On Nov 16, 2010, at 10:57 PM, Ethan Merritt wrote: > Bleah. Virtually none of those are human-readable, no matter what the > wikipedia page may choose to put as a heading title. > > What kind of data are you dealing with? PDF would indeed be an odd format for > diffraction images, but it would be miles better than most of the formats on > the list you point to.
The operative word is "dataset", which is a subset of all things "data". A dataset should be in a format that 1. can be validated 2. is structured 3. is machine readable A pdf file guarantees none of the above. It is a presentation format and is not optimized for validating, structuring, or ensuring the machine readability of the data that it might contain. I'm not advocating for any particular serialization format. So this isn't about JSON v. XML religion wars. This is JSON or XML versus a file format that is basically designed to ferry presentation information between printers or computer screens. James
