Schema creators ignore Miller’s [law concerning the cognitive limits of human beings] at their peril. User and software developers alike can and will subvert the technically elegant designs in order to work within comfortable cognitive limits.
–“Soft Issues Surrounding Industry Standard Schemas”
(4:13:17) bushwald: Would you us DTDs or XSDs?
(4:14:05) c0wb0yd: DTD
(4:14:10) c0wb0yd: but it depends on what for
(4:14:17) bushwald: Just in general.
(4:14:25) c0wb0yd: yeah in general, DTD is still better
(4:14:30) bushwald: What cases would you use either one?
(4:15:08) c0wb0yd: well, if I needed to do fine tuned data verification in the schema, or manage complex relationships between the entities, I’d use XSD
(4:15:34) c0wb0yd: but then again, I might just do that in code, because I haven’t seen how gracefully collecting XSD errors is.
(4:15:46) c0wb0yd: does it just barf and say “hey, invalid DOM asshole!”
(4:18:02) bushwald: Someone just asked me the same question, and I gave pretty much the same response.
(4:18:25) bushwald: I just applied Cotés Indicisiveness Razor: Err on the Side of Simplicity.
(4:18:29) c0wb0yd: use DTD if you want it to be readable.
(4:18:34) bushwald: Exactly.
(4:19:03) bushwald: I was like, “you gotta be a little worried when there’s a whole series of documents for the ‘spec,’ and the first one is 100 pages and’s called ‘Introduction.’”
(4:19:47) c0wb0yd: heh.
(4:30:56) c0wb0yd: yeah, shit that’s too complicated goes down the shitter.
(4:31:24) c0wb0yd: BALL HAIR!
