On Tue, Mar 11, 2008 at 10:08 PM, Johannes Schindelin<Johannes.Schindelin@xxxxxx> wrote:> Hi,>>> On Tue, 11 Mar 2008, Nguyen Thai Ngoc Duy wrote:>> > On Tue, Mar 11, 2008 at 8:28 PM, Johannes Schindelin> > <Johannes.Schindelin@xxxxxx> wrote:> >> > > On Sun, 2 Mar 2008, Nguyn Thái Ng÷c Duy wrote:> > >> > > > Signed-off-by: Nguyn Thái Ng÷c Duy <pclouds@xxxxxxxxx>>> > >> > > This fix is completely independent of the rest of your series.> >> > Not really. It would have no impact (good or bad) if sent separately.> > But it is required by the series (otherwise it would die() under certain> > condition because prefix cannot be recomputed properly).>> Well, it _is_ independent. Either you say "we want to have this as an> example where you can call setup_git_directory() twice" (which I would not> find all that bad), or you say "this was calling setup_git_directory()> twice, fix that".>> In any case, it should be separate from this series.
My point stands still. It is merely cleanup without harm or goodnessif seperated from the series. This patch does not fix the realproblem.
> Back to the subject of this series: as I stated already, I do not like the> separation of work-tree and git directory, and I especially do not like> how much you had to work outside setup.c. I think changing the semantics> for all callers was not necessary, and in fact, not desirable.
For one, setup_work_tree() prototype change is necessary to deal with(gitdir autodetection + core.worktree + setup_work_tree()).
For setup_*gently() changes, I think I can makesetup_git_directory_gently() retain its current semantics by removing06/10 and 07/10. But I do think it's nice to have setup_*_gently()does things gently _and_ easily-predictable (current _gently() setupsworktree under certain conditions and not otherwise, which I consideruneasily-predictable). Patch 06 and 07 are mainly cleanup as a resultfrom _gently() changes. Are those most intrusive to you?
> Ciao,> Dscho>-- Duy��.n��������+%������w��{.n��������n�r������&��z�ޗ�zf���h���~����������_��+v���)ߣ�m
