Re: [PATCH] Docs: git checkout --orphan: `root commit' and `branch head'
- From
Michael Witten <mfwitten@gmail.com>
- Date
- Sep 28, 2011, 15:06 UTC
- Message-ID
- <CAMOZ1Bs5GKE19nd+9sVQuEH+zRCug0OTGXAqpDjEgq=_pKzZxA@mail.gmail.com>
- In-Reply-To
- <vpq39fglo8y.fsf@bauges.imag.fr>
On Wed, Sep 28, 2011 at 14:45, Matthieu Moy <Matthieu.Moy@grenoble-inp.fr> wrote:
Show 26 quoted lines
> Michael Witten <mfwitten@gmail.com> writes: > >>> Even if you use it normally: >>> >>> # I want to create a new root commit with a different tree >>> git checkout -b new-project >>> # hack for a while >>> git status >>> # ?!? why are files shown as 'deleted', 'moved', 'modified', I said I >>> # was creating a new project! >>> >>> Also, with your proposal, we would need to add two flags to "commit" >>> (--no-parent and --force), which is one of the first command beginners >>> learn, while in the current state we have just one for "checkout" to do >>> the trick, and newbies do not use or read the doc for checkout, so it's >>> not scary for them. >> >> Well, those are mildly interesting scenarios. I can only say that I don't >> think we should cater to people who either have amnesia or work casually >> on a repository for short bursts every few months or so. > > How is the "git status" issue above linked to any kind of amnesia? > > When hacking to create a rootless commit, it seems legitimate to me to > run "git status", and it seems _very_ confusing that "git status" still > refers to the commit you don't want as parent.
I think "confusing" is perhaps the wrong word. How about "annoying"?
I suppose it is true that neither "--orphan" nor "--no-parent" is good enough alone. For instance:
# I want to create a new root commit with a slightly different tree git checkout --orphan new-project # hack just a bit here and there git status # ?!? My slight alterations are obscured by hundreds of these # "new file" status notifications!
> (I don't get the relation between your reply and the paragraph right > above it either)
Is that a passive aggressive French way to state that I didn't respond to your concern?
I think it's moot now, anyway, given that both "--orphan" and "--no-parent" quite possibly have their logical purposes. However, perhaps:
git checkout --orphan
should be renamed:
git checkout --no-parent
in order to match:
git commit --no-parent