What Makes a Good Language?

1
What Makes a Good Language?

• Lecture 3

2
The first formal languages

• The evolution of writing systems to record
  language reveals the first steps toward formal
  notation and unambiguous communication.
• The worlds oldest written documents are in
  Sumerian - small 3 dimensional tokens used for
  counting things like grain, wine and cattle
• At first writing was composed of individual
  pictographs

3
The first formal languages

• Gradually civilization moved to ideographs,
  symbols for ideas
• meaning understood by convention, not by the
  picture
• earliest known algorithms are due to Babylonians
  - a base 60 notation for arithmetic, including
  fractions

4
Designs of High-level Computer Languages

• Mechanisms to design still must be perfected
• every known language has its shortcomings
• usually the more popular languages have less
  shortcomings

5
Popularity

• some reasons for success or failure of language
  may be external
• a government mandate (COBOL or Ada)
• strong computer manufacturer support (FORTRAN)
• excellent text (Griswolds SNOBOL4 text)
• use as theoretical study (Pascal and LISP)

6
Attributes of a Good Language

• Clarity, simplicity, and unity
• language provides both a conceptual framework for
  thinking about algorithms and a way of expressing
• aid the programmer before coding
• unified set of primitives used in developing
  algorithms
• conceptual integrity - minimum number of
  different concepts and rules for combining to be
  as simple and regular as possible
• FORTRAN had 3 but not higher dimensional arrays

7
Syntax

• Affects the ease with which a program may be
  written, tested and later understood and modified
  (readability)
• many languages contain syntactic constructs that
  look identical put are radically different.
  Example, SNOBOL, a presence of a blank char,
  which is an operator, in a statement may entirely
  alter its meaning. Or a(1) versus a1
• How many ways can you write in C to add to a
  variable x? or as

8
Famous FORTRAN example

• Do 10 I 1.5
• a(I) X B(I)
• 10 continue
• FORTRAN ignores all spaces
• so this was interpreted to
• do10i 1.5
• destroyed a rocket!

9
Hamming code

• Provides the greatest distance between
  representations
• catch errors easily
• each representation is at least separated by two
  distinct places
• Notice the syntax of the Do was too close to
  the syntax of the assignment.

10
Orthogonality

• Able to combine various features of a language in
  all possible combination and still make sense.
• Expressions and conditionals are orthogonal if
  any expression can be used within the conditional
  statement
• this make a language easier to learn (fewer
  exceptions and special cases)
• also program will compile without errors even if
  it contains a combination that does not make
  sense
• this is why orthogonality is controversial

11
Naturalness for the application

• Program structure reflects the underlying logical
  structure
• provide appropriate data structures, operations
  and control structures

12
Support for abstraction

• The jump made from the abstract data structures
  to the solution built with the language data
  structures and operations
• class scheduling - need the structures of
  student, class section, instructor, lecture room
  - need the operation of assign student to a class
  section
• what support is provided?
• Programmer should use language self-contained
  abstractions without knowing details of
  implementation
• It is desirable to have constructs that factor
  out recurring patterns, e.g., subprograms

13
Ease of program verification

• Reliability is a central concern
• verification - a program correctly performs its
  required function
• proven correct by a formal verification method
• informally by desk checking
• tested with test input data and results checked
  against specifications

14
Programming environment

• Presence of an appropriate programming
  environment may make a technically weak language
  easier to work with
• special editors, testing packages
• facilities for maintaining, modifying multiple
  versions

15
Portability of programs

• Transportability - independent of the features of
  a particular machine
• standardized definitions
• Ada, FORTRAN, C all have standardized definitions
• ML come from single source implementation
  allowing the language designer some control over
  portable features of the language

16
Cost of use

• Program execution - optimizing compilers,
  efficient register allocation, efficient run-time
  support
• important in large production programs that will
  be executed repeatedly
• Cost of program translation - compiled many times
• Cost of program creation, testing, and use - a
  inefficient language may take less programmer
  time (Smalltalk)

17
Cost of use

• Cost of program maintenance - greatest amount of
  dollars spent

18
Language Standardization

• What describes a programming language?
• int i i (1 2 ) 3
• Is this valid C? What is the the value of i?
• How would you answer this?
• Read the definition in the language reference
  manual to decide what the statement means.
• Or Write the program on your local computer
  system to see what happens
• or read the definition in the language standard

19
Language Standardization

• The first and second option are tied to a
  particular implementation. Is that
  implementation correct?
• Often, language design involves some intricate
  details and one vendor may have a different
  interpretation from another, yielding a slightly
  different execution behavior.
• What about additional features

20
Standards

• Proprietary Standards
• Consensus Standards
• In the US, the American National Standards
  Institute, ANSI
• programming language standards are assigned to
  committee X3 of the Computer Business Equipment
  Manufactures, or CBEMA
• partially technical and partially political
• consensus process

21
The FORTRAN 77 string problem

• Strings and substrings were desirable features
• most implementations already had such features
• several feasible implementations of substrings
• if M abcdefg then bcde could be M25 or
  M24 (start at 2 and go 4 chars) or start at
  the right M36
• no consensus, left out (expedient)

22
Standards

• Timeliness.
• When do we standardize a language?
• Conformance.
• What does it mean for a program to adhere to a
  standard and for a compiler to compile to a
  standard?
• Obsolescence.
• When does a standard age and how does it get
  modified

23
Timeliness

• FORTRAN was standardize in 1966 when there many
  incompatible versions
• Ada was standardize before any implementations
• balance
• LISP - no standard reference (Scheme, Common
  LISP, IBCL) all exist are similar, but
  incompatible

24
Conformance

• A program is conformant if it only uses features
  defined in the standard
• A conforming compiler is one which when given a
  conformant program, produces an executable
  program that produces the correct output
• However, extensions can be added and programs can
  use those extensions.

25
Obsolescence

• Standards have to be reviewed every five years
  and either be renewed or dropped
• most standards require backward compatibility
• A feature is obsolescent if it is a candidate
  feature that may be dropped in the next version
  of the standards
• warns users that the feature is still available,
  but in the next 5 to 10 years will be dropped.
  Fair warning to rewrite code using that feature

26
Obsolescence

• A deprecated feature may become obsolescent with
  the next standard, hence may be dropped after two
  revisions
• gives a 10-20 year warning.
• New programs should not use either class of
  features.