Re: Subversion security design

> The fact that people haven't [added different access control
> strategies] suggests that it's not as important as we're often
> led to believe.

That may be true.

Another possibility is that CVS isn't and never has been truly distributed
(remoted yes, distributed no), and that there is no need for a distributed
access control mechanism in a non-distributed system. In fact, many people
seem to use cvs-acls, which is sufficient in the non-distributed case.

I'm not suggesting that anybody commit huge energy to building a distributed
solution. I'm merely suggesting that we should understand the requirements
lest we accidentaly design an interface or protocol that precludes solving
this later.

shap
Received on Sat Oct 21 14:36:05 2006