either define concepts or set of objectives

put some effort into some kind of summarizing statement that is a position
	intended more as notes for someone to write a well crafted document

to have impact outside of this group, its important to keep this in mind that everyone involved 
is not all from top universities


what all undergraduates should know
	shriram's suggestion
	2 roles: expose students to things that they don't get elsewhere in the curriculum
		other is the deep concepts we've talked about herein

	problem is that if there is no PL person on the faculty that can teach the elective


From group A
	features vs concepts

	reflection: understanding the underlying architecture of an implementation - Bryan Smith (interpreter
	that can understand its insides)
	
	if we focus on what all undergraduates to have
		outcomes/objectives/competencies
			non-PL core course: 
			programs of data

	if we focus on PL concepts (abstract concepts)
		we need some layering
		
		modularity/interfaces
		information hiding
		data abstraction
		different runtime models: call stacks/frames, address arithmetic, memory model, organization, malloc
		class hierarchies, objects and classes
			programming with state
		higher order functions
		persistent and ephemeral data structrues
			mutation/linked lists/ purely functional data structures
		understanding why your code is correct
			invariance, abstract types, invariant governs a data struture and operations assume/guarantee it


		concurrency/parallelism
			sync primitives, CML
		data vs task parallelism

		control, state, naming
			exceptations, objects, binding/scope

		cost model

		very trendy to say it: mapreduce 
			stateful vs statelessness is vital
			html -- stateless, functional programming
				the way webservices are built, functional
				load balancing
		tie-in with mathematics
		absense of state


		pruning
			have a way of writing down an abstraction
			concurrency and parallelism
				- deterministic vs nondeterministic
				- concurrency: composition (different outcomes)
				distinction between the two is what is important
					models of parallel computation
						data, task
					concurrency without parallelism
						ML argument John 
					
		statelessness
		programming with immutable data
			persistent is immutable
			josh box programming in ML using Java syntax


		naming -- from outside of Pl

		ways of programming
			lay on top of way of programming
				can program in a purely higher order fashion
		

		streams could go under functional programming
			but too specific

		dynamic method dispatch
			could go under control
			make it dynamic dispatch -- much more general, even interpretation 

undergraduate PL 
	teach it from a more formal perspective
	a lot more depth -- consistency across 

	monads -- type systems modal that segregate types
		dynamic/static typing, binding
----------------------------------------------------------------------