niczsoft

Working as opensource project maintainer is a big task. I have been doing it for some time for the RVM project thanks to Engine Yard.

Problem

A big part of the maintenance is accepting others code for the project; it is very important to handle it well so the code author wants to get back to us and help again. As much as Ruby community is opensource oriented there are few things that can be improved. It happened a few times to me and other developers that the code was merged into projects without preserving authorship. This is a basic error made by maintainers, so I want to share some thoughts on how I try to avoid that problem when maintaining RVM. This problem mostly happens when the we want to change the commits, but is not limited to it and happens also without changes.

Easy way

When you do not care about number of commits you can merge the commit from Github or using hub and then add new commit on top updating it to proper state mentioning the ticket number in commit:

git commit -m "Formatting ..., update #2195"

For example see https://github.com/wayneeseguin/rvm/pull/2195

Limit number of commits

The most basic method when you want to limit the number of commits is to apply changes from the committer and then commit it using the --author flag. I used this for https://github.com/wayneeseguin/rvm/pull/2238 – getting the proper result with involving author would require too much work as for small commit, so I did the required changes to the code (just changing order of paragraphs) and committed it using:

git commit -m "..., closes #2238" --author "Name <author@email.com>"

It is important that the author matches exactly the original author name and email (from their commits) so it does not confuse others or tools used (like git annotate). Also make sure to mention the pull request (close #2238) so it is automatically closed and your commit is linked with it.

Work with the author

The slowest way is to actually talk to the author and explain to him/her what changes are required to make his/her commit accepted. For example, this happened in https://github.com/wayneeseguin/rvm/pull/2187 where most of the discussions happened on IRC or Skype sometimes in comments. This way the person contributing gets best experience and getting back to coding on your project is much easier for him/her.

MINASWAN

Matz Is Nice And So We Are Nice

Most important for maintaining your project is treating other developers with respect, they are human beings, being impolite just builds a wall and ensures they will not get back to you or your project.

Other ways?

Obviously there have to be other ways to handle this, please comment below if you use another flow.

Over a year ago I have posted instruction how to setup simple git server, today I had to do it again but already having code so here is new instruction.

Create remote repository:

ssh mpapis@niczsoft.com -C "git init --bare repos/library3.git"

Create local repository:

git init
git add .
git commit -m "initial commit"

Tell local repository to synchronize with remote repository:

git remote add origin mpapis@niczsoft.com:repos/library3.git
git config branch.master.remote origin
git config branch.master.merge refs/heads/master
git push

You might be interested in the original post: fastest way to get git server

After long playing around with my prompt I finally made it stable and thought it’s time to share :)

So edit your ~/.bashrc file and add following lines on the end:

shopt -s promptvars dotglob histappend no_empty_cmd_completion cdspell xpg_echo

function parse_git_dirty {
  echo -n $(git status 2>/dev/null | awk -v out=$1 -v std="dirty" '{ if ($0=="# Changes to be committed:") std = "uncommited"; last=$0 } END{ if(last!="" && last!="nothing to commit (working directory clean)") { if(out!="") print out; else print std } }')
}
function parse_git_branch {
  echo -n $(git branch --no-color 2>/dev/null | awk -v out=$1 '/^*/ { if(out=="") print $2; else print out}')
}
function parse_git_remote {
  echo -n $(git status 2>/dev/null | awk -v out=$1 '/# Your branch is / { if(out=="") print $5; else print out }')
}
export PS1='$(ppwd \l)\u@\h:\[33[33m\]\w\[33[0m\]$(parse_git_branch ":")\[33[36m\]$(parse_git_branch)\[33[0m\]$(parse_git_remote "(")\[33[35m\]$(parse_git_remote)\[33[0m\]$(parse_git_remote ")")\[33[0m\]$(parse_git_dirty  "[")\[33[31m\]$(parse_git_dirty )\[33[0m\]$(parse_git_dirty  "]")>'

I know it looks a bit complicated, unfortunately it is … this is wired bash rule that escape sequences are evaluated before evaluation of functions/variables evaluation.

Some examples of prompt using this script:

mpapis@papis:~/old_laptop/nicz-projects/content2:master> touch a
mpapis@papis:~/old_laptop/nicz-projects/content2:master[dirty]> git add .
mpapis@papis:~/old_laptop/nicz-projects/content2:master[uncommited]> git commit -m "added a file"
mpapis@papis:~/old_laptop/nicz-projects/content2:master(ahead)>git push origin master
mpapis@papis:~/old_laptop/nicz-projects/content2:master>

To make it more useful the prompt is also colored to distinguish between states of git repo.

For lazy users the script could be also replaced by very easy version, which prints git status before each prompt line (only where git is applicable):

export PS1='$(git status 2>/dev/null)\[33[0m\]\n$(ppwd \l)\u@\h:\[33[33m\]\w\[33[0m\]>'

Note: download the code from here http://niczsoft.com/files/2010/05/my-git-prompt.txt

I was searching for a fast and easy way to setup git server, and I found one, but I was not fully happy with it, so lets have a look on another way:

ssh mpapis@niczsoft.com -C "mkdir -p repos/library2.git; cd repos/library2.git; git --bare init"
git clone mpapis@niczsoft.com:repos/library2.git
cd library2/
touch test.txt
git add .
git commit -a -m "1st"
git push origin master

This is good way to keep Your private repo, just as central repo or backup, no packages needed, just git on both sides

The same can be done on local filesystem like shared storage(nfs or smb):

cd /remote
mkdir test.git
cd test.git/
git --bare init
cd
git clone /remote/test.git/
cd test/
touch test.txt
git add .
git commit -a -m "1st"
git push origin master


From now on it will be working by simply calling “git pull” and “git push”.