On 12/28/2009 01:03 PM, Stefan Sperling wrote:
On Mon, Dec 28, 2009 at 12:21:34PM +0100, Miha Vitorovic wrote:
I'm just a lurker, so my apologies if I'm missing something obvious,
but isn't it much simpler to support client side hooks on a Windows
only client (like Tortoise clearly is)?
It's up to the authors of hook scripts to make sure they run on the
machines they need to run on.
True, but to make it useful from admin point of view (at least if I'm
the admin), is to have hooks on the server, so that each client
connecting to the repository gets them. Otherwise, what is the point of
hooks if it is up to the user to install them, if they want to? Such
hooks would be no different from wrapper scripts.
I mean, clearly a CLI svn client solution should support server
(admin) enforced client side hooks that can run on all the OSes that
the CLI client is capable of running on. And that doesn't seem like
a simple task to me.
Letting a Subversion server run code on a client machine would be a
gaping security hole. The client-side hooks would not be enforced from
the server side. They would be installed and configured on the client
machine they are supposed to run on. The server would not even know
about them, just like it doesn't know about the client configuration file.
I agree about the security implications, but see above about wrapper
scripts.
In enterprise environments admins often control what software is
installed on workstations and how it is configured, so they can
install and configure client-side hooks for their users.
Stefan
Miha
