| Discussion |
Keywords: Saver, Question, Suggestion
Remote Name: 18.104.22.168
I think we're pretty close to agreement on this, but I'm confused about your proposed use of Categories. If you use the * functionid prefix to designate templates, you don't need to use Categories to tell them apart. The rest of the functionid could convey additional information indicating which template a given clone came from. I'm a little unclear about your concept of a "cloned" configuration, however. If all you have is templates and clones, it would suffice just to know which ones are templates. Is there another category of configs besides templates and clones? Is a "cloned" config one that is identical to a template, unchanged in any way? If so, what happens when the user makes a change; does it suddenly turn into a non-clone?
Assuming that you do have, or eventually intend to have other useful ways of categorizing configurations that reside in a common metafile, I wouldn't waste the Category field on words like "Clone" or "Template". Instead, I would envision some sort of coding scheme that provides a useful hierarchical ordering. (You could also make use of a pattern matching or other Category filtering mechanism, as proposed in svma0022.)
Let me illustrate how you might make the best use of the Category field:
1. Start by defining some useful dimensions of categorization, e.g. job type, customer, geographic region, time period, etc.
2. decide on a compact encoding scheme. In many cases you already have code table. Add new encodings as needed. It may be useful to avoid using the same code value for different dimensions, if you want to simplify pattern matching.
3. decide on the concatenation sequence for assembling the hierarchical Category code. Since you can order by Category, ascending or descending, this should be chosen to be the most useful logical organization.
So my suggestion is that you forget about the fact that you changed all the Categories yesterday, and give further thought to a more useful Category scheme, presumably with input from Alan and Nancy as well. Meanwhile, just use the * functionid convention to distinguish templates from everything else.