Re: Overriding ~/.gitconfig using GIT_CONFIG

On Fri, 2011-08-12 at 12:39 -0700, Junio C Hamano wrote:

Junio C Hamano [off-list ref] writes:

quoted

Richard Purdie [off-list ref] writes:

quoted

Looking through the manuals/code, it suggests I should be able to do:

GIT_CONFIG=/dev/null git XXX

and all should work happily. It doesn't though. As an example, with a
~/.gitconfig, "GIT_CONFIG=/dev/null git fetch --all" is clearly
accessing the file in ~ and then acting upon it.

If the manual says the above is expected for any value of XXX, then that
is a bug in the manual since mid 2008, I think.

See dc87183 (Only use GIT_CONFIG in "git config", not other programs,
2008-06-30).

I _think_ these days a workaround to force a known config is to set HOME
to a value that has a known .gitconfig (or no such file), and decline
usage of /etc/git.config by exporting GIT_CONFIG_NOSYSTEM.

Side note. Here is what dc87183 says:

commit dc87183189b54441e315d35d48983d80ab085299
Author: Daniel Barkalow [off-list ref]
Date:   Mon Jun 30 03:37:47 2008 -0400

    Only use GIT_CONFIG in "git config", not other programs
    
    For everything other than using "git config" to read or write a
    git-style config file that isn't the current repo's config file,
    GIT_CONFIG was actively detrimental. Rather than argue over which
    programs are important enough to have work anyway, just fix all of
    them at the root.
    
    Also removes GIT_LOCAL_CONFIG, which would only be useful for programs
    that do want to use global git-specific config, but not the repo's own
    git-specific config, and want to use some other, presumably
    git-specific config. Despite being documented, I can't find any sign that
    it was ever used.
    
    Signed-off-by: Daniel Barkalow [off-list ref]
    Signed-off-by: Junio C Hamano [off-list ref]

It clearly explains the reason why LOCAL_CONFIG was removed (the reader
does not have to agree with "I can't find any sign that it was ever used",
though), but I cannot read from the first paragraph the reason why it was
felt necessary not to honor GIT_CONFIG in other programs, i.e. "was
actively detrimental" is not backed by any example in the paragraph. I can
sort of sense from "Rather than argue over..." that there may have been a
discussion on the list, and reading the archive from that timeframe may
reveal why many felt it was not a good idea.

Daniel, do you recall the context?

I went digging and this looks like as good a summary as any of the posts
around that time:

http://marc.info/?l=git&m=121476432303314&w=2

It sounds like if you specified GIT_CONFIG when making a clone it would
end up writing the config file specified rather than .git/config.

My problem isn't that I want to specify a specific .gitconfig file, I
just need it to ignore the one in $HOME. I'm happy for the .git/config
file to be used, in fact I need it to be.

I noticed 8f323c00dd3c9b396b01a1aeea74f7dfd061bb7f was committed which
removed GIT_CONFIG_NOGLOBAL support which is the other way to address
the problem. Could we add that back?

I appreciate I can set $HOME to something but that means creating an
empty directory to point at and feels rather like a work around rather
than a solution.

Cheers,

Richard

-- 
Linux Foundation
http://www.yoctoproject.org/