[Opensst-developer] Re: Proposition to externalize config variables

[<S.S...@au...>]

Pascal,

Pascal Steichen wrote:
> Well, as the client has to fill, set the config in his config file, I
> think the use of a simple properties file would be easier accepted and
> understood as a "complicated" xml file. XML is really powerfull, and I'm
> pretty impressed about your xml messages, but for a config file I think
> a properties file would do the thing, no need of a dtd, no need of
> updating syntax and/or style, etc... it's pure text !

I dont think a end user should ever directly modify this file.  A UI 
should be provided.  Therefore, the exact file format is not very 
revelant (as long as we are talking about the end user).
For us, developpers, I agree to say that any ascii file format will be 
easier to use that a binary format (serialized class)
Using XML allows to leverage the programming techniques used everywhere 
else within the programs.
:-)
Anyway ... as with many other subject, I guess we will have as many 
opinion as people involved :-)))


>>On the other hand, since we are in the context of a web application :
>>i really like the idea of having the user's properties stored on the
>>server (and the file being exchanged through opensst messages). This
>>solution solves all deployment issues we are facing 
> 
> 
> That sounds really good to me. The user properties could be stored on
> the server, and ciphered with his public key, so only he can use them.
> But I don't know if you want to go that far ?

I realize later last nigth that this was a silly proposition from my 
side.  The purpose of the config file is to give the proxy some config 
file some info that will allow it to, amongst other, connect to the server.
In this regard, the config file can not be stored on the server !
Unless, we are 100% sure that we won't ever need to store any connection 
related information.

I'd like to have the advise from other on that topic before choosing.

> The modif is in the OpenSSTServlet class (the proxy config on the client
> is handled by the JWS :)), here are the stuff I added : 

OK, that's another good reason to use JWS.  I never checked that aspect 
of JWS.  Really a good point.
On the other side : is there really a reason to use a proxy on the 
server-side ?

Thanks for the code.

> Eh, another thing, I heard that there is an initiative about testing
> interoperability of openSST and idx-pki (from idealx), do you know more
> about that, cause that would interset me aswell ?

I don't know the exact status of this.  Check that with Christophe 
Feltus and/or Alexandre Dulaunoy.

They are booth on the ope...@li... mailing list

Seb