Preliminaries

Report, Progress.

Make sure and submit. I also want to know that you’re on track.

How’s that going? Someone stuck?

Additional Bonus - TRACE surveys

Reminder: If TRACE eval scores completion % >= 85% I’ll add 2 overall grade points to class grade average.

Last time

Implementing our own lambda calc language Implementing languages in terms of host languages.

Qs

Types

Why and where?

Perhaps the most pervasive formal methods are those that are not even viewed as formal methods anymore! These include the pervasive use of static type systems in mainstream languages like Java and C# ….

– Ranjit Jhala et al. Formal Methods Successes and Directions

A division and a trade-off

A division and a trade-off between the heavyweight formalisms, for general correctness properties, but often cumbersome for programmers to use, and require lots of sophistication from the programmer.

A kind of lightweight automation.

On the other hand, we have the kinds of less powerful but easier to apply, implement and use. Language implementers can build these into automated language tooling

Programmers who don’t know the formalisms can still use them!

By far the most common of these are type systems.

A type system is a tractable syntactic method for proving the absence of certain program behaviors by classifying phrases according to the kinds of values they compute. Benjamin Pierce, TAPL

Types: not just for CS

  • Russell to avoid antimony (1902)
  • STLC 1940
  • More expressive type systems (Per Martin-Lof + forward) (open research questions still!!)

Note also: classifies terms wrt run-time values

Type systems provide a static approximation of the program (fragments)’ run-time behavior

Compositionality

We calculate the types of “bigger” terms compositionally, from only the types of the immediate subexpressions.

Static > Conservative approxmation of what you want.

A type system can assure us that accepted programs don’t have certain errors, but doesn’t mean that the rejected programs do have those errors.

“Certain errors”

Rules out specific kinds of bad behaviors, not all bad behaviors.

  • Right # args? yes
  • Div-by-0? Prolly not!

“Bad behaviors” ~= “run-time type errors.”

Typically checkers built-in and fully automated

Lots of room, space for design

benefits (Ancillary)

  • Correct Documentation
  • Enforce Abstraction
  • *Detecting errors
  • Surprisingly useful

Inference rule (schemas)

These “inference rules” are actually rule schemas; each schema represents the infinite set of concete rules that we can obtained consistently by replacing each metavariable by all phrases from the appropriate syntactic category

“If we have established the statements in the premise(s) listed above the line, then we may derive the conclusion below the line.”

These can be inferences about values, or communicating processes, or in our case, types.

————————————————-

Last time:

We established

  1. t ∈ Terms
  2. v ∈ Values
  3. An evaluation function.
  4. We “disallowed” bad programs. We didn’t say what to do.

General structer “inference rules”

  aoeuaoeuo   345345345443
  ————————————————-
  choeurh

Writing some rules, for example!

  ne : Nat 
————————————————-
 (zero? ne) : Bool


  ne1 : Nat    ne2 : Nat 
———————————————-—————————
   (+ ne1 ne2) : Nat


————————————————————
 true : Bool

————————————————————
 false : Bool

—————————————————— (natural? ne)
 ne : Nat 


 ne : Nat
—————————————————— 
 (sub1 ne) : Nat 

  t : Bool c : τ  a : τ
—————————————————— 
  (if t c a) : τ 

Now

The most basic property of type systems is safety (also called soundness): well-typed terms do not “go wrong.”

We have already chosen how to formalize what it means for a term to go wrong: it means reaching a “stuck state” (Definition 3.5.15) that is not designated as a final value but where the evaluation rules do not tell us what to do next.

What we want to know, then, is that well-typed terms do not get stuck.

We show this in two steps, commonly [3] known as the progress and preservation theorems.

Progress: A well-typed term is not stuck (either it is a value or it can take a step according to the evaluation rules).

(base)

Preservation: If a well-typed term takes a step of evaluation, then the resulting term is also well [4] typed.

(induction)

Conservativity vs. expressiveness

Job security: wanting to permit more programs to type check without devolving into chaos or further overburdening the programmer, that’s the good stuff.

Updated: