Is there a good programming language out there

1
Is there a good programming language out there?

• Allen TuckerSeptember 19, 2008
• http//www.bowdoin.edu/allen

2
A good programming language is a conceptual
universe for thinking about programming. A.
Perlis

• This raises four questions
• 1. Whats in this conceptual universe?
• 2. How should we think about programming?
• 3. What should a good programming language look
  like?
• 4. Is there a good programming language?

3
1. A languages conceptual universe

• Concepts about computing environment
• Machine architecture/platform
• Run-time behavior
• User interface
• Networking clients and servers
• Data persistence
• Concepts about programmers and programming
• Application domain
• Programmer skills
• Organization individual vs team
• Program goals

4
2. How should we think about programming?

• Programming combines modeling and coding.
• Elements of modeling
• Literate style
• Programming in the large (frameworks)?
• System and network platform
• Application requirements and use cases
• Elements of coding
• Syntax and semantics
• Abstraction
• Efficiency
• Development environment (IDE, version control)?
• Correctness and documentation

5
3. What should a good programming language look
like?

• Historical influences
• Application domains
• Programming paradigms
• Quality metrics

6
3a) Historical influences

• Programming languages evolve over time
• 1960 - there were 4 (count em) major languages
• Algol, Fortran, Cobol, and Lisp
• 1977 - I wrote a book that covered 6 prominent
  languages
• Algol, Fortran, Cobol, PL/I, RPG, and Snobol
• 1986 - the 2nd edition covered 11 languages
• Pascal, Fortran, Cobol, PL/I, Snobol, APL, C,
  Lisp, Prolog, Ada, Modula-2
• 2001 Bob Noonan and I wrote a book covering 4
  languages and 6 paradigms
• Java, C, Scheme, Prolog
• 2007 - the 2nd edition covered 9 languages
• Java, C, Ada, Perl, Smalltalk, Python, Scheme,
  Haskell, Prolog

7
(No Transcript)
8
3b) Application domains also have language
preferences

• Artificial Intelligence
• Scheme, C, Prolog
• Computer Science Education
• C, C, Java, Python
• Science and Engineering
• Fortran, C
• Information Systems
• Cobol, SQL
• Systems and Networks
• C, C, Ada
• World Wide Web
• Perl, PHP, Python

9
3c) Programming Paradigms

• A paradigm is a problem-solving pattern that
  utilizes a conceptual universe for thinking about
  programming.
• So a paradigm utilizes a programming language.
• Here are six very different paradigms
• Imperative
• Object-oriented
• Functional
• Logic (declarative)?
• Event-driven
• Concurrent
• But different paradigms utilize different
  languages.

10
Imperative Programming

• Based on the von Neumann-Eckert computing model
• Program and data coexist in memory
• Program a sequence of commands
• State values of all variables at run time
• Large programs use procedural abstraction
• Example imperative languages
• Cobol, Fortran, C, Ada, Perl

11
The von Neumann-Eckert Computing Environment
12
3d) Quality Metrics

• A good language should have ten qualities
• Literacy
• Semantic clarity
• Rich primitive types and structures
• Abstraction
• Reliability
• Efficiency
• Objects and events
• Concurrency in the large
• Correctness tools
• Support

13
Quality 1. Literacy

• Writable
• Small vocabulary (instruction set)?
• E.g., Java vs. Scheme
• Simple syntax
• E.g., C/C/Java vs. Python
• Readable
• Clear structure and naming conventions
• Documentation support

14
Quality 2. Semantic clarity

• A binding associates an object with a property.
  Examples a variable and its type, or
• a variable and its value
• A language is strongly typed if the type of every
  variable in a program is known whenever it is
  referenced.
• Strong typing is related to clarity about
  binding, which leads to semantic clarity.
• A language is orthogonal if its semantics are
  built on a
• small, mutually independent set of primitive
  operations.
• Fewer exceptional rules lead to semantic clarity

15
Quality 3. Rich Primitive Types and Structures

• Numbers
• Currency
• Strings
• Dates and times
• Lists and list functions
• Hash maps (associative arrays)?
• Traversals
• Relational databases

16
Quality 4. Abstraction

• Data
• Programmer-defined types/classes
• Class libraries
• Procedural
• Programmer-defined functions
• Function libraries

17
Quality 5. Reliability

• A language is reliable if
• A program behaves the same when run on different
  platforms
• E.g., Fortran 95 vs Fortran I
• Type errors are detected
• E.g., Haskell vs C
• Semantic errors are properly trapped
• E.g., C vs C
• Memory leaks are prevented
• E.g., Java vs C

18
Quality 6. Efficiency

• Embedded systems
• Real-time responsiveness (e.g., navigation)?
• Failures of early Ada implementations
• Web applications
• Responsiveness to users (e.g., Google search)?
• Database applications
• Efficient search and updating
• AI applications
• Modeling human behaviors

19
Quality 7. Objects and Events

• An object oriented (OO) program models a
  collection of objects that interact by passing
  state-transforming messages.
• OO is a good way to model real-world
  applications.
• Example OO languages
• Java, C, C, Python, PHP

20

Events

• Computation is best modeled as a community of
  persistent entities coupled together by their
  ongoing interactive behavior ... Beginning and
  end, when present, are special cases that can
  often be ignored. Lynn Stein
• Event-driven programs do not control the sequence
  in which input events occur.
• Example event-driven languages
• Perl, Python, Java applets, PHP forms

21
Imperative vs. Event-Driven Programs
22
Event-driven Applications

• GUI
• Games
• Graphical design
• Web-based
• E-commerce (e.g., amazon.com)?
• Volunteer scheduling (e.g., hfoss.bowdoin.edu)?
• Embedded systems
• Cell phones
• Security alarm systems

23
Quality 8. Concurrency in the large

• Two or more instruction streams (threads) running
  asynchronously and sharing data.
• Characteristics
• More efficient than single instruction stream
• More realistic
• Have unique, fundamental complexities
• Traditionally studied in the context of operating
  systems

24
Concurrency example web pages

• To render a web page (a shared resource), we
  need
• A thread for loading each image
• A thread for rendering text
• A thread for responding to user input
• e.g., a stop button
• Threads cannot all write to the page
  simultaneously
• They need asynchronous control
• But they must be concurrent
• They must be responsive to the user

25

• Asynchronous control is needed
• Whenever a thread requires exclusive access to
  some resource,
• Whenever a thread needs to exchange data with
  another thread,
• To prevent the resulting value of a shared
  variable from being different depending on the
  execution order of two threads (a race
  condition)?
• To prevent a thread from waiting for an event
  that will never happen (a deadlock).

26

• Concurrency in the large
• the client-server model

client
client
client
network
server
Shared resource
27
Quality 9. Correctness tools

• To treat programming scientifically, it must be
  possible to specify the required properties of
  programs precisely.
• The importance of formal specifications must
  ultimately rest in their utility - in whether or
  not they are used to improve the quality of
  software or to reduce the cost of software.
• J. Horning

28
Why correctness?

• A correct program does exactly what its
  specifications require, no more and no less.
• A formally correct program is one whose
  correctness can be proved mathematically.
• Until recently, correctness has been just an
  academic exercise.
• Now it is becoming a key element of
  safety-critical systems
• Correctness can ensure higher levels of quality
  than traditional testing
• e.g., the Paris Metro System

29
Practical problems with correctness

• Scaling up
• Teaching the methodology
• Practicing the methodology
• Lack of good tools
• Duplication of effort
• E.g., The Java elements JML and ESC/Java2
  support
• Correctness proof (ESC/Java2)
• Complements traditional Unit testing (Junit)?
• Modeling (JML)
• Complements traditional Coding (Java)?
• Documentation (JML)
• Complements traditional Documentation (Javadoc)?

30
Quality 10. Support

• A well-supported language has
• Free access (open source)?
• Good texts and tutorials
• Significant application domain (and programming
  community)?
• Good development environments (IDEs)?

31
4. Is there a good programming language?

32
4. Is there a good programming language?

• Yes

33
4. Is there a good programming language?

• Yes
• My vote goes to PHP/Apache/MySQL

34
But wait, you say

• This isnt a language
• PHP and MySQL are two different languages, and
• Apache is a web server.

35
But wait, I say

• This is a language
• PHP and MySQL are well-integrated, and
• We cant program client-server applications
  (concurrency in the large) without a web server.
• Anyway, let's assume (for arguments sake) that
  PHP/Apache/MySQL is a language,
• So now we can look at PHP/Apache/MySQL through
  the lens of our 10 quality criteria.

36
PHP/Apache/MySQLs qualities

• Literacy yes
• Semantic clarity yes -- even w/dynamic types
• Rich primitive types and structures yes --
  see next slide
• Abstraction yes
• Reliability yes
• Efficiency yes
• Objects and events yes
• Concurrency in the large yes
• Correctness tools no -- see 2nd/3rd next slides
• Support yes

37
Examples of PHP/Apache/MySQLs Rich Primitive
Types and Structures

• Dates and times
• today date()
• september9 mktime(0,0,0,9,19,2008)
• Associative Arrays
• population array(CTgt3502309, MAgt
  6449755,
• RI gt 1057832, ME gt 1317207,
• NH gt 1315828, VT gt 621254)
• So populationMA gives 6449755
• MySQL integration
• my_query SELECT state, pop FROM censusData
• result connection-gtquery(my_query)
• population mysql_fetch_array(result,
  MYSQL_ASSOC)

38
The challenge for integrating correctness tools

• Documentation is essential for software
  development
• Javadoc (and PHPdoc) have keywords for
  specifying features of classes and methods.
  E.g.,
• _at_return specifies value returned by a function
• _at_param specifies parameters required by a method
• Formal specifications are essential for
  correctness proofs
• JML has features for specifying the same. E.g.,
• \ensures and \return clauses
• \requires clauses
• But these two are not integrated.
• So the programmer has to write all these
  specifications twice!

39
PHP/Apache/Mysqls Main Weakness

• Lack of correctness tools
• Suggestions for fixing this weakness
• Integrate formal specifications within PHPdoc
• motivator for writing better documentation
• supplement for unit testing
• preparation for more formal proof
• Use JML and ESC/Java2 experiences for guidance

40
A PHP/Apache/MySQL Application

• RMH Homebase A volunteer scheduling for the
  Ronald McDonald House in Portland, ME
• Web-based, client-server
• Database
• Designed and Programmed by four Bowdoin students
  (and me)?
• Open source (see http//sourceforge.net/projects/r
  mhhomebase)?
• Fully developed during spring semester 2008
• See http//hfoss.bowdoin.edu and login as guest
  (no password) to experiment with the system.

41
Conclusions

• If you become a computing professional
• prepare to spend a lot of time learning
  languages.
• Expect programming languages to be the last to
  keep pace with the evolution of computing.
• Expand your notion of programming language
  beyond the conventional.
• Have fun with programming languages.
• They will have fun with you!

42
References

• Tucker and Noonan, Programming Languages 2/e,
  McGraw-Hill 2007
• http//highered.mcgraw-hill.com/sites/0072866098/
• 2. PHP/Apache/MySQL Web sites
• http//www.php.net/
• http//www.apache.org/
• http//www.mysql.com/
• PHP/Apache/MySQL Software Projects
• http//hfoss.bowdoin.edu/
• http//www.hfoss.org/

43
Questions?