Re: flush buffer using socket recv() with non-blocking flag

[Scott Klement <midrange-l@xxxxxxxxxxxxxxxx>]

Hello,

I'm not sure that I understand?   Let's see if I do...

You are sending multiple requests:

req = 'COMMAND TO DO SOMETHING';
send(fd, req, len(req));
req = 'COMMAND TO DO SOMETHING';
send(fd, req, len(req));
req = 'COMMAND TO DO SOMETHING';
send(fd, req, len(req));

In this example, 3 requests.  Now you want to get the response from the 
3 requests, but you want to make sure you get the response from the last 
one?

If so, why not use a counter?  (Keep receiving until you've received the 
3rd response.)

I don't understand what this has to do with blocking/non-blocking?



On 1/27/2011 3:24 PM, hockchai Lim wrote:
> I've a socket client application that calls the send() api to send some
> request data to the sever and then calls the recv() api to receive the
> result from it.  A fairly typical socket client program.  But because this
> is a critical application, I would like to have a way to ensure that all the
> result data that I'm getting from the recv() api calls are for the last very
> last send() request that I've just sent out.  So, I'm thinking, I should
> call the recv() api in non-blocking mode before I call the send() api to
> flush out all the recv() buffer.  Would this be a wise idea?
>
> thanks