Re: sockets library

From: tommysvensson (tommysvensson_at_[hidden])
Date: 2001-12-19 07:07:20

I've been involved in networked application programming for quite a 
while and as many other of you, I have written and used my own socket 
class library. My library was more or less just a wrapper class and a 
couple of helper classes similar to the socket-lib submitted by Rob 
Tougher.

I never really liked it. Please do not ask me why ;-) To me this is 
the wrong way of doing it. I've used the ACE framework in the last 
couple of projects and it sure eases the burden to develop safe, 
robust and portable network programs.

So, I completely agree with Jeremy!
It would be really cool to see the concepts of ACE brought into the 
Boost Sockets and maybe "carried even further". Would that be ACE on 
steroids? ;-)

Btw. The first book on ACE was published by Addison-Wesley this very 
week!  C++ Network Programming, Volume 1: 
http://www.amazon.com/exec/obidos/ASIN/0201604647
Also, check out the POSA2 book (networked app. patterns realized with 
ACE): http://www.amazon.com/exec/obidos/ASIN/0471606952

/Tommy

--- In boost_at_y..., Jeremy Siek <jsiek_at_c...> wrote:
> 
> I still feel strongly that the C++ type system should be
> used to prevent typical errors when using a socket library.
> The process that one has to follow with the C/POSIX API can
> be encoded in classes and objects, thereby ruling out lots 
> of mistakes. The ACE framework does this to a certain extent,
> and I'd like to see this carried even further. The following 
> message contains a prototype of this kind of interface.
> 
> http://groups.yahoo.com/group/boost/message/15273
> 
> Cheers,
> Jeremy
> 
> -- 
> Jeremy Siek                          http://www.osl.iu.edu/~jsiek
> Ph.D. Student, Indiana Univ. B'ton   email: jsiek_at_o...
> C++ Booster (http://www.boost.org)   office phone: [phone]