From: Mark Rodgers (mark.rodgers_at_[hidden])
Date: 2001-07-24 14:58:14
> I think I've caught my breath a bit and would be willing to help Beman
> refactor these guidelines according to what appears to be the emerging
> consensus... but I would like to be absolutely sure that it's going to be
> worth the investment. Is there really (or can there be) general agreement
> about which things should be guidelines and which should be requirements?
> there really a consensus that something derived from the proposed document
> is appropriate? How can we answer these questions?
Well I can only offer my view on this. We'll have to see if we can gain a
IMHO anything that affects the external interface could be a requirement.
Thus we really do care about such things as
- How files are named and what directories they go in.
- How classes are named.
- How public and protected members are named.
- What namespaces are used.
- How macros are named.
- Use (or not) of exception specs
What we don't care about much is implementation details because these
really are a QOI issue. Here we can provide guidelines which are
suggestions how people can improve the (perceived) quality, but they
should be no more than that. Thus guidelines can include such things
- Organisation of files (e.g. #include order, log location, etc)
- Naming of private members
- Expression spacing and bracketing
- Spacing of definitions and declarations
- Declarations and initialisation
- Comments and documentation
- Class Organisation
The rest of the stuff seems to be more advice on design, and could go in
yet another document.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk