In its current incarnation, IPP relies on HTTP 1.1, which by default
maintains persistent TCP connections between HTTP client and HTTP
server. The client and/or server keeps a cache of open connections
between clients and servers for a period of time, so they can be
reused for subsequent requests. The cache lifetime for TCP connections,
as well as the total number of concurrent TCP connections maintained,
is implementation dependent.
With persistent connections a client can make as many requests and
receive as many responses as it wishes prior to the closing the
connection. NOTE: the server or the client may elect to close a
connection so both client and server should be prepared to handle
an unexpected connection close operation.
Daniel Cogswell wrote:
>> I've been working on a prototype that uses the IPP and I have run into a
> question which I would be interested in hearing some comments on.
>> I think of HTTP as not being a connection oriented protocol. I know I'm using
> the term loosely here, but, as I understand it, when you open up a connection
> using http you can write and then you can read once. You can't do the
> read/write read/write kind of activity. There is, however, the CreatePrintJob
> command in IPP which wants to maintain a connection. That is, it wants this
> read/write read/write connection. How does IPP plan to maintain the connection
> when it begins to append documents to the job?
>> Dan Cogswell