[PATCH/RFC 0/3] setup: stop ignoring GIT_WORK_TREE (when GIT_DIR is unset)

From: Jonathan Nieder <jrnieder@gmail.com>
Date: Jan 19, 2011, 12:37 UTC
Message-ID: <20110119123732.GA23222@burratino>
In-Reply-To: <7v1v4aknij.fsf@alter.siamese.dyndns.org>

Junio C Hamano wrote:

> Jonathan Nieder <jrnieder@gmail.com> writes:

Show 13 quoted lines

>> +++ b/setup.c
>> @@ -419,6 +419,11 @@ static const char *setup_discovered_git_dir(const char *gitdir,
>>  		return NULL;
>>  	}
>>  
>> +	if (getenv(GIT_WORK_TREE_ENVIRONMENT)) {
>> +		warning("GIT_WORK_TREE without explicit GIT_DIR is deprecated");
>> +		return setup_explicit_git_dir(gitdir, cwd, offset, nongit_ok);
>> +	}
>> +
>
> My knee-jerk reaction is that calling this "deprecated" is probably
> confusing. git merely failed to notice misconfiguration

After staring at test results, I agree. The rule, before nd/setup, seems to be something like this:

 - if GIT_DIR is set, you're safe
 - if GIT_DIR is unset and .git is in the current directory,
   GIT_WORK_TREE will work by accident
 - otherwise, git errors out.

The patch below should be more robust.  It prints a warning:

	warning: pretending GIT_DIR was supplied alongside GIT_WORK_TREE

As the night winds on I am less sure that that warning is a good idea. The .git discovery behavior seems relatively safe. What is more worrisome for the future of the use of GIT_WORK_TREE independent of cwd is the interaction with pathspecs.

Jonathan Nieder (3):
  tests: cosmetic improvements to the repo-setup test
  tests: make the setup tests briefer
  setup: stop ignoring GIT_WORK_TREE and core.worktree

 setup.c               |   27 +-
 t/t1510-repo-setup.sh | 5166 +++++++------------------------------------------
 2 files changed, 729 insertions(+), 4464 deletions(-)