notes from Bayjax Meetup: http://www.meetup.com/BayJax/calendar/108524

Doug Crockford on the json saga

– json already existed in nature, but crockford discovered it

– it wwas being used at netscape in he form of array literals in communication 5 yrs before crockford discovered it

– crockford’s first transmission used js embedded in html in a frame for cross-browser compatability

– they set document.domain to get around the same origin policy

– backslashes are tolerated in json so we can put html in json

– original name was JSML, but that conflicted w/ a pre-existing java protcol

– json’s good for interserver and db applications

– some of his customers balked at usage because it wasn’t standard, so crockford put up a website to standardize it

– json is the intersection of modern programming languages

— simple values

— sequence of values

— collection of named values

— an intersection is easier to find than a union

– js is brilliant for state machines

– most ppl implement json parsers using eval, but this must be guarded by regex to valiudate the json, which slows it doen

– the latest version of ecma script implements a native JSON.parse, which is ver fast

– ajax was an important influence on json uptake

– improvements

— strip comments 

— comments broke the parser

— comment parsing greatly increased complexity

— alignment w/ yaml

— added e notation to numnbers

– no version number

— everything’s crap until 3.0, but we avoid this by not having version numbers

— perhaps someday it will be replaced

— at least one piece of the stack will remain constant forever

– minimalism

— can fit on the back of a business card

– influences

— lisp s-expressions

— perhaps the greatest influence

— rebol

— al built upon a represenation of data, which is then executable

— rebol is a brilliant language

— JS, Python, NewtonScript

— Brendon Ike is a brilliant guy, so it’s no accident that it has brilliant aspects

— all were developed in isolation at the same time

— NeXT

— OpenStep property lists

— XML

— how did it become so popular?

—- html reduced it to basics, made it more basic, and made it easier to get everything to balance

—- A-level CTOs threw it out, but the B- and C-level developers embraced it and outnumbered the A-levels

— john seely brown said “maybe only something this simple could work”

—- he thought that the future was in loosely coupled systems

— xmlsucks.org

— some guy named pault lists all the xml alternatives

– disruption

– the 1st rule of workmanship: use the right tool for the right job

– xml arose out of: one tool to rule them all

– where did the idea come from that every data format should be a document format?

— runoff was one of the first

— GML

— Scribe

— the first place where document formate were done well

— separated format from markup

— if the web had been based on scribe instead of sgml, it would be a better place today

— scribe was the first time where doc format was used for data

– license

— MIT + “the software shall be used for good, not evil”

– the logo

— related to the ambihelical hexnut 😛

— a square and a circle w/ a twist

— data interchange we can believe in

– questions

— what would make html better

— make it extensible

— to be able to define new tags using css

— is there case-sensitivity in unicode?

— maybe.  use lower-case in the meantime

— what would you like to see replace json

— jsonp is brilliant

— currently, we can’t easily represent simple bags

— remove the quotes from the keys

— a = []; a[0] = a; => infinite loop when run thru parser

— schema-less langs

— i don’t care

— brilliant work in schema-less dbs now

— why no commnets

— because ppl were using them to communicate w/ the parser