I would say "just
use a cachable POST." But I think both of us would say that you really
don't need this for true queries. Every query I've ever seen can be
formulated into a URI-friendly scheme. The trick is not to try to go
through some other query language. You wouldn't try to embed an XQuery
in a SQL query or vice versa. You would rethink your problem in terms of
the query language you are using. Well, URIs are a sort of query
language. You need to rethink your problem in terms of them.

There are legitimate uses for something like GET+BODY, where the thing
you are doing isn't really a query. "Here is a JPEG, please return it to
me as a GIF." But this is a rare requirement and cacheable POST is good
enough considering the deployment cost of a new method and the serious
likelihood of accidental abuse.

 Paul Prescod
Received on Wednesday, 30 January 2002 20:34:46 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:32:30 UTC