The Jelmer "fork" of bzr.dev

[Jelmer Vernooij]

Russel Winder wrote:
> On Sun, 2009-04-12 at 10:54 +0300, Alexander Belchenko wrote:
>   
>> Jelmer Vernooij пишет:
>>     
> [ . . . ]
>   
>>> Or we could (yuck) special-case google code.
>>>       
>> This way contradicts to the Zen of Python:
>>
>> 	Special cases aren't special enough to break the rules.
>>     
>
> True, but what if the rules are wrong?
>
> In this case the user (me :-) knows that the repository at the end of
> the line is a Subversion repository and is being accessed with HTTP so
> requires no authentication.
>
> The fact that Bazaar is trying to see if the HTTP is actually a Bazaar
> repository before seeing if it is a Subversion repository, violates my
> expectations.  If there is no authentication check then there is nothing
> observed by the user and so no problem.  Where there is an
> authentication check user expectation is violated.
>
> So the core of the problem here is the removal of the svn+http: protocol
> (which made everything explicit) in favour of just having http: (and
> therefore having to introduce reflection on the remote connection).
>
> The appears a strong prima facie case for reverting the decision to
> remove svn+http.
The core of the problem is the fact that Bazaar doesn't behave as
expected with http:// and prompts the user before it's checked all
formats to see if there's one that works without authentication.

svn+http:// is just a workaround for the problem. In the end, we should
work on making http:// work properly, as that is the first thing that
users will expect to work.

Anyway, I've reverted the removal of svn+http(s):// for now in the 0.6
branch.

Cheers,

Jelmer