Some features I use: Cherry-picking commits across branches. Merging branches. Rebases for forward-porting your work onto a new upstream. Bisects for determining which commit introduced a bug. Push/pull for sharing. Format-patch for sending patches around. Submodules.

Two of these I haven't done: Cherry-picking (well I did once, it went so wrong I had to reset the repository), and Bisects. That sounds very useful.

If you want to get a deeper understanding of what git's doing, read gitcore-tutorial(7). It's pretty easy to understand git's internals, and once you know them, it's easier to understand what the user commands are actually doing. Then, if something goes wrong, it's much easier to recover.

I am a guy who usually puts too much info into commit messages instead of code comments. Git blame recently helped me to find out why I had added a particular line to a lengthy if condition.

Svn blame would have done the same... Where's the git-specific use?

As with many svn features, blame requires access to the server and is also quite slow. In git, this sort of thing is instantaneous and therefore viable to do all the time if you like.

Lol! Never knew svn blame existed :-). I suppose tracking a change over multiple merges may not be easy?

I think we use a lot of PR flow in our team. It's really powerful if you collaborate or plan to collaborate with others as it allows you to get together and look at the commits to ensure code quality (in theory).

Even the command line git itself is really nice because it gives you the current condition of your repo and suggestion on what you can do next. You have changes, if you want to commit use this command or if you want to add use this command.

there are many 'minor' features that are occasionally useful. For example, when you make a stupid mistake, reflog can make you gain a couple of minutes.