Re: How are the same file modified on different brances stored physically?

amishera <amishera2007@gmail.com> writes:

First, read existing bundled and on-line documentation: git(7), "Git User's Manual", "Git Magic", "Git for Computer Scientists", "Git in Nutshell". One of them should explain git concepts to you.

Let's assume for simplicity that project consist of two files, 'foo' and 'bar' in top directory. You have created side branch, using "git checkout -b test-branch" for example. You have modified file 'foo', and committed those changes (either "git add foo; git commit" or "git commit -a". This created three new objects in repository (in object database): a "blob" object containing new contents of file 'foo', new "tree" object with 'link' to old 'bar' and new 'foo', and "commit" object, with given commit message, with 'parent' to previous commit, and with 'tree' to newly created "tree" object. It also advances refs/heads/test-branch to new commit.

Now you switch to previous branch, with "git checkout master". You modify file 'foo' and commit changes. Again, this creates three new objects: "blob", "tree" and "commit". This means that there are three blob objects with contents of 'foo' and one blob object with contents of 'bar'.

Of course after repacking ("git gc") contents for file 'foo' might be stored as base objects (usually largest version) and binary deltas from base. But those are implementation (engine) details.