help: shared mainline over ssh

[Emanuele Olivetti]

Eric Siegerman wrote:
> ...
> It sounds as though what you're aiming for is more a "centralized
> with local commits" workflow, but without requiring users to type
> "commit --local" most of the time or to do the unbind/rebind
> dance.  Is that correct?  On the assumption that it is, most of
> your plan looks OK, with the following exceptions.
>
>   

Thanks a lot for answering.

>
> One possibility is to have your users use "bzr merge --pull"
> instead.  

Excellent. I wasn't aware of it, I'll forward to the users of
my repositories.


> ...
>
> As an aside, there's another problem too, but it has nothing to
> do with bzr pe se:
>   
>> # now I'm on to the server...
>> chmod -R g+rwxs,o-rwx .  # the sticky bit will ensure same
>>     
> permissions 
>   
>> for future files created by other users
>>     
>
> Umm, are you *sure* you want to set all the files setgid?  Far
> safer would be to do that only for the directories:
>     chmod -R g+rwx,o-rwx .
>     find . -type d | xargs chmod g+s
>
>   

Yes you are right, this is another detail I missed. Thanks
again!


I have one further note. I guess that my case is definitely not infrequent,
i.e., small group of users, easy ssh access, preferring local commits
and pull/push from time to time, no real branches. I use sometimes git and
this simple setting is described here as the basic way to set up a server:
http://progit.org/book/ch4-2.html

I'm wondering why there is nothing like that in bzr docs, I mean at this 
level
of detail. Of course you can figure it out from the concepts in the user
guide etc., but then the lack of step-by-step instructions leads to mistakes
(as in my case).

Kind Regards,

Emanuele