backups + restores

• Git does 2 things:
  • Record subset of the history of your project
  • Keeps track the Future of your project

What can you do in Git?

• Creating a commit, that "moves part of the future into the history" (according almighty Egg)
  • This can include fancier changes, including merges of lines of development in the past
  • Typically the merges need some human assistance

Getting start:

$ git init # init a git reposotory, happens whne you start a new project
$ git clone # clone an existing repo
# .git is the repository that stores all the version info

After the clone, you can:

• look at the source code
• change the course code

$ git diff # tell you the differences of the changes that you made
# compares the most recent version to the changes that you make

• Look at status of dev tree

$ git status # tells you the current status: which files are modified, uncommitted etc.

.gitignore has the information for what git is ignoring, or not keeping track of

Q: Which of your softwares' files should not be put under Git control?

• If you can regenerate it automatically (mostly)
• It's machine-dependent (always)

Sh-style Globbing patterns

/abc*def matches only files at top leaves

!abc.o *.o will ignore every .o file but abc.o

• Check the history version

  $ git log # there's a lot, lot, lot of options
  # for instance, git log --pretty=fuller
  

  This will outputs:

  1. metainfo about each commit
  2. the commit message(written by hand, by the committer-the person who ran 'git commit', the author-person who wrote the patch)

• git ls-file list all the files that git is in control of

  grep Error $(git ls-file)
  git grep Error # identical to above
  

• To identify a commit, we can find a special value(40 chars) along with the commit that's the check sum calculated by the SHA-1 checksum. 1 in 2^160 chance that 2 SHA will collide. The SHA itself is extremely difficult to reverse-engineer

• So how should we use it? git log A..b will log everything e such that A < e < B, where A and B are ids. We don't have to use the full id and can just use first few characters too.

  Also, there are the special alias HEAD that points to the most recent history

  We can also do log by tag name like git log v3.8..HEAD

  C^ just before C C~3 3 commits before C

  git diff C^! = git diff C^..C

• git show C showing everything about this particular commit C

• git blame C --FILE look at all the changes of FILE at commit C and the commit messages