couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "matt j. sorenson" <m...@sorensonbros.net>
Subject Re: Http 1.1 keep alive
Date Wed, 06 Nov 2013 14:47:24 GMT
On Tue, Nov 5, 2013 at 6:22 PM, Jens Alfke <jens@couchbase.com> wrote:

>
> On Nov 5, 2013, at 3:57 PM, Stanley Iriele <siriele2x3@gmail.com> wrote:
>
> > This snippet I found in the couchdb core api section of the documentation
> > seems a bit misleading to me. It looks like its saying" keeping
> connections
> > is hard so tear them down”..
>
> I agree, that quote is very misleading! It seems to be confusing
> “connection” with “request”: HTTP requests are stateless, but the network
> transport will reuse a TCP connection across multiple requests. Someone
> should rewrite it.
>
> The quote appears here:
>         http://docs.couchdb.org/en/latest/intro/api.html#revisions
> It appears at a lot of other URLs, says Google, but it seems to originate
> from the “Complete Guide” book at http://guide.couchdb.org.
>
> —Jens


So, statelessness is defined more at the application level (meaning
couchdb); and is fundamentally a pattern of RESTful interfaces - meaning
each request and response contains anything & everything necessary to
communicate application state... is that more/less correct? It's been quite
a few years since I *skimmed* Fielding's paper ;)

Where-as the http level stuff is more of a transport layer thing?

Perhaps I'm mis-using the OSI layer references[1] here, but logically it
helps me to think of the statelessness at a layer above, and
not-really-caring-about-or-impacted-by, the connection/keep-alive in the
protocol.

[1] http://en.wikipedia.org/wiki/OSI_model

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message