Vim & Vision

You searched for articles tagged with Git.

Git's Index Permalink

Git Added about a month ago

Plasmasturm has some excellent things to say about git's index, including:

So what is the index good for? The key to understanding it is how it interacts with git diff. Once you add something to the index (also referred to as staging it), it disappears off the diff. You can pass --cached to see what changes you have staged, but by default, it doesn’t show you the changes that you have asserted are ready for commit. When I first read about this, it sounded outright stupid to me. Why would anyone want that? Turns out: because it is hugely helpful. Consider: when a merge fails, the successfully merged diff hunks are staged, but conflicted hunks are not – which means that git diff will show only conflicts, and the successfully performed part of the merge doesn’t cloud the diff. Furthermore, the way to mark files with conflicts as merged is to stage them after manual resolution, which makes them too disappear from the diff. Maybe this is why Linus introduced the concept in the first place, being that the main part of his job is to perform merges all day long. But that’s far from the only circumstance in which the index has been useful to me.

Git GUI Permalink

Git Added about two months ago

The GUI you get from:

$ git gui

... is a good way to look at all your changes (with diffs), selectively stage them, commit and push. This is mostly just an easy shortcut but I find it easier to compose a commit message this way, while you can see the diffs for changed files.

Git Squashing Permalink

Git Added less than a year ago and last edited less than a year ago

If you decide to choose a development workflow where you are working on lots of branches (perhaps a branch per feature), then you will want to merge branches back to master regularly - like when you've finished your feature. If you have branches for smallish items then you might only want the whole feature to appear under one commit in your master branch - even though your feature branch .

This is when you should use git's commit squash.

$ git branch
* master
$ git checkout -b feature-01
$ # EDIT A FILE
$ git add .
$ git commit -m "Edit 01"
$ # EDIT ANOTHER FILE
$ git add .
$ git commit -m "Edit 02"
$ git log
 ... output here lists Edit 01, Edit 02 and all previous log messages ...
$ git checkout master
$ git merge --squash feature-01
$ git commit -m "Feature 01"
$ git log
... output here lists Feature 01 and all previous log messages ...

So Edit 01 and Edit 02 now appear in your log the one Feature 01 commit.

If you don't use squash then all the commits from your branch will also individually appear in the commit log for master.

A squash acts like a regular merge in svn.

Git Push Error Permalink

Git Added less than a year ago

After rearranging my remote repos I received this error locally after trying to push upstream (from local to the new remote repo):

! [remote rejected] master -> master (hook declined)
error: failed to push some refs to 'ssh://myserver/myproject.git'

... make sure the hooks in your hooks/ directory are all chmod -x (unless of course you have purposely enabled one.) This folder is full of default/example hooks that should be only enabled/modified if you require them. If they are executable they are enabled.

Using Git for Server Configs Permalink

Git, Shell, Apache Added less than a year ago and last edited less than a year ago

I dev on a VM that has at least two projects running with incompatible apache httpd.conf files. The differences include User/Group, Location etc, so I switch between the configs and then restart apache httpd when I want to work on a particular project.

Firstly, I understand that some or all of my different settings could be merged into one conf file and made to work on different virtual hosts etc, but in some situations you can't do this because: you only want one project running off your dev server at a time due to performance/load issues; or you have a specific file you need to use as a httpd conf file from your codebase etc.

So - I used git to help me switch between configs.

# cd /etc/apache2
# git add apache2.conf
# git commit
# git tag projA
# vim apache2.conf # or cp your apache2.conf for projB into place
# git add apache2.conf
# git commit
# git tag projB
# echo "/etc/init.d/apache2 restart" > .git/hooks/post-checkout
# chmod a+x .git/hooks/post-checkout

... now you are setup! So when you want to switch to projA you do this:

# cd /etc/apache2 && git checkout projA

... and to switch back to projB use:

# cd /etc/apache2 && git checkout projB

... and your apache will automatically restart after each checkout.

Yes you could do this just by copying your files into place but I think this is a neat use of git and one that will scale over many projects and many revisions of the conf files.

UPDATE

Instead of using tags, use branches! That way you can update the conf files, commit, and then continue using the projA and projB identifiers.

The Bull Shark's Guide... Permalink

Subversion, Best-Practice, Agile, Testing, Git Added less than a year ago

... To Modern Software Development is a concise list of technical points for a software house to implement. It has a basis in Agile methodologies but doesn't preach a certain one.

Modern is being proactive and efficient. It is being prepared, having backups, examining information, deploying with minimal mistakes and being able to recover and review when mistakes are made. Modern is being well-tested and writing maintainable code. Modern is being able to release regularly and easily.

Using Git to Rollback Bad Deployments Permalink

Git, ThisBlog, Apache Added more than a year ago and last edited less than a year ago

Git just keeps rocking my world.

OK, here is how I deploy my blog and here is how git saved me last night when the deployment went bad:

I git-pull into a webroot directory (well, the directory that is setup to serve my python) from my main git repo (which is also hosted on the same machine.)

When I updated the blog software last night I did a git-pull, restarted apache and saw the error (yes I should've caught it in dev, shoulda-coulda-woulda but-but-but.)

So, error noted. What do I do to rollback to the last working version?

$ git log

... and I copy the SHA hex-string of the last working commit. Then we rollback with:

$ git checkout <copied_hex_string>

... and restart apache. Ta-da! Now the blog is still up while I do the bug fix.

OK so what happens when I've commited the fix and I want to make the blog update?

$ git pull

... puts all the new changes in the git repo but doesn't fast-forward it to the HEAD of master automatically. Why? Because you aren't in master anymore - you've checked out a previous commit. So all you need to do now is:

$ git checkout master

... restart apache and you're set.

EDIT: Here is some info to rollback to the previous commit in a more general way:

$ git checkout HEAD~1
...  # you are now at the previous version
$ git checkout master
... # you are now back to the latest version

Git Remote Repos - After The Fact Permalink

Git Added more than a year ago

Git is easy to setup locally. You might be in a directory, working on a new project, and you just do a:

local$ git init
local$ git add .
local$ git commit

... and suddenly you're all setup. But what if you've done this locally, and you want to use a remote repo as a backup of this local repo?

Setup the Remote Repo

remote$ mkdir yourproject.git
remote$ cd yourproject.git
remote$ git init --bare

Tell your Local Repo About the Remote One

local$ git remote add origin ssh://yourserver.tld/~/yourproject.git

Push Upstream

local$ git push origin master

... and you're done! From then on you should be able to simply do a:

local$ git push

... to continue to "backup" your repo.

Git - Trailing Whitespace Permalink

Git Added more than a year ago and last edited more than a year ago

If Git is complaining about trailing whitespace in the files you are trying to commit, and you don't think it should be (that's your decision) then you can stop it whinging by changing your config:

$ git config core.whitespace -trailing-space

Now add and commit again.

Want this to work on all your git repos, not just the current one?

$ git config --global core.whitespace -trailing-space

Git Problems - Dangling Objects Permalink

Git Added more than a year ago

If, when trying to commit, you ever get a message like:

unable to create temporary sha1 filename sha_hash File exists

... from git, you need to do a few things from within the root level of your git repo.

Run fsck

$ git fsck
# should show you some dangling objects

Run garbage-collect

$ git gc
# should show you some progress as it prunes old dangling objects

If committing still doesn't work after this, you need to try:

$ git prune

... because git-gc by default only prunes dangling objects that are older than two weeks, whereas prune gets them all.

My error still hadn't gone away after this. Turns out I had accidentally done some git adding as root instead of the user I currently was. So I needed to do this:

$ sudo git gc
$ sudo git prune

Then I:

$ sudo chown -R myusername: .git/

... to make sure all the objects belonged to me. Committing then worked fine!

Older Posts ... (Nothing Newer)

Colophon

Django Python 960.gs Git Vim NetBSD Nginx

The Author

The author is a software engineer living in Australia. He sux at guitar, loves camping, doesn't like cake, does like coffee and is a lazy home brewer.

Links

ChoppingBoard, Project365, RageQuit