- 1.1 Git clone tiki
If you master svn and want an explanation of Git in terms of differences with Subversion, the SVN Crash Course document is fairly short.
Also see Git and SVN combined workflow.
The main repository is git at git.tiki.org:tiki.git You need to have your public ssh-rsa key registered to that server if you are going to merge the accepted pull requests from the Gitlab mirror. Ask luciash <firstname.lastname@example.org> or Fabio <email@example.com> if you need that direct access.
An automated mirror. This is what you should clone if you want to do pull requests on https://gitlab.com/tikiwiki/tiki
No need to ask anyone, but you DO need to upload you ssh key to https://gitlab.com/profile/keys
Tiki development has been going on for nearly 20 years, and the code base is huge. For these two reasons the repository has a huge size and a regular clone will take a lot of time and disk space.
But it is possible to create a shallow clone, which will bring just the last source code version.
git clone --depth 1 --no-single-branch firstname.lastname@example.org:tikiwiki/tiki.git tiki
This is quite fast, and uses about 292M of disk space. Probably best for all developers.
git branch -a works like most people expect and it's easy to switch branch.
If the purpose of the clone is Tiki development, including master and other branches, it is not needed to clone the entire repository, buy you my want to use --depth 100 so you have some useful history. (this will inflate the size to about 455M however.
git clone --depth 1 --branch master email@example.com:tikiwiki/tiki.git tiki
The smallest option, but you can't use
git branch -a like you normally would. instead you need to use
To add another branch, you need to use the variant below.
This option is best to checkout on production servers. It uses about 187M of disk space
git clone --depth 1 --branch master firstname.lastname@example.org:tikiwiki/tiki.git tiki git remote set-branches --add origin '19.x' git fetch --depth 1 #REALLY important on the first fetch, otherwise that whole branch will be fetched with all history. That will use up 1.4G and take a really long time. #For normal operations you can just git fetch normally, it won't un-shallow the branch.
It now uses about 187M of disk space, which is barely more than a single branch. This option is best if you rely on expensive data connections and/or have very limited disk space.
git clone email@example.com:tikiwiki/tiki.git
This is quite slow, and uses about 3.1G of disk space. However subsequent operations are not any slower, and you have full local history. This is best if you have the time for the initial clone, and you need to work completely offline.
git clone --reference
Full git repository is 3.1G. That's because the whole history is stored locally. Since Tiki setup depends on a lot of libraries, changing branches in a single repository might be impractical, and having one repository for each directory would consume a lot of redundant disk space.
To solve that, start by having a local mirror repository that will be used as reference for all other repositories. This way you can have several repositories that share the local object database. It's like doing a pre-cache of the repository:
git clone --mirror firstname.lastname@example.org:tikiwiki/tiki.git tiki-reference
(This will take some time, have a coffee).
Supposing a developer wants to clone also the branch 19.x, but without cloning the entire repository, the following command should be used.
git clone --branch=19.x --reference=tiki-reference email@example.com:tikiwiki/tiki.git tiki19x
Repeat for other branches.
Now just commit and push normally, and everything should work fine.
In practice, git is often based on ssh, which may be confusing if you've never used it before. Among other things, it will ask to accept the key of the remote server the first time you use it.
clone --mirror firstname.lastname@example.org:tiki.git tiki.reference Cloning into bare repository 'tiki.reference'... The authenticity of host 'git.tiki.org (220.127.116.11)' can't be established. ECDSA key fingerprint is SHA256:bsX1VZ1KCXiUlequpZc6f+JGBG0xO4HpUSyYnVrXbMs. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added 'git.tiki.org,18.104.22.168' (ECDSA) to the list of known hosts. email@example.com's password:
You can read about changed files, untracked files and how far is your local repository from the remote repository by using the
git status command. More information can be found by typing
git help status.
git status --ignored
The command below will list all ignored files present on the working copy. This is useful when trying to commit a file that should be on repository, but it is ignored by a rule present in
git status --ignored
git status -u
Sometimes, depending the task, it is possible that working copy contains a lot of untracked files. But it is possible to check working copy status excluding those files.
git status -u no
There are other occasions when there is a need to check all untracked files, including files on sub-directories.
git status -u
To read the commit log, container authors, data and commit messages, you can use the command
git log --author --since
It is possible to filter logs by author and date, like the snippets below demonstrate.
git log --author=jonnybradley # --author can be used multiple times git log --author=jonnybradley --author=drsassafras
git log --since=2018-03-01 # --since is the start date, and --until is the end date git log --since=2018-03-01 --until=2018-04-31
git log --since=2018-03-01 --until=2018-04-31 --author=jonnybradley --author=drsassafras
For more options, try
git help log
Before committing, the changes you did should be marked as added. So, when you tell git to commit your changes, only files you marked as added will be committed.
echo 'My new super file' > src/foo.txt echo 'Other super file' > src/bar.txt git add src/foo.txt src/bar.txt
It is possible to add all changes from a folder at once. It also works to add changes from entire working copy, but it is not a good practice.
echo 'My new super file' > src/foo.txt echo 'Other super file' > src/bar.txt git add src git commit -m '[NEW] Add new super feature to Tiki'
git rm src/foo.txt git commit -m '[FIX] Removed unused file'
After all new or changed files are marked to be committed by using
git add, or files were removed by using
git rm, it is time to commit these changes to your local repository.
git status git commit -m '[NEW] Descriptive message about my changes'
There are occasions a developer commits changes from someone else. In this case, it is possible to give the commit the correct author identity.
git commit --firstname.lastname@example.org -m '[FIX] Options list for editable icon preference'
When there are changes just on versioned files, there is shortcut to commit these changes at once, by adding
-a parameter on git commit.
git commit -a -m '[NEW] Descriptive message about my changes'
Creating branches is easy and fast on Git. It is a good idea to have different branches for tasks, features or experiments and keep master branch clean, so you can fetch updates regularly without dealing with conflicts.
# create the branch git branch task-icon-picker # and then switch to branch git checkout task-icon-picker
There is a shortcut to create a branch and then switch to it.
git checkout -b task-icon-picker
If this local branch is needed for other team mates, it is possible to make it available on remote repository.
git push -u origin task-icon-picker
Let's suppose you are working on branch *bugfix-foo*, and you want update your branch with branch master changes.
git checkout bugfix-foo # switch to branch git merge master # bring master changes to current branch
To send your new commits from your local repository to the remote repository, just use the command
git push. The command below will send all changes on local branch to the remote branch on remote repository.
It is a good practice to stay always up to date with remote repository your team is using. This way you avoid a lot of conflicts. To get remote changes and bring to local repository and working copy, use the command
The command below will fetch updates from remote repository.
After fetching changes, the local repository and branch should be updated with it. Supposing the branch to be updated is the
master, and the current local branch is also
master, it is possible to apply changes with the command below.
git merge origin/master
There is a shortcut to fetch and apply changes on current branch, like below.