[ee122] confusion about client behaviour
dank at eecs.berkeley.edu
Wed Sep 26 19:51:45 PDT 2007
Davide Cerri wrote:
> in the specs of the project the client specifies a server to connect
> as described below..
> "The client should compile to a binary called http client and will be
> executed using the command,
> http client <server> <port>, where <server> and <port> indicate the
> appropriate port on the
> particular server to which you wish to connect. In particular, your
> client should be able to connect to
> the server at ilinux3.eecs.berkeley.edu, TCP port 7788, run by the
> EE122 staff. (Note: an
> announcement will be sent when this server is available for testing.)"
This is the part A behavior.
> but then
> "The program should take a "full" URL (e.g., a URL of the form
> http://ilinux3.eecs.berkeley.edu:7788/index.html) as input from
> standard input; translate
> the URL into a valid HTTP 1.0 GET request as specified above; and send
> it to the server specified in the
> URL on the port number specified in the URL (or port 80 if no port
> number is specified):
This is part B behavior, and your goal for this milestone.
> so my question is, does the server
client you mean?
> disconnect from the original server
> to connect to another server or do we have to use select and handle
> multiple connections?
The server will close the connection on the client, so you will have to
perform connect() every time. This is explained in the spec. The client
should *not *be using select() to multiplex connections. Please reread
the spec if this is unclear to you.
The difference between part A and part B behavior has been explained in
this newsgroup before.
More information about the ee122