On Fri, Feb 26, 2010 at 11:44 PM, David Brodbeck
<brodbd_at_u.washington.edu> wrote:
> On Feb 26, 2010, at 2:03 PM, Bob Archer wrote:
>> Baring that setting up apache for anon requests and svn for authenticated.
>
> That would also work. In our case we didn't really want the added complexity of Apache -- plus I was worried about user confusion. I already have a fair number of people who get http: and svn: mixed up from time to time. Having http: work in some situations but not others would just add to their perplexity.
You can also put them both on Apache, then there's not that much
confusion. That's how we do it: two Location blocks in Apache, both
backed by the same repository:
- http://svn.example.com/public_svn: the anonymously accessible one.
Goes over plain http, doesn't require authentication, but uses an
authz file to limit the paths that can be read.
- https://svn.example.com/svn: the "real" one for the developers. Goes
over https, requires authentication, and doesn't use an authz file
(allowing us to set "SVNPathAuthz off", which avoids the performance
loss normally associated with this setup).
Of course you can also have them both over https (it doesn't really
hurt the anonymous one).
Johan
Received on 2010-02-27 00:42:47 CET