I erred by working locally on a local repo’s "main" branch, instead of branching out with a new branch and modifying code there. I brought more modifications to the code than I thought I would, all without pushing anything to remote "origin". A quick $ git reflog --date=iso | grep pull yielded the hash that corresponds to the pull from "origin/main", from which I should have branched out in the first place (and didn’t), say it’s "D".
Before restoring my local "main" to the prior state of interest, "D", I created a "new-branch". Now I have:
E' (new-branch)
/
B--D--(E) (main)
So currently E’ reflects all changes locally made to D. What I would like is to end up with all changes reflected in E’ to be as if they :
E' (new-branch)
/
B--D-- (main)
Not sure that the two schematics above are fully correct representation. In any case it’s all local. Would doing
$ git checkout main
$ git checkout -b new-branch
$ git checkout main
$ git reset --soft <sha1_for-D>
revert all changes to previous state "D" on my local "main", while preserving changes stored in "new-branch" as "E’". But is this strictly equivalent to
$ git checkout main
$ git switch -c new-branch
$ git checkout main
as suggested here?
Continuing to work on "new-branch" will eventually lead to a merge into "main" after testing. Will that future "merge" unfold any differently from a merge that would have occurred with no soft reset on "main" ?
>Solution :
Your chart is a bit misleading, as there seems to be no commit E'. After git checkout -n new-branch both branches point to the same commit E like this:
C--D--E (main) (new-branch)
If you then checkout main and reset it to commit D as you proposed this will result in
E (new-branch)
/
C--D (main)
You might want to consider using git reset --hard or at least switch back to the new-branch immediately after the reset to not have this somewhat confusing state of the repository with lots of uncomitted local changes that git reset --soft leaves behind.