"Buhrmaster, Gary" <[email protected]> writes: > I am going to suggest an alternative. XML format. It is ugly, but > provides arbitrary complex levels, and can represent even binary data > if needed. I tend to have a hate relationship with XML, but it does > give one flexibility.
XML is an okay wire representation because there are a lot of parsing and validation libraries available, but it's spectacularly awful at being human-editable. If we were designing a data interchange format, I'd be inclined to agree with using XML, but for human-editable configuration files, I think we need to go a different direction. > Another format that has been used is YAML. It is certainly more human > readable friendly, and the library is under a compatible license as I > recall. The impression that I'm getting from the general trends in this area is that JSON is displacing YAML as the XML alternative of choice. I wouldn't mind using JSON, although it has the drawback of not having the section concept that's part of the Kerberos profile and Windows INI formats. You can, of course, create an equivalent of sections with nested objects and that would work fine, but I think the clear syntactic separation of sections from settings in the profile and INI formats is useful for humans. -- Russ Allbery ([email protected]) <http://www.eyrie.org/~eagle/> _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
