On Mon, 2003-12-29 at 12:52, Fred L. Drake, Jr. wrote:
Changing the keytype allows different interpretations for keys. The only thing required of keys at the lowest level of the parser is that keys do not contain spaces. The "basic-key" datatype is used to create a case-insensitive handling to keys. Setting keytype to "identifier" would require that all keys in sections of that type be Python identifiers, and would be case-sensitive. Another keytype might allow email addresses to be used as keys:
Ah, that does sound useful! What I think the documentation needs is a kind of overview-slash-motivation-slash-real-life-examples. I'd start with an example configuration file, explain why it was designed that way, then work backwards to the schema and any components being used. That's how I tend to work anyway -- I think about what it is I want my end-users to be able to configure, i.e. I design the user interface for the .conf file first, then I figure out what the schema and component schemas I need to support that. -Barry