Personal branch strategy

• My personal fork does not duplicate anything in HDFGroup/hdf5 - features and bugfixes only!
• HEAD points to this branch, named NOPE, which is essentially empty
• I clone via the clone_here script included in this file:
  • Clones my personal fork
  • Adds HDFGroup/hdf5 as a remote named canonical
  • Sets the canonical push URL to nonsense to avoid creating branches there
  • Fetches canonical and checks out the develop branch
• When starting a feature branch, I branch off of the development/feature branches in HDFGroup/hdf5
• I try to prefix features and fixes with something useful, like "10" for merges to hdf5_1_10
• Anything unprefixed was probably spun off of develop
• Add a BRANCH.md file to anything likely to be long-lived
• Delete branches after the changes have been pushed as far as they are needed

Why do I do this?

• Syncing personal copies of HDFGroup/hdf5 is a waste of time
  • Also, doing this poorly can generate ridiculous log entries showing every merge commit made in your personal copy
• No wading through canonical HDF5 stuff to find my personal branches
• Easily spin branches off of HDFGroup/hdf5 and move code around
• No name clash - git checkout develop resolves to canonical/develop
• Works like a personal extension to HDFGroup/hdf5