From: Wincent Colaiuta <win@wincent.com>
Date: Mon, 05 Nov 2007 15:26:36 GMT
Subject: Re: [PATCH] Implement selectable group ownership in git-init
Message-ID: <5C7AE6D3-5846-4E04-85CA-75B6C2108504@wincent.com>
In-Reply-To: <472F3212.4090800@gmail.com>

El 5/11/2007, a las 16:09, Francesco Pretto escribió:

> Wincent Colaiuta ha scritto:
>>
>> using the administration tools designed for managing access, just  
>> like
>> every other SCM (and every server, and every piece of software  
>> which can
>> be accessed by many on a multi-user system).
>>
>
> I don't agree in general: in SCMs and other multi-user softwares,  
> the access
> control configuration can be safely postponed just because it's in  
> their
> standard usage pattern that the access should be conditioned by a  
> daemon
> to be configured later. It's not the case of git, just because git  
> is very
> tied to *nix permissions.
> But as it is now, it could seems that it's good to put committers in  
> the (for
> example) git group, just because you have a git administrative account
> git:git . This is caused, imo, by the fact that the flow of creating  
> a shared
> repository for a specific work/project group with git-init run by an
> administrative user (as it should be) is something like this:
>
> 	- Do it wrong;
> 	- Fix it immediately.
>
> I don't like the "Do it wrong" part. I'm trying to produce a sane and
> transparent patch to implement the selectable group just in case of  
> repository
> first initialization. Why do I care so much of first time users?  
> Dunno, but
> I think it's important.

What's stopping you from using "sudo -u"?

Wincent