I have a CMS theme installed on my machine. I'm tracking changes to it via git
and decided to back it up on GitHub so I could share those changes.
The theme as provided is also available on GitHub. On my machine I have added
this as a remote upstream. Now I can easily see the changes between my master
and the remote upstream by using the following command:
git diff --color master upstream/number
If I could add the remote upstream on GitHub I could easily share these changes.
Is it possible to set this relationship on GitHub?
I have tried the following:
git push -u origin upstreambranch
which adds an upstreambranch to the master on GitHub. However trying to
compare both branches doesn't work, the result I get on GitHub is that: "There
isn't anything to compare"
Is there an alternative way to compare these?
If the problem is "main and master are entirely different commit histories.", the following will work
git checkout master
git branch main master -f
git checkout main
git push origin main -f
The Short Answer
It looks like GitHub won't let you compare the branches because they don't
actually share any of the same history at all, even though they may share
much of the same files and code.
Here is a screenshot of the temporary fork I made of your repo, where I tried to
compare master with the upstreambranch, like you described. Notice the error
message:
It says:
There isn't anything to compare.
master and upstreambranch are entirely different commit histories.
The Long Answer
You probably downloaded the original source and added it to a completely new
repo instead of cloning the original repo, right? Doing that will make it so
that the history of your repo will be completely different from the
history of the original repo, since your new repo won't have any of the same
commits with the same sha IDs.
You can see that by doing a reverse log of your master branch and the
upstreambranch:
# Your first commit, see commit sha
git log --reverse master
commit c548d7b1b16b0350d7fbdb3ff1cfedcb38051397 # <== HERE
Author: Padraic Stack <padraic.stack#nuim.ie>
Date: Wed Apr 2 15:11:28 2014 +0100
First commit of everything
# First commit sha of the original repo
git log --reverse upstreambranch
commit 105a12817234033c45b4dc7522ff3103f473a862 # <== THERE
Author: Jeremy Boggs <jeremy#clioweb.org>
Date: Mon Feb 22 16:00:53 2010 +0000
Creates repo directories for the Seasons theme.
Solutions
If you redo your commits on top of the original history, you should then be able
to compare the branches. There are several different ways that you can redo your
commits, including
git rebase --onto
and
git cherry-pick
You also can redo each commit manually, if you have to.
I had a similar situation, where my master branch and the develop branch I was trying to merge had different commit histories. None of the above solutions worked for me. What did the trick was:
Starting from master:
git branch new_branch
git checkout new_branch
git merge develop --allow-unrelated-histories
Now in the new_branch, there are all the things from develop and I can easily merge into master, or create a pull request, as they now share the same commit hisotry.
I solve my issue using these commands
git checkout [BRANCH]
git branch master [BRANCH] -f
git checkout master
git push origin master -f
You can force update your master branch as follows:
git checkout upstreambranch
git branch master upstreambranch -f
git checkout master
git push origin master -f
For the ones who have problem to merge into main branch (Which is the new default one in Github) you can use the following:
git checkout master
git branch main master -f
git checkout main
git push origin main -f
The following command will force both branches to have the same history:
git branch [Branch1] [Branch2] -f
From the experiment branch
git rebase master
git push -f origin <experiment-branch>
This creates a common commit history to be able to compare both branches.
This looks like undesirable behavior on github's part, but it's fairly easy to fix. What you want to do is to rebase your branch on a reasonable (any reasonable) commit in the existing history. What you can do is to fetch the github repo and find which tree in its history is most similar to the one you started with. Start this way:
git remote add github u://r/l
git fetch github
myroot=`git rev-list master --max-parents=0`
root_tree=`git rev-parse $myroot^{tree}`
github_base=`git log --pretty=%H\ %T github/master | sed -n "s/$root_tree//p"`
With any luck, that will find you a commit in the github history that has the exact tree you started with. Assuming it does,
git rebase --onto $github_base $myroot master
and you're done.
If that doesn't find a matching tree, you get to find a nearest approximation. Here's one way to get a rough estimate of the differences:
git log --pretty='echo %H $(git diff-tree -p -b -U0 '$myroot:' %T|wc -l)' github/master \
| sh
which will count the lines in a minimized diff between the tree of each commit in the github/master history and your root tree. It seems reasonable to hope for a nice small difference, you could eyeball the actual diffs on it before calling that the github_base commit and doing the rebase above.
Terminology
First, let's get some terminology out of the way...
upstream <= The remote git repo (likely whose master or release branch is in production)
forked-repo <= The remote [experimental git repo] (https://docs.github.com/en/github/getting-started-with-github/fork-a-repo) also known as "origin".
local repo <= The files and directories that you work with on your local workstaion, which you likely got by running a git clone my-forked-repo.git command
local index <= Also known as your local git "stage", i.e., where you stage your files before pushing them to you remote repo.
Github workflow process
Next, let's talk about the process of getting your changes to the upstream repo:
The process is generally to work on a feature branch and then push said branch, and open a Pull Request, either to your forked-repo's master branch or to the upstream's master branch
Create a feature branch by running git checkout -b FEATURE_BRANCH_NAME
Add/delete/modify files project files.
Add files by running git add .
Commit your files to your index by running git commit -m'My commit message'
Push your staged files by running git push origin FEATURE_BRANCH_NAME
Solution for entirely different commit histories
The master and upstreambranch are entirely different commit histories message can occur when you've forked a git repository and have changed your git history.
For example, if you fork a repo and pull your forked repo to work on it locally...
If then you decide to rewrite the entire application and then decide it's a good idea to deleting all existing files, including the forked-repo's .git directory. You add new files and directories to recreate your app and also recreate your .git directory with git init command.
Now, your application works great with your new files and you want to get it merged into the upstream repo. However, when you push your changes you get that "...entirely different commit histories..." error message.
You'll see that your original git commit will be different in your new local directory and if in your remote fork (as well as your upstream). Check this out by running this command in your current directory: git log --reverse master. Then running the following: pushd $(mktemp -d); git clone https://github.com/my-forking-username/my-forked-repo.git; git log --reverse master; popd
You must fix your local .git repo to match your remote my-forked-repo if you want to push your commits and subsequently perform a pull request (in hopes of merging your new updates to the upstream/master branch).
git clone https://github.com/my-forking-username/my-forked-repo.git
cd my-forked-repo
git checkout -b my-new-files-branch-name
# Delete all files and directories except for the .git directory
git add .
git commit -m'Remove old files'
# Copy your new files to this my-forked-repo directory
git add .
git commit -m'Add new files'
git push origin my-new-files-branch-name
Create a PR on GitHub and request to merge your my-new-files-branch-name branch in your my-forked-repo into master.
Note: The "...entirely different commit histories..." error message can also occur in non-forked repos for the same reasons and can be fixed with the same solution above.
A more simple approach where you can't mingle with the master.
Consider i have master and JIRA-1234 branch and when i am trying to merge JIRA-1234 to master i am getting the above issue so please follow below steps:-
From JIRA-1234 cut a branch JIRA-1234-rebase (Its a temp branch and can have any name. I have taken JIRA-1234-rebase to be meaningful.)
git checkout JIRA-1234
git checkout -b JIRA-1234-rebase
The above command will create a new branch JIRA-1234-rebase and will checkout it.
Now we will rebase our master.
git rebase master (This is executed in the same branch JIRA-1234-rebase)
You will see a window showing the commit history from first commit till the last commit on JIRA-1234-rebase. So if we have 98 commits then it will rebase them 1 by 1 and you will see something like 1/98.
Here we just need to pick the commit we want so if you want this commit then don't do anything and just HIT Esc then :q! and HIT ENTER.
There would be some changes in case of conflict and you need to resolve this conflict and then add the files by
git add <FILE_NAME>.
Now do git rebase continue it will take you to rebase 2/98 and similarly you have to go through all the 98 commits and resolve all of them and remeber we need to add the files in each commit.
Finally you can now push these commits and then raise Pull Request by
git push or git push origin JIRA-1234-rebase
This happened for me because I created a repo from GH, but then I also added a README. In doing so I created a commit on my remote.
Then I went and created a new repo locally, made some changes and committed. Then I pushed it to the 👆repo and tried to make a Pull Request.
But my remote's initial commit was different from my local's commit, hence this error message. GitHub itself even warns you against this:
Create a new repository on GitHub.com. To avoid errors, do not initialize the new repository with README, license, or gitignore files. You can add these files after your project has been pushed to GitHub.
GitHub Docs
Similarly if you're creating a new repo, GitHub will quietly suggest that you skip initializing the repo. Rather just define the repo.
tldr the very first commit has to be identical, you can't merge 2 commits that don't have an identical initial commit.
If you know from which commit issue started, you can reset your branch to that commit and then merge them.
this is 100% works in any situation :
1)create new folder in your machine
2)clone the remote repository to the new folder
3)delete all files and folders except for the .git folder
4)add your project files that you are working on to this new folder you created
5)open terminal
6)cd new_folder_path (path to the new folder you created)
warning : don't type > git init
7) > git add .
8) > git commit -m "write anything"
9) > git push URL(url of the remote repository)local_branch_name:remote_branch_name
This happened with me yesterday cause I downloaded the code from original repo and try to pushed it on my forked repo, spend so much time on searching for solving "Unable to push error" and pushed it forcefully.
Solution:
Simply Refork the repo by deleting previous one and clone the repo from forked repo to the new folder.
Replace the file with old one in new folder and push it to repo and do a new pull request.
I solved that problem. In my case when i did “git clone” in one directory of my choice without do “git init” inside of that repository. Then I moved in to the cloned repository, where already have a “.git” (is a git repository i.e. do not need a “git init”) and finally I started do my changes or anything.
It probably doesn’t solve the problem but shows you how to avoid it.
The command git clone should be a “cd” command imbued if no submodule exists.
I got this error when initializing a GitHub repository with a README file, and then trying to push my existing local git repository to it. This resulted in a main branch with the README file, which was the Default branch, and a master branch with my code, and they couldn't be merged.
But since I didn't actually have anything important in my main branch (if you need to keep the data from both your branches, check out PaianuVlad23's Answer instead), I managed to solve the problem by changing the Default branch to the master branch, and then delete the main branch, like this:
When in GitHub, click your user icon in the top right of the window.
Choose "Your repositories", and then click your repository name.
Under the repository name, choose the tab "Settings".
From the pane on the left, choose "Branches".
Under the headline "Default", change the default branch from the one you want to delete (in my case main) to the one you want to keep (in my case master).
Now, click the tab "Code" under the repo name.
Under the tab line containing "Code" etc, you'll see a place where it says "2 branches". Click it.
Find the branch you want to delete, and click the trash bin icon on the right on that line.
Now, your repository has only one branch, which is the one you want to push your local changes to! 🙂 Just as if you hadn't initiated your repository before pushing to it, as #mfaani's answer in this thread suggests you do it.
I found that none of the answers provided actually worked for me; what actually worked for me is to do:
git push --set-upstream origin *BRANCHNAME*
After creating a new branch, then it gets tracked properly. (I have Git 2.7.4)
I don't think we have same case here, but still someone else may find it helpful.
When similar error occurred to me, it was going to be the first merge and first commit. There was nothing in on-line repository.
Therefore, there was no code on git-hub, to compare with.
I simply deleted the empty repository and created new one with same name.
And then there was no error.
I got this error message, because I was migrating an application from SVN to GitHub and it's not enough to invoke a git init in the location of the source code checked out from SVN, but you need to invoke a git svn clone in order to have all the commit history.
This way the two source codes on GitHub will have a mutual history and I was able to open pull requests.
I had an issue where I was pushing to my remote repo from a local repo that didn't match up with history of remote. This is what worked for me.
I cloned my repo locally so I knew I was working with fresh copy of repo:
git clone Your_REPO_URL_HERE.git
Switch to the branch you are trying to get into the remote:
git checkout Your_BRANCH_NAME_HERE
Add the remote of the original:
git remote add upstream Your_REMOTE_REPO_URL_HERE.git
Do a git fetch and git pull:
git fetch --all
git pull upstream Your_BRANCH_NAME_HERE
If you have merge conflicts, resolve them with
git mergetool kdiff3
or other merge tool of your choice.
Once conflicts are resolved and saved. Commit and push changes.
Now go to the gitub.com repo of the original and attempt to create a pull request. You should have option to create pull request and not see the "Nothing to compare, branches are entirely different commit histories"
Note: You may need to choose compare across forks for your pull request.
Top guy is probably right that you downloaded instead of cloning the repo at start.
Here is a easy solution without getting too technical.
In a new editor window, clone your repo in another directory.
Make a new branch.
Then copy from your your edited editor window into your new repo by copy paste.
Make sure that all your edits are copied over by looking at your older github branch.
I had mine solved by overriding the branch:
My case: I wanted to override whatever code is in the develop with version_2.
delete the local copy of conflicting branch:
git checkout version_2
git branch -D develop
checkout a fresh branch from the version_2 and force push to git:
git checkout -b `develop`
git push origin `develop`
I didn't need to rebase. But in my case, I didn't need to take code from my old code.
first: pull from remote repo
merge or rebase
finally: push to remote repo
finish
When you are pull/merging feature to main and are in the main branch in the terminal, I successfully used 'git pull origin feature --allow-unrelated-histories'.
Before using this command, I had the same message about completely different commit histories, and I think it's because I accidentally pushed to main after committing to the feature branch. Then I tried some of the solutions offered here like rebase, which allowed me to merge my code, but I still had the compare and pull notifications through git, and it was a one time fix. By one time fix I mean I still got the different commit history message the next time I tried to merge a feature branch's code to main.
Another source from a google search offered the --allow-unrelated-histories fix, and it permanently works exactly how I wanted it to. The branches were merged and now I can merge without error messages and the compare and pull notifications work through git.
I'm sure there are consequences for people who didn't have the same problem as me, but I didn't lose any code and now my repo is clean. Also, I'm also an amateur coder and the question is older so maybe this command wasn't available when the question was asked or I'm not understanding the issue correctly.
I wanted to copy commit history of "master" branch & overwrite the commit history of "main" branch .
The steps are:-
git checkout master
git branch main master -f
git checkout main
git push
To delete master branch:-
a. Locally:-
git checkout main
git branch -d master
b. Globally:-
git push origin --delete master
I am using eclipse, git and github. A friend of mine forked my github repo and changed some code. There was a pull request, I accepted the changes to my github repo. Now I tried to commit my own files and I am not able to 'commit and push'. As you can see in my screenshot there are no visible differences in the file. What procedure will solve this conflict?
Assuming your friend's work got merged into master and you are on branch your-feature-branch then:
git checkout master
git pull
git checkout your-feature-branch
git rebase master
will attempt to rebase your work on top of master, which already includes your friend's changes. The rebase procedure will process your commits one at a time and inform you of any conflicts which you can then resolve. Each file with conflicts will have conflict markers that tell you the changes introduced by each branch. After resolving the conflicts stage your changes then do
git rebase --continue
Repeat until rebase finishes. If at any point you become confused you can always abandon the rebase with
git rebase --abort
Once your work is rebased push your-feature-branch to origin, open a pull request, then merge.
I got the same issue sometimes, without a visible reason. Using a git rm on the directory containing the bad file, followed by a git reset --hard solved the issue for me.
git rm --cached -r DIRECTORY
git reset --hard
I created a local git repository, and I push changes from it to a gitosis remote that I created with
git init my_git
git remote add origin git#server:my_git
... various adds and commits
git push origin master:refs/heads/master
Now, I edit and commit changes locally in eclipse, and when I commit, I see (using qgit) that it moves my master branch to that version.
However, it also shows me that origin/master is at the previous version.
git status on command line shows me everything is up to date:
$ git status
# On branch master
nothing to commit (working directory clean)
I can see the differences in versions with
git diff origin/master
If i do git push on my command line, then qgit shows me the origin/master branch is now at same place as my master.
I can't work out how to configure the "remote / push" or "remote / configure push to upstream" dialog in eclipse to do the same thing as a command line git push to move the origin/master to the same level as the master.
I always have to do the command line push to make the origin/master come up to the same place as master.
Q1. Can anyone tell me how to do this in eclipse?
Q2. What is the command line version of git push doing that the eclipse version doesn't do?
Q3. Are my assumptions that master is my local HEAD pointer and origin/master is the remote server's view of the current HEAD correct?
Going by the relevant part of egit's documentation you can either:
click the "Add all branches spec" button, to push all of your local branches to ones with the same name in the remote repository, or
(the much safer option) just select master under both "Source ref" and "Destination ref" to only push your master branch
The remote-tracking branch origin/master is usually updated by git fetch (which is part of what git pull does), but with command line git, the remote-tracking branch is also updated on a successful push to the branch in the remote repository that's being tracked. It's possible that Egit, being based one of the pure Java implementations of git, JGit, rather than the command-line tools, doesn't update origin/master on a successful push in the same way. If that's the case, you can just do a fetch to update origin/master.
Update: It seems that this is a known bug in EGit (not the underlying JGit) - the bug report is here:
Push does not update remote tracking branch
An update to this, I was using eclipse Helios, and I've upgraded to Indigo, with the latest version of egit, and the fix appears to be active, as I'm no longer having to pull after a push.
I would like to do some experimental work in a hg project. So I would like to create branch, commit to it. And if the experiment works, I can merge it back to main branch.
In git, I can do
$ git branch experimental
$ git checkout experimental
(edit file)
$ git commit -a
$ git checkout master
I've read A Guide to Branching in Mercurial. It said hg branch feature. But what is next?
I don't follow.
$ hg branch experimental
(edit file)
$ hg commit
$ hg update default
First note that git branch is different from hg branch. Branches created with hg branch are permanent and live in a global name space, whereas branches made with git branch are transient. The nearest equivalent of git branch is hg bookmark: bookmarks can be renamed and deleted and behave more like Git-branches.
I've recently written a guide for Mercurial bookmarks. Compare this with the named branch guide. Both guides contain worked examples of how to use (named) branches in Mercurial for keeping track of the development. It shows how to merge branches and how to close them or delete the bookmark when you are done.
If it's not a big feature (i.e. the branch doesn't have to have a name), it's quite simple.
Let's say your repository is at changeset X. You work on the feature as much as you like, commit, commit, commit and if you're happy with the result, continue as if you knew it would work all along. ;) If you aren't happy, do a hg update X and continue development from there. All the work you did on your experiment will become an anonymous branch.
Strangely enough, it appears that Git doesn't provide such a way to work with anonymous branches which is what might be confusing you.