Notes from Thursday's What breakout group

Focus Group B: What
	Chair: Jim Larus (Microsoft Research)
	Olin Shivers (Northeastern Univ.)
	Robert Harper (CMU)
	Stephen Freund (Williams College)
	Franklyn Turbak (Wellesley College)
	Gary Leavens (University of Central Florida)
	Peter Sestoft (ITU)
	Chandra Krintz (UC Santa Barbara)

Peter Sestoft not present

Focus on high level initially (ideas) on what:
objectives/outcomes (X's are top five ranking by participants of
most important)
------------------------
-metaprogram / analyze (program as data)	XXXXX
-choose paradigm and write program	XXXXX
specific system and tools	XXXX
computational model	XXXXX
evaluate language XX
loop invariants
limits of program analysis
pl concept implementation X
create/compose modules
learn a new PL XXXX
	



tension between programming methodology and language

different courses: PL, SWE, programming, theory

paradigm as a term: undefined, obsolete attempt on taxonomizing languages, 
avoid its use

programs are a form of expression 

REARCHITECTING:

1) analysis
	understanding a variey of individual concepts and what they are used for
	- what they are, how to deplay
	- use to understand existing languages
	-----------------------------------
	static&dynamic semantics
		- typing rules, syntax, specification
		- how to operate on, analyze, execute
		- pick a model for a problem, evaluate a language
	

2) synthesis
	Integrate concepts into a PL to get something done
	- build a language yourself
	- understand deeply an existing language
	- how to implement
	- how to integrate into a language design
	nice abstract model
	-----------------------------------
	connect specifics to broader concepts and examples
	- formal model must come first -- directly related to abstraction

3) implementation
	interpreter, type checker, parallel scheduler
	concretize concepts into a computing fabric
	notational framework
	-----------------------------------
	parser
	interpreter
	type checker
	translation/compilation
	debugging
	model checking
	tools
	runtime systems
	GC
	tree manipulation (compilers course?)
	sum of product data structures
	pattern matching (compilers course?)

	abstract interpreation (not sure where/how to do this in an ugrad course)
		Europe does it early and often


	parser/interp/type checker: structure of state of a machine
	
	implement a DSL
	understand pieces of a language


What are the concepts (one axis with which to look at the world:)
1) finite data structures
2) functions
3) inifite data structures
4) lambda calculus
5) actor model
6) process calc model
7 concurrency - mode of composition, non-determinism
8 parallelism (deterministic, cost model changes, w/ dynmamics / work-efficient versus speculative (which is concurrency) -- efficiency issue only
9) cost model
10) relational/logic programming model
11) constraint based programming
12 mutations and state
13 laziness
14) modularity and interfaces
15 regex/cfg
16 process calculus

value in education in the long term should be emphasized / our goal

example of these concepts in other areas

function compostion / unix pipes (2)
plan 9, tex (4)
narces, dolev/yao model (4)
modeling business processes 7, 16
network protocols, 15
os scheduleser 7
FSA 15
unix select 7 (nondeterm choice on I/O options)
map reduce 2,3
perl 15
web services 5,6
	locality not really in our list, continuations


- having a variable and replacing it by its binding - substitution (TeX)
- alpha conversion is everywhere (plan9, naming, narces, unguessible secrets,
dolev-yao model (cryptography)
- language features are types
	models of execution (types)

- we have an obligation to be prescriptive

- separate implementation from abstraction

- Appreciate the power of formalism
	ambiance, admin domains, continuations (call stack)
	persistent/ephemeral data structures