It’s interesting how clear the requirements about a SCM system should have when Linus talks about it. One day I should experiment another SCM tool, back at my days in the university we used none of these for most of the time. Actually, I had to interact with several data files that would be very nicely done inside a Database but I accessed it directly from disk. Savages =)
Interesting to see how people at that audience is not familiar with something that seems standard to me right now.
The only security work that can be done is through network of trust, and this should be the only way that development is done as well.
Distribution means, no one is special.
Git does not track files, it tracks your contents.
You might have as many branches as you like
Mercurial has the same model as git and some differences to implementation details
Centralized systems cannot work as well as distributed ones
Your functions should have names short, sweet and to the point
You should not have a single repository for every part of your system. It’s better to have a super project that have pointers to other git repositories.
History must always be seens on a project basis