users@jersey.java.net

[Jersey] Re: stateless REST

From: Robert Gacki <robert.gacki_at_contenttrace.org>
Date: Fri, 10 Jun 2016 17:31:59 +0200

Dum_b_ pipes, smart clients.

Akward. ;)

Robert

Am Freitag, den 10.06.2016, 17:30 +0200 schrieb Robert Gacki:
> What do you mean by directly? What is direct? I count using a browser
> as direct interaction with a server, that serves me webpages or a
> download. HTTP itself is just a protocol that is mostly human-
> readable
> - a reason for why its so successful and easy to adopt.
>
> The 'stateless' _only_ refers to HTTP or HTTP-like protocols (like e-
> mail). Maintaining state between server and client is the
> responsibility of the client and server application. Because of that,
> Fielding explains things like idempotence or what HTTP verbs are
> responsible for. It is often summarized as "Dump pipes, smart
> clients".
>
> To summarize it in my words: So the transport layer is stateless, the
> application layer may maintain state.
>
> Robert
>
> Am Freitag, den 10.06.2016, 10:34 -0400 schrieb cowwoc:
> >
> > On 2016-06-10 4:33 AM, Robert Gacki wrote:
> > >
> > >
> > > You also said that REST is for computers, not humans. Sorry, but
> > > that
> > > is just wrong. REST describes the basic principles of how the web
> > > works. It was created at a time, where HTML clients were the
> > > common
> > > kind of client to use HTTP. Fielding just explained that HTTP
> > > allows
> > > other kinds of clients. This includes clients for humans and for
> > > automation.
> > My point was that humans do not interact with a REST server
> > directly. 
> > They always go through an intermediary (a web browser in most
> > cases) 
> > that submits HTTP requests on their behalf.
> >
> > Humans expect a stateful conversation with the server. REST
> > mandates
> > a 
> > stateless conversation. The browser abstracts this implementation-
> > detail 
> > away from the human.
> >
> > Gili
> >