mina-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Emmanuel Lécharny <elecha...@gmail.com>
Subject Re: [MNA 3] SSL support
Date Thu, 03 May 2012 11:52:44 GMT
Le 5/3/12 12:11 PM, Emmanuel Lécharny a écrit :
> Le 5/2/12 8:00 PM, Julien Vermillard a écrit :
>> +1
> Ok, I have committed some changes, but I'm still not pleased with what 
> we get at the end.
> The problem is that in order to get a TCP session secured, we have to 
> store the SSLContext somwhere. If we stores it into the 
> AbstractTcpServer/Client, then we stores it in two different places. 
> The very same for the initialization : we have to implement the init 
> code in two classes...
> If we want to have this filed and method shared, we should have them 
> in the AbstractIoService, but then, it will be visible for the UDP 
> classes...
> Another solution would be to get rid of this SSLContext field, and to 
> move it into the IoSessionConfig TCP implementation...
> I do think, atm, that it would be a better solution.
> thoughts ?
> Btw, the IoSessionConfig hierarchy is a bit limited, we may want to 
> improve it by creating a TcpSessionConfig and an UdpSession config 
> instead of the DefaultSocketSessionConfig...
Some more thougts : securing the session has nothing to do with the 
service itself. At this point, having a method like 
IoService.isSecured() makes no sense.

I think I added this method whith things like HTTPS or LDAPS in mind, 
and it was a bad move. It does not reflect what's going on : first the 
SSL layer works on a session, second with StartTLS, you can perfectly 
well switch from a non secured to a secured session, live.

I'll remove the isSecured() and setSecured() methods from the IoService.

Emmanuel Lécharny

View raw message