Re: Questions related to the use of HttpClient classes

[Eric Johnson]

Om,

Responses follow....

Om Narayan wrote:

Please validate my understanding of how HttpClient and the other classes
work together.
1. When I do "httpclient = new HttpClient", the object is created by default
with the SimpleHttpConnectionManager.
2. I create "post = new PostMethod(url)" and call
httpclient.executeMethod(post). At this point httpclient takes the url in
the post object and using the SimpleHttpConnectionManager, creates a
HttpConnection object.  This connection is retained by the
SimpleHttpConnectionManager and (possibly?) reused.
The connection will be reused if you invoke another request that uses 
the "Simple" connection manager.

3. After connection is established, data is posted, and the result
retrieved.
4. I do post.releaseConnection().  Does the connection get closed at this
point? Is the HttpConnection object still around? What do I need to do if  I
had wanted to have the connection stay alive (because it takes time to
re-establish the connection)?
"releasing" a connection should not be confused with "closing" the 
connection.  "Releasing" ensures that it can be safely reused by a 
subsequent request.  For the most part, releasing a connection is 
entirely unnecessary with the "SimpleHttpConnectionManager", as it only 
has one connection, and it will "release" it internally before handing 
it out for a subsequent request.  Closing a connection only happens when 
necessary - as in a "Connection: close" header, an Http 1.0 response, or 
any of the other various standard reasons for forcing a connection 
closed.  As a general rule of thumb, HttpClient only closes the 
underlying socket connection when absolutely necessary (unless you are 
using HTTPS and a proxy server, as I recall).

5. Is post.recycle() related to connection in any way? What is the purpose
of this? Is PostMethod ctor an expensive operation?
 

I think that previous discussion on the "recycle" operation has 
indicated to me that this is a largely spurious function.  It might be 
useful, but I suspect it is mostly premature optimization.  The code in 
our application simply makes a new "Method" object for each new 
request.  Of course, we don't make millions of requests, which others 
might be, and they might report differently.

6. SimpleHttpConnectionManager doc says "This manager makes no attempt to
provide exclusive access to the contained HttpConnection". Does this mean
that calling httpclient.execute() method in this case is not thread-safe?
 

As soon as you do an "execute" on the next method using the 
SimpleHttpConnectionManager, the previous response data becomes 
invalid.  In this sense, it is not even "single thread" safe.  For 
example, code written like this: execute A, execute B, read A, read B 
will fail on the attempt to read A.  Used across multiple threads, even 
worse things can happen.  I recommend using 
MultiThreadedHttpConnectionManager.

7. If I wanted to use "connection pooling" should I be using
MultiThreadedHttpConnectionManager instead?  Or do I need to implement my
own pooling?
It should be sufficient to use the MultiThreadedHttpConnectionManager, 
which actually serves multiple tasks, one of them being pooling.

I am sure I have other questions, but this will do for now. :)
Thanks.
Om.

Good luck.

-Eric Johnson

---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]