presenting the fundamental abstractions

[John Yates]

On Monday 2007-09-03 Martin Pool wrote:

> #3 raises the question of whether we want users to think "there are
> three components which can be under a bzr dir" or "a bzrdir can be of
> several types."  The former is closer to how it really works in Bazaar
> so it may be better for the ui to match.

This is the clearest statement I have seen of this.  Sadly Martin stops
short of enumerating those three components.  At present I am unaware of
any introductory bzr documentation that presents such a picture.  Instead
the documentation presents a panoply of composites and a set of scenarios
that may guide one towards a selection.

My style of learning a complex system is reduction and synthesis.  I need
to identify fundamental components, the effects of primitive operations on
those components, and rules for structural and/or functional composition.
I doubt that I am unique in needing to master systems in this manner.

One thing that I love about git(7) is that it presents git at essentially
this level (http://www.kernel.org/pub/software/scm/git/docs/#Discussion).
Chapter 4 of the Hg's book (http://hgbook.red-bean.com/hgbook.pdf) has a
similar discussion.

Admitted that both of these examples are somewhat more concrete than might
be desirable in introductory material, especially for bzr where many of the
fundamental components have multiple implementations.  None the less, I do
find it much easier to abstract from the concrete than to be offered a set
of related composites and have to derive an all-subsuming generalized model.
(Had I enjoyed such challenges I would have become a quantum physicist:-)

/john