Re: Making a case for cases

[Ralph Lange <Ralph.Lange@bessy.de>]

 On 05.08.2010 18:00, Steiner, Mathias wrote:
> I believe the solution is to have project leaders design the names into the system from the start.  This might even make them think about controls ahead of time ;->

In a perfect world .... Well, good luck with that. It would be really 
great if it worked.

> There needs to be an authority to make sure the proposed  names pass muster.  ":HVRd" isn't really worse or better than ":RdHV," but we don't want both.

I would suggest that first thing after your convention is agreed upon, 
you create a web-based parser, so that people can play around, validate 
their name ideas, and get used to the names.

The list of signal names will be ever growing. Put the authoritative 
list on the web and make people use existing names if they match the 
case, having the naming tsar add a new signal name, if it is something 
not covered yet.

Also, splitting up the signal name in two parts keeps your lists shorter:
- the signal (i.e. the unit, such as temperature, pressure, field, 
power, ...)
- the signal domain (i.e. the type, such as readback, setpoint, status, 
command, switch, selector)

Combining these two parts using CamelCase might yield quite nice-to-read 
yet short signal names, e.g.
PwrSt (power status), PwrSw (power switch), RstCmd (reset command), 
VoltSet, ... you get the idea.
Caveat: you might have to integrate things like a "voltage threshold 
setpoint" or a "temperature range selector" or reading back a setpoint, 
which need a bit more thought.

> A small committee might be indicated; I envision three or four people from different departments, preferably the kind who hate meetings...

A name oligarchy ... might be more fun than just one tsar.

Good luck!
Ralph