On Thu, Feb 28, 2008 at 6:26 PM, Johannes Schindelin<Johannes.Schindelin@xxxxxx> wrote:> Hi,>>> On Wed, 27 Feb 2008, Nguyễn Thái Ngọc Duy wrote:>> > This was impossible earlier because git_dir can be relative. Now that> > git_dir is absolute, I see no reason for worktree setup inside> > setup_git_directory_gently().>> I do see it, though. Why make the users work harder? To make clear separation of worktree usage? I think it's good to addsetup_work_tree() where you really need it. Right now in commands thatuse setup_git_directory_gently(), it is (to me) really hard to tellwhether an option require worktree. Another point of removing worktree setup from _gently(): worktreesetup can die() while _gently()'s purpose is, well, do it gently. Ifit's not safe, let the caller decide. Also having to deal with prefix while you don't need worktree at allis a bit of work that can be eliminated with this series. > If you want to get> the git directory, chances are that you want to work with a worktree, too. Yes. That's why I keep worktree setup in setup_git_directory(). > And you really cannot properly separate worktree detection from git> directory detection: in most of the cases, you will find them at the> _same_ time (if .git/ is the git directory, the working directory is .). No I don't separate worktree detection. I only avoid moving toworktree unless necessary. Right after _gently() you can tell if youhave a worktree and where it is. > So I am mildly negative on the thrust of your patch series.>> Ciao,> Dscho> -- Duy���m�ka����zwm���˛��m���ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
