From: Eric Blake <ebb9@byu.net>
Date: Mon, 11 May 2009 20:04:25 GMT
Subject: Re: [PATCH 0/6] Add core.autocrlf=true on cygwin by default during tests
Message-ID: <loom.20090511T195910-370@post.gmane.org>
In-Reply-To: <1242070141-2936-1-git-send-email-Don.Slutz@SierraAtlantic.com>

Don Slutz <Don.Slutz <at> SierraAtlantic.com> writes:

> 
> This is a patch set to do the work from mail message:
> 
> http://kerneltrap.org/mailarchive/git/2007/8/7/254415
> 
> It is split into 6 parts.  This is because the tests do not work on
> Linux if autocrlf=true.
> 
> 1) The change to autocrlf=true by default for cygwin

I help maintain the cygwin packaging of git, and I don't particularly like 
patch 1/6.  Cygwin very much recommends that users stick with binary mounts, 
where cr's are not inserted/stripped by default, and enabling autocrlf in that 
situation is asking for problems (namely, the same set of problems as you would 
have on Linux with autocrlf=true).  While adding an environment variable is 
useful for the case of a user building git on cygwin with the intent of using a 
text-mode mount, it goes against the grain of cygwin's recommendation of not 
using text-mode mounts in the first place.  I would rather see the default 
remain false, even on cygwin, unless the person building git takes steps to 
activate it themselves.

-- 
Eric Blake