Re: D-BUS versus GDOMAP (WINDOWS users please note)

[Rogelio Serrano]

On 2004-08-31 15:51:45 +0800 Richard Frith-Macdonald 
<richard@brainstorm.co.uk> wrote:

> On 31 Aug 2004, at 08:04, Rogelio Serrano wrote:
>
> > I think DO is more like the libdbus layer not the message bus 
> > layer.
> Yes, the encoding and transport part of DO (eg NSPortCoder and 
> NSSocketPort) 
> is conceptually similar to libdbus,
> While the nameserver part of DO is (very roughly) comparable 
> to message bus, 
> it's actually quite different.
[snip]
> That sounds like a good idea.   The main issue (after 
> implementing basic 
> functionality to start/stop services by name of course), is 
> how to combine 
> ease of use with security.  While dbus provides a specific 
> security protocol 
> to authorise connections via a variety of mechanisms, this 
> really rather 
> misses the point.  Passing authentication tokens and 
> encrypting stuff is 
> fairly straightforward (the GNUstep DO system can already do 
> it) ... what's 
> important is working out how to easily configure the processes 
> to have the 
> correct security tokens and enforce the security policies.  We 
> don't have any 
> way to do that now, and d-bus wouldn't help.
>
> IMO what would be good would be to write a proxy class to 
> handle security 
> issues over DO (a server would only vend these proxies rather 
> than vending 
> objects directly), and write easy to use gui and command line 
> tools for 
> configuring policy and security tokens for applications.  The 
> NSConnection 
> class could then be trivially modified to check that, when 
> setting a root 
> object for the connection, the object was an instance of the 
> security proxy 
> class.

Hmmm... DTE? Amoeba like capabilities? A user space security 
server? Selinux?

--
Blood is thicker then water... And much tastier
                            John Davidorff Pell