From: Jonathan Nieder <jrnieder@gmail.com>
Date: Fri, 27 Nov 2009 15:33:49 GMT
Subject: Re: Breaking expectations in 1.7.0, was Re: What's cooking in git.git (Nov 2009, #06; Wed, 25)
Message-ID: <20091127153349.GA24647@progeny.tock>
In-Reply-To: <alpine.DEB.1.00.0911271536100.4521@intel-tinevez-2-302>

Johannes Schindelin wrote:

> I recently read more and more about 1.7.0 being expected to break 
> expectations, and more and more expectations about more and more being 
> expected to be broken there.
> 
> This is a very slippery slope.  You have been warned.

I was trained by 1.6.0 and 1.5.0, I guess.  I think it would be hard to
argue that 1.5.0 breaking compatibility in some places (the behavior of
'git add', for example) was a bad thing.  I also liked the major 1.6.0
change, but I did have the benefit of being warned far in advance.

Of course, the best break with compatibility is the one that does not
break expectations.

I’m on the fence about the default git grep behavior for that reason.
I think --full-tree really would be a better default.  Every once in a
while, I really do find myself wondering why git grep has not given me
the results I expected; it can take a moment to remember that this is
because I am in a subdirectory.  But it is not clear it is worth the
change. 

Am I right to infer you are less conflicted about this, or was it just
a good opportunity to bring to mind that important principle?

Jonathan