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

[Albert York <alfromme@xxxxxxxxx>]

That approach should work - most of the time. However, if the response
doesn't come back fast enough your program will fail on the recv() and
not realize there is still data coming.

With the select() statement you can specify a timeout value. The
select() waits for a response from the socket indicating its status
and will return to your program when it receives a response or it
times out. If there is data waiting the select() returns immediately
and you know you need to recv() the data. If it times out then you can
continue. The timeout value is set in microseconds and/or seconds.

Look on Scott's web page for an explanation on how to use it.

Albert

On Fri, Jan 28, 2011 at 6:38 AM, hockchai Lim
<lim.hock-chai@xxxxxxxxxxxxxxx> wrote:
> Scott,
> It should always be a one to one relationship between the send and recv.  My
> application sends one command and will recv one respond back from the
> server.  Yes, in theory, I could send all three commands and then recv all
> three responds back at once.  But this senario will not happen in this
> application.
>
> The server is actually not a true socket server application.  It is actually
> a modem (v.35) type application.  The network guy puts a CISCO route, that
> speak TCP/IP, in between my socket app and the actual modem server.  So, my
> app actually speck to this CISCO route and it then sends the data to the
> target server.  Sometime if the CISCO route is not configured correctly,
> I'll see the respond being echo back to me (seeing the same respond twice).
> So, I'm kind of worry that this echoing problem might throw my app to become
> out of sync on this one to one send and recv relationship.
>
> Albert,
> I've never had a chance to use select() api.  So, I'll have to look more
> into this.  I was thinking recv() in non-blocking mode will flush this
> echoing problem and hence keeps my send() and recv() in sync.