Lists Home |
Date Index |
On Tuesday 23 April 2002 15:07, you wrote:
> Didier replies:
> Can you explain to me how an HTTP POST brings more side effects than an
> HTTP GET.
GETs are supposed to be idempotent; caches are welcome to cache GET requests
and service later requests to that resource without hitting the server again,
etc. In most cases, the only difference between performing one GET and
performing two GETs should be in the server's logs!
POSTs are assumed to have some effect upon the world, so that performing two
POSTs is not necessarily the same as performing one POST.
> So, I am anxious to read your arguments of how an HTTP POST
> will bring more side effects than an HTTP GET (in the context of fetching
> an XML document and incorporating it into an infoset). This will expand my
> knowledge :-)
There's nothing stopping developers from making 'GET
http://www.missiles.gov/launch' destroy the world, and requiring people to do
a POST to some URL (perhaps even the same URL!) to find out the current
weather in Alaska, indeed.
But by convention, 'fetching a[n XML] document' ought to be done with GET,
and things involving activating behaviour (RPCs, basically, although people
here get a bit touchy about that word) done with POST, and things involving
overwriting a document with PUT, and so on.
> Didier PH Martin
Alaric B. Snell
http://www.alaric-snell.com/ http://RFC.net/ http://www.warhead.org.uk/
Any sufficiently advanced technology can be emulated in software