Joshua Jensen [off-list ref] writes:
git rebase --onto mybranch START_SHA END_SHA
In the middle, I decide to run 'git rebase --abort'.
Just as the documentation states, it performs a checkout of END_SHA as
the restored branch. END_SHA has nothing to do with the originating
branch, and confusion ensues.
Is there a reason why 'git rebase' should not store off the
originating branch and use that for an --abort, instead of <branch>
which is END_SHA?
When END_SHA is an actual branch name (which by the way is almost always
how I use --onto in my attempts to transplant my topics), and when I found
out that the conflicts I see while rebasing the topic to a different
starting point (i.e. --onto) is too much to handle for too little gain, I
would not appreciate if --abort took me to the --onto branch, which is
totally uninteresting for the purpose of what I was attempting to do,
namely, to deal with the topic.
If the command took me back to the tip of the topic that I failed to
rebase, I can continue attempting to whip it in shape using different
strategies from there (e.g. merging an older part of upstream into the
topic before merging the topic back to the upstream).
----- Original Message -----
From: Junio C Hamano
Date: 10/22/2010 1:15 PM
Joshua Jensen[off-list ref] writes:
quoted
git rebase --onto mybranch START_SHA END_SHA
In the middle, I decide to run 'git rebase --abort'.
Just as the documentation states, it performs a checkout of END_SHA as
the restored branch. END_SHA has nothing to do with the originating
branch, and confusion ensues.
Is there a reason why 'git rebase' should not store off the
originating branch and use that for an --abort, instead of<branch>
which is END_SHA?
When END_SHA is an actual branch name (which by the way is almost always
how I use --onto in my attempts to transplant my topics), and when I found
out that the conflicts I see while rebasing the topic to a different
starting point (i.e. --onto) is too much to handle for too little gain, I
would not appreciate if --abort took me to the --onto branch, which is
totally uninteresting for the purpose of what I was attempting to do,
namely, to deal with the topic.
If the command took me back to the tip of the topic that I failed to
rebase, I can continue attempting to whip it in shape using different
strategies from there (e.g. merging an older part of upstream into the
topic before merging the topic back to the upstream).
As if turns out, I wasn't expecting --abort to take me to the --onto
branch but rather, the originating branch. Let's assume I am working in
a branch called JBranch. I run:
git rebase --onto master START_SHA END_SHA
I decide to abort the rebase. I expect to be returned to the place I
started from: JBranch. After all, I am aborting the rebase. I don't
want to proceed. I decided it was the wrong thing to do, so just put me
back where I was.
Anyway, that was my thought pattern. Since that isn't how it works,
I'll just have to remember where I was when I started the rebase,
perform the abort, and checkout the originating branch.
Thank you for your time.
Josh
Joshua Jensen [off-list ref] writes:
I'll just have to remember where I was when I started the rebase,
perform the abort, and checkout the originating branch.
"git checkout -" (or its generalization "git checkout @{-<n>}").
P.S. This is discounting whether there is bug in "git rebase --abort"
after using it with `--onto` or not.
--
Jakub Narebski
Poland
ShadeHawk on #git