squashing commits all the way to the initial leaves 2 commits

Question

I am learning git and trying out squashing. I'd like to squash all commits on the master and be left only with initial one. For example, my commit history looks like this:

2c825e339b60702b7b48c2ea022e473341d89a7d (HEAD -> master) another edit
f676a10db6a0cfe3cebb8f84aad971b481483181 second file
dc55b957a94bc4a3d6c1b6d8c134407aa4c8316c initial

After that I ran git rebase -i dc55b957a94bc4a3d6c1b6d8c134407aa4c8316c

The usual way described all over the net is to replace every pick with squash, but the first one. After doing that I am presented with this commit history:

2067070140c38f8ecf5f70894e6267bfee614d85 (HEAD -> master) second file dc55b957a94bc4a3d6c1b6d8c134407aa4c8316c initial

How come the squash didn't work all the way to the initial commit?

Moreover, if I run the same squash command again, there is only 1 pick statement in the interactive window. Changing it to squash triggers error and leaving it to pick does nothing. I have found

git reset --soft HEAD~1
git commit -m "1.0.0"

to be an effective way of squashing only 2 commits. How exactly can I use the git rebase --squash to achieve the same result?

Answer 1

The commit you specify in the rebase command is the parent of the first commit you want to modify or move. This is difficult for the first commit of the repository as it doesn’t have any parents. Instead you can use the --root option:

git rebase -i --root

This should give you all the commits.

Answer 2

You had to do git reset --soft HEAD~2 and then do git commit --amend -m 1.0.0