Re: Scripted clone generating an incomplete, unusable .git/config

From: Stefan Naewe <stefan.naewe@atlas-elektronik.com>
Date: Nov 11, 2010, 07:55 UTC
Message-ID: <4CDBA15A.6090301@atlas-elektronik.com>
In-Reply-To: <AANLkTik7-QzrMKDpV=W4dqpuguZsAr5yrMELmHu5NZMd@mail.gmail.com>

On 11/11/2010 12:21 AM, Dun Peal wrote:

Show 36 quoted lines

> This is a weird issue I ran into while scripting some Git operations
> with git 1.7.2 on a Linux server.
> 
> When running the git-clone command manually from the command line, the
> resulting repo/.git/config had all three required sections: core,
> remote (origin), branch (master).
> 
> When running the exact same git-clone command manually from the Python
> scripted, the resulting repo/.git/config was missing the `core` and
> `remote` sections.
> 
> Here's a bash log fully demonstrating the issue:
> 
>   $ python -c "import os; os.popen('git clone
> git@git.domain.com:repos/repo.git')"
>   [...]
>   $ cat repo/.git/config
>   [branch "master"]
>           remote = origin
>           merge = refs/heads/master
>   $ rm -Rf repo
>   $ git clone git@git.domain.com:repos/repo.git
>   $ cat repo/.git/config
>   [core]
>           repositoryformatversion = 0
>           filemode = true
>           bare = false
>           logallrefupdates = true
>   [remote "origin"]
>           fetch = +refs/heads/*:refs/remotes/origin/*
>           url = git@git.domain.com:repo/repo.git
>   [branch "master"]
>           remote = origin
>           merge = refs/heads/master
> 
> What's causing this?  Is it a bug?

Same for me with git version 1.7.3.2 on Debian Etch. Seems to be a problem with the popen() returning too early or the interpreter dying too early.

This works though:

$ python -c "import subprocess; subprocess.call(['git', 'clone', 'git://host/repoo.git'])"

Regards,
  Stefan

-- 
----------------------------------------------------------------
/dev/random says: I is knot dain bramaged!