[cvsnt] Latest Update (2.0.26) and ACL

Tony Hoyle tmh at nodomain.org
Wed Feb 11 14:49:26 GMT 2004


On Wed, 11 Feb 2004 07:41:15 -0700, Glen Starrett <grstarrett at cox.net>
wrote:


>Tony, in light of this scenario, what do you think?  I'm thinking I 
>agree with the amendment that any ACL with an unspecified branch should 
>apply to all.  Basically implying that the previous behavior had an 
>implicit "(main)" branch specified on it if no specific branch was 
>specified, and now it would be saying that no branch specified means 
>'all branches without specific permissions for this user | default'.

We have a number of permissions on the HEAD branch that don't apply to
branches (to stop the release people accidentally editing the HEAD
branch and vice-versa, which happened a few times).  To duplicate the
behaviour of HEAD to branches would break this completely.
 
It needs another solution... It seems even the simple 'default' fix
breaks things, and there isn't the language in the ACL mechanism to
talk about inheritence (eg. if one branch is derived from another,
should it take the permissions of its parent branch? Should there be
branch owners?).

I'm going back to the old way for now (since it's been like that since
it was first implemented and people are used to it).  If the ACL
mechanism is to be replaced it's best to do it properly.

Tony




More information about the cvsnt mailing list