Git: Difference between revisions

From XPUB & Lens-Based wiki
No edit summary
Line 1: Line 1:
=install Git=
https://git-scm.com/book/en/v2/Getting-Started-Installing-Git
https://git-scm.com/downloads
==Mac==
Git is installed when you install [https://developer.apple.com/xcode/ Xcode]
to check if it is installed run in the command line:
git --version
If it responds with a version number, Git is installed
=Start Git: init=
To start using Git, you need:
* a dedicated folder for you project, where all files and sub-folders of the project live
* navigate with the command line to that folder: <code>cd path/to/folder</code>
* initialize Git: <code>git init</code>
The last step with create a hidden <code>.git/</code> folder where all the Git data will be stored. 
=3 Git moments=
http://codingdomain.com/git/partial-commits/git-staging-area.png
For local (in your computer) Git operations, there are 3 main moments:
* working: editing files 
* stage: adding (tracking) files to git
* commit: span-shooting the changes performed on files
==Stage==
Staging refers to the action of making files ready for commit.
* <code>git add</code> when run for the 1st time on file or folder, it ask Git to start tracking it
** on a file that is being tracked, but has since changed, it include those changes to be commited
** There might be files you might not want to track and therefore you don't add them
* <code>git status</code> overview or staged moment:
** displays ''tracked'' and ''untracked'' files
Trick: If you wanna ''stage'' from files which Git is tracking run: <code>git add -u</code>
==Commit==
A commit is version of your project. When you complete some part of the work, you commit!
* <code>git commit -m "message about what happened in that commit"</code> creates a version or ''commit'' of your project. A accompanying message describes what happen in that ''commit''.
* <code>git log</code> show the ''history of commits'', where each commit is shows an author, date, message and hash (a long string of letter and numbers that identifies the commit)
=Undoing things=
==unstage==
Unstage, means bring added changes out or the stage moment.
In other words, before commiting, the changes you made to a file, then <code>git add that.file</code>, can be undone by running:
<code>git reset HEAD file.name</code>
Warning: those changes will be lost!
==checkout a specific commit==
The state of files in a specific commit can be checked out using:
<code>git checkout commit-hash</code>
This command will detach the HEAD (like the reading head of a K7) from the latest commit. In other words, it will let you go back in time to a previous commit.
To keep it simple, you can't do much there, but look around, copy content from files and save them in new (untracked) files. 
When you are done, you can return the head to the latest commit by running
<code>git checkout master</code>
==reverts==
Reverts the head to a previous commit:
<code>git revert commit-hash</code>
But instead of erasing history, creates a new commit
-----
=resources=
[http://ndpsoftware.com/git-cheatsheet.html#loc=index; Git Cheatsheet]
[http://git-scm.com/ Git home]
[https://wubthecaptain.eu/articles/why-i-dont-support-github.html blog on Github Terms of Services]
http://eagain.net/articles/git-for-computer-scientists/
http://eagain.net/articles/git-for-computer-scientists/


== Trees, Branches, and Blobs ==
 
=Remotes / collaboration=
 
Git facilitates the coordination of contributions from different collaborators in a project.
 
Each contributor has a local repository, that is a ''''clone'''' of the ''remote repository''.
 
 
'''The remote repository acts as the central node from which all the users will receive - ''pull'' - and send - ''push'' contributions.'''
 
 
==creating a remote==
In pzwart1 the git server is located in <code>/home/git/</code>
 
With the repositories stored  <code>/home/git/git/</code>
 
To create a remote, 1 person(normally the admin) start a ''remote (bare) repository'' to push and pull code to and from:
<source lang="bash">
cd /home/git/git
mkdir null.git
cd null.git
git init --bare
</source>
 
==set remote & first push==
(Note: This step only needs to be performed by the admin, whom will usually set a remote in the local repository).
 
Inside local folder <code>null-local/</code> dir:
<source lang="bash">
git remote add origin git@pzwart1.wdka.hro.nl:/home/git/git/null.git #add origin remote
git remote -v #view the existing remotes. Only origin remote was added
  origin git@pzwart1.wdka.hro.nl:/home/git/git/null.git (fetch)
  origin git@pzwart1.wdka.hro.nl:/home/git/git/null.git (push)
</source>
 
==cloning the remote==
To clone the remote git needs an authentication from its users, usually using '''ssh keys'''.
 
===authentication===
ssh key pairs (a public and private key) is an authentication method for clients to access servers through <code>ssh</code>. E.g. For my computer to ssh to pzwart1.
 
How do we go about doing that:
* in our local computers we need to [https://git-scm.com/book/en/v2/Git-on-the-Server-Generating-Your-SSH-Public-Key generate an ssh-key pair]
* in our local computer a plain text file  <code>~/.ssh/id_rsa.pub</code> stores contains our ''public key''
* we give our public key to the git admin,
* the adim adds the ''public key'' to the git server's <code>/home/git/.ssh/authorized_keys</code>
* we are ready to clone, push and pull to the remote.
 
===clone===
<source lang="bash">
git clone git@pzwart1.wdka.hro.nl:/home/git/git/amazing_project.git
</source>
 
 
 
----
 
 
= git commands: Trees, Branches, and Blobs =


  git init
  git init

Revision as of 11:47, 23 September 2019



install Git

https://git-scm.com/book/en/v2/Getting-Started-Installing-Git

https://git-scm.com/downloads

Mac

Git is installed when you install Xcode

to check if it is installed run in the command line:

git --version

If it responds with a version number, Git is installed

Start Git: init

To start using Git, you need:

  • a dedicated folder for you project, where all files and sub-folders of the project live
  • navigate with the command line to that folder: cd path/to/folder
  • initialize Git: git init

The last step with create a hidden .git/ folder where all the Git data will be stored.

3 Git moments

git-staging-area.png

For local (in your computer) Git operations, there are 3 main moments:

  • working: editing files
  • stage: adding (tracking) files to git
  • commit: span-shooting the changes performed on files

Stage

Staging refers to the action of making files ready for commit.

  • git add when run for the 1st time on file or folder, it ask Git to start tracking it
    • on a file that is being tracked, but has since changed, it include those changes to be commited
    • There might be files you might not want to track and therefore you don't add them
  • git status overview or staged moment:
    • displays tracked and untracked files

Trick: If you wanna stage from files which Git is tracking run: git add -u

Commit

A commit is version of your project. When you complete some part of the work, you commit!

  • git commit -m "message about what happened in that commit" creates a version or commit of your project. A accompanying message describes what happen in that commit.
  • git log show the history of commits, where each commit is shows an author, date, message and hash (a long string of letter and numbers that identifies the commit)

Undoing things

unstage

Unstage, means bring added changes out or the stage moment. In other words, before commiting, the changes you made to a file, then git add that.file, can be undone by running:

git reset HEAD file.name

Warning: those changes will be lost!

checkout a specific commit

The state of files in a specific commit can be checked out using:

git checkout commit-hash

This command will detach the HEAD (like the reading head of a K7) from the latest commit. In other words, it will let you go back in time to a previous commit.

To keep it simple, you can't do much there, but look around, copy content from files and save them in new (untracked) files.

When you are done, you can return the head to the latest commit by running

git checkout master

reverts

Reverts the head to a previous commit: git revert commit-hash

But instead of erasing history, creates a new commit



resources

Git Cheatsheet

Git home

blog on Github Terms of Services

http://eagain.net/articles/git-for-computer-scientists/


Remotes / collaboration

Git facilitates the coordination of contributions from different collaborators in a project.

Each contributor has a local repository, that is a 'clone' of the remote repository.


The remote repository acts as the central node from which all the users will receive - pull - and send - push contributions.


creating a remote

In pzwart1 the git server is located in /home/git/

With the repositories stored /home/git/git/

To create a remote, 1 person(normally the admin) start a remote (bare) repository to push and pull code to and from:

cd /home/git/git
mkdir null.git
cd null.git 
git init --bare

set remote & first push

(Note: This step only needs to be performed by the admin, whom will usually set a remote in the local repository).

Inside local folder null-local/ dir:

git remote add origin git@pzwart1.wdka.hro.nl:/home/git/git/null.git #add origin remote
git remote -v #view the existing remotes. Only origin remote was added
   origin	git@pzwart1.wdka.hro.nl:/home/git/git/null.git (fetch)
   origin	git@pzwart1.wdka.hro.nl:/home/git/git/null.git (push)

cloning the remote

To clone the remote git needs an authentication from its users, usually using ssh keys.

authentication

ssh key pairs (a public and private key) is an authentication method for clients to access servers through ssh. E.g. For my computer to ssh to pzwart1.

How do we go about doing that:

  • in our local computers we need to generate an ssh-key pair
  • in our local computer a plain text file ~/.ssh/id_rsa.pub stores contains our public key
  • we give our public key to the git admin,
  • the adim adds the public key to the git server's /home/git/.ssh/authorized_keys
  • we are ready to clone, push and pull to the remote.

clone

git clone git@pzwart1.wdka.hro.nl:/home/git/git/amazing_project.git




git commands: Trees, Branches, and Blobs

git init
git add *
git commit -m 'hello git'


init, clone

git init
git clone git://...

status, add, diff

tracked/untracked "staged"

file status lifecycle, p. 20 gitbok

git status

Add a file called ".gitignore" to ignore "temporary" or other files you don't want to track.

*.[oa]
*~
*.pyc


git diff

Shows changes to unstaged files only

git diff --cached

Show what you've staged so far

commit

git commit -m 'added new quote'
git commit -a -m 'edited links'

Stages every file (that's being tracked) -- ie "auto add"

git commit --amend

Add forgotten changes to last commit.

rm

git rm readme.txt
git rm --cached readme.txt

Remove from the stage, but not the file! (untrack)

git mv

(technically git "sees" these movements already, same as mv, remove, add: but the command is there for convenience)

log

git log
git log -p -2
git log --stat
git log --pretty=oneline
git log --graph

remote

git remote
git remote -v
git add remote remote-name URL

Where remote-name is whatever you want it to be called, and the URL is a remote git URL.

git remote show origin
git fetch remote-name

pull

pull = fetch + merge

push

git push [remote-name] [branch-name]
git push origin master

tags

Show SMW ?!


<!?> Find/link workflow article for git + web

Auto-Completion

contrib/completion/git-completion.bash

Add to .bashrc:

source ~/.git-completion.bash

(or)

cp git-completion.bash /etc/bash_completion.d/

(linux)

cp git-completion.bash /opt/local/etc/bash_completion.d/

(mac)

branch

git checkout -b iss53

create + checkout a branch, short for:

git branch iss53
git checkout iss53
git checkout master
git checkout -b fix
git commit -a -m 'fixed something'
git checkout master
git merge fix
git branch -d fix

"Fast-forward" reflects that the changes were "upstream" of the current working state.


conflict

Generally, you want to commit all your changes before switching branches (checkout)


If we each write to master -- lots of conflicts? Contrast with creating branch for each person ?!

Recipes

Cloning from git.xpub.nl

  • Use the https address
  • When prompted, enter your username + password for logging into gitea.

Pusing a folder to git.xpub.nl

WARNING: These recipes are based on an older install of git.xpub.nl...

This recipe requires that you know some things such as:

  • USERNAME: your username to ssh to git.xpub.nl. You will also need to know your password or (better) have already copied an ssh key to the server
  • SOMEFOLDER: the folder on your laptop (local computer) that you want to push
  • NAMEOFREPO: the name for the repo on the server... typically the same as the name of the folder but it shouldn't be the same as another repo.

Step 1: SSH to the server and run the bash script "newrepos.sh"

   ssh USERNAME@git.xpub.nl
   cd /var/www/git.xpub.nl
   ./newrepos.sh NAMEOFREPO "Description of my great project"

This spits out a message like:

   ALL GOOD!
   push URL: git.xpub.nl:/var/www/git.xpub.nl/repos/NAMEOFREPO.git

You can logout now.

Step 2: Turn your folder into a git repository

You can skip this step, if your folder is already a git repo.

   cd path/to/SOMEFOLDER
   git init

You may want to make and add a ".gitignore" file to avoid adding files you *don't* want in the repo. And then add things with

   git add myproject.py
   git add subfolder
   ...

And commit them

   git commit -m "First commit"

Step 3: Add the repo as a "remote" to your repo

   cd path/to/SOMEFOLDER
   git remote add pzi USERNAME@git.xpub.nl:/var/www/git.xpub.nl/repos/NAMEOFREPO.git

Note that you should add your USERNAME for ssh'ing to git.xpub.nl

Now finally you can push and "set the upstream" to that remote as the default...

   git push -u pzi master