Re: [PATCH v4 0/5] Patches to avoid reporting conversion changes.

>
> On Mon, 7 Jun 2010, Finn Arne Gangstad wrote:
>
>> I think it would be best if git at this time could decide that the
>> affected files also become dirty. The ideal commit is one that
>> both alters the .gitattributes _and_ the affected files at the same
>> time, and git should make it easy to create that commit.
>
> I agree in the case of newly added attributes. In the case of 
> repositories already containing unnormalized files this however leads to 
> problems.
> eg
>
>   Consider the case above, but a while later when the repository has been
>   fixed at HEAD. If an old version from before the normalization is
>   checked out, the index will once again become dirty, which means that
>   git will refuse the user to check out some other version unless the
>   --force flag is given. Excessive use of --force is not a good thing.
>   If the user is aware of the problem, and checking out old versions is
>   a common operation, toggling the suggested option might be a good
>   solution.