John Peacock <jpeacock@rowman.com> writes:
> I'm not saying the patch is anything except a way to ensure that _upon
> creation_ the repository has the correct ownership and permissions
> under most Unix-like filesystems. You can also view it as a way to
> explicitly set the permissions correctly, as opposed to the current
> behavior which sets the permissions based on the running user and
> their umask.
I know what you are saying. And what I'm saying is, don't add
complexity to our system and hand a false sense of security to our
users under the guise of helping them. It's not helpful. You *are*
going to see just as many bug reports as before, except instead of "I
can't access my repository," they'll be "I can't access my repository
even though I created it with --owner and --user. Why didn't this
work?"
> And supporting that is on my todo list (with help from Brane
> probably). It is a much bigger issue to deal with it in the Win32
> world (where permissions are in some ways more flexible and yet even
> more likely to get messed up in non-trivial ways).
Great. I'll look forward to your patch when it is complete.
> The current docs on permissions and ownership are of limited use under
> Unix, and of no use under NTFS (IMNSHO). I am comfortable under
> various *nix variants, yet the existing documentation took me some
> thinking to know how to do it right the first time. Just because I
> have [lots of] Unix experience doesn't mean I've had to do a great
> deal of set up for shared database file access (which is a much more
> specialized use-case).
Dude, then we just need to fix the docs. We don't need to start
throwing options at the problem. If the problem at its core is one of
user ignorance, then the solution is education -- plain and simple.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Nov 24 18:10:08 2003