Advanced Computer Science Teaching with the Shared Source Common Language Infrastructure

1
Advanced Computer Science Teaching with the
Shared Source Common Language Infrastructure
2
Overview

• In this session we will discuss
• What is the Shared Source Common Language
  Infrastructure (SSCLI)?
• Key SSCLI components
• Differences between SSCLI and .NET
• Working with the SSCLI
• Using the SSCLI for teaching
• Deploying the SSCLI in a teaching lab environment

3
What is the SSCLI?
4
ECMA Common Language Infrastructure (CLI)

• An international standard for a virtual execution
  environment
• produced by Microsoft
• Implementation of this standard is a subset of
  the .NET framework
• The specification (ECMA 335/ISO 23271/2 ) defines
• common type system
• intermediate language
• component model
• A minimum implementation consists of
• a C compiler
• CLI execution engine
• component frameworks

5
SSCLI A Shared Source CLI

• Microsofts free, modifiable and redistributable
  implementation of the CLI, codenamed Rotor
• Platform independent
• Currently executes on MS Windows, FreeBSD and Mac
  OS/X
• Third party ports to Linux, Windows CE and others
• Virtually identical to .NET 1.1 except for
• retooled JIT compiler and garbage collector
• C generics support (Gyro)
• Source written in C, C, C, Perl and assembly
  language
• Other implementations of CLI are Ximians Mono
  and DotGNUs Portable.NET

6
Implemented for Learning

• Many areas of interest (3 million lines of code!)
• Good, real-world example of how software is built
• Contains useful component-based frameworks
• Standard programming constructs
• Web-service oriented networking and remoting
• Contains compilers, tools, and utilities
• C, JScript, and IL JIT compilers
• Assembler, disassembler, linker, debuggers
• Configuration utilities
• Extensive tests and build harness

7
Relationship of SSCLI and .NET Framework

• Both SSCLI and .NET implement complete ECMA
• SSCLI is a superset of ECMA, subset of commercial
  .NET Frameworks
• Shared source CLI implementation differs in
• JIT and garbage collector replaced with more
  portable, more approachable, implementations
• Only supports C and JScript as development
  languages
• Windows-specific features not included
• COM interop
• Windows Forms
• It Just Works!(managed/unmanaged C
  integration)
• Commercially competitive features not included
• ADO.NET (database access)
• enterprise services
• NGEN (install-time, JIT-ahead precompilation)
• ASP.NET (web application / web service
  development)

8
The .NET Framework
9
SSCLI
10
Platform Adaptation Layer

• SSCLI is built on top of a Platform Adaptation
  Layer (PAL)
• The PAL
• isolates CLI implementation from the OS and
  hardware
• runs co-operatively with native code
• implements security and control, threading and
  exceptions
• offers equivalent functionality to many Win32 OS
  functions
• eases porting to new platforms - a student
  project?

11
Working with the SSCLI
12
demo

• Building the SSCLI
• Were setting this off now as it takes a while!

13
The SSCLI Distribution

• SSCLI can be freely downloaded from the Web
• Packaged as single compressed file archive
• 1.9 million lines of code in first release (now
  3 million)
• 1.15M of C and C
• 625K of C
• 125K of CIL (intermediate language)
• Smattering of assembly code, Perl (test
  harnesses) and Tcl/Tk (UI)
• 5900 source files (9700 total)
• 2900 tests
• Build output
• 1200 defined types
• About 20 dynamically loadable libraries
• About 22 executable programs

14
The SSCLI Source Tree

• Four major areas in source code
• Runtime execution engine
• Base class library
• Compilers and tools
• Portability layer, tests, and build
  infrastructure
• Other important points of interest
• License
• Documentation
• Samples

15
SSCLI Execution Engine

• Heart of component-oriented infrastructure
• Converts metadata, resources, and CIL (on disk as
  PE/COFF image) into running code
• JIT compilation and IL verification
• Cross-language exception handling
• Language-agnostic, object-capable, type system
• Automatic heap and stack management
• Dynamic code loading
• Evidence-based security (code access security)

16
Common Intermediate Language

• Basis for most managed code execution in the CLI
• Simple stack-based machine model see vm
  directory
• Typeless opcodes, defined in opcode.def
• Verifiable subset see fjit directory for
  verification
• Signed and unsigned via opcode, not type
• CIL code is JIT compiled to run under CLI
  management
• Type layout, control, and dispatching
• Typed variable argument lists, dynamically typed
  pointers
• Tail calls, virtual dispatch, call via function
  pointer
• Rich set of conversion operations, calling
  conventions
• Exception handling
• Garbage collection

17
Frameworks

• A minimal toolkit for the 21st century
  programmer
• Base class library is a key feature of the ECMA
  CLI spec
• Collections, arrays, strings, and other compound
  data types
• Globalization and formatting
• System services (threads, I/O, synchronization,
  etc.)
• Security
• SSCLI augments this with
• Floating point and extended arrays libraries
• Networking, regular expressions, and XML
  libraries
• Access to runtime infrastructure
• Reflection and custom attributes
• Remoting, unmanaged interop, serialization, and
  marshaling
• AppDomains, Assemblies, GC, other execution
  engine features

18
Compilers and Tools

• Full-featured C compiler
• also used in build process
• JScript compiler written entirely in C
• Additional developer tools
• clix, the shared source CLI program launcher
• Assembly tools resource compiler, assembly
  linker, metainfo metadata viewer, assembler,
  disassembler
• Debuggers cordbg managed command line debugger,
  plus a debug extension for working on managed
  code from C/C via windbg

19
demo

• Developing software with the SSCLI
• Debugging with cordbg

20
SSCLI Build Process

• Complex build which occurs in two phases and is
  driven by several build tools
• nmake, build, Perl scripts
• Initial phase builds
• PAL using native toolchain
• Build tools (nmake, build) against PAL for
  portability
• resource compiler and PAL runtime using newly
  compiled build tools
• Main phase has complex interdependencies
• Unmanaged frameworks, compilers, and tools, are
  built
• Managed portions of runtime and frameworks are
  built using compilers built previously
• Assemblies and runtime are configured
• Possible to rebuild individual SSCLI components
• Allows teaching activities to focus on smaller
  parts of the SSCLI and rebuild those only
• e.g. Hull JScript compiler modification exercises
• Good example of how large software systems are
  built
• Unfamiliar to most students

21
SSCLI Test Suite

• Incredibly handy when modifying code!
• Two major areas managed code and PAL
• PAL suites
• Test driver script is pvtrun.pl
• Used on new PAL implementations to verify
  compliance with PAL specification
• Quality suites
• Test driver script is rrun.pl
• Tests for three primary aspects of SSCLI
  functionality
• IL verification
• BVT smoke tests many small programs
• Tests for the BCL

22
Using the SSCLI for Teaching
23
Possibilities for Teaching

• The SSCLI implementation touches upon many
  aspects of a standard Computer Science syllabus
• Programming languages and skills
• Object-oriented development
• Framework design
• Software testing
• Security
• Networking
• Virtual machine implementation
• etc, etc, etc
• Many possible uses in the curriculum
• Allows students to see the relevance of many of
  the aspects of the curriculum in the
  implementation of a large commercially-produced
  software product

24
The SSCLI at Hull

• Hull University uses both the .NET Framework and
  the SSCLI for teaching at Master level
• MSc in Distributed Systems Development
• .NET provides
• overreaching example of distributed systems
  concepts, techniques and technologies
• SSCLI provides
• down to the metal implementation details
• excellent environment for enhancing testing/
  debugging skills
• We introduce three ways in which the SSCLI is
  used at Hull

25
MSc Distributed Systems Development

• Our Masters course aims to
• Give advanced coverage of modern distributed
  computing techniques
• Develop an understanding of contemporary managed
  code environments and how they work
• Develop skills in working with large codebases
• Develop active practitioners
• Provide hands-on practical experience underpinned
  by advanced theoretical concepts

26
Software Maintenance Exercise

• Designed to improve students skills in
• Navigating and comprehending unfamiliar code
• Maintaining code written by others
• Designing and implementing for maintenance
• Requires students to make modifications to the
  JScript compiler
• Modify the abstract syntax tree of the compiler
• JScript compiler written in C gt easy
  introduction to SSCLI
• JScript compiler can be built and deployed
  independently

27
JIT Profiling Exercise

• Designed to provide
• an introduction to key aspects of the virtual
  machine and JIT compiler implementation
• An appreciation for how JIT compilation works to
  transform intermediate code to native code
• A basis for further, more detailed study
• Requires students to
• Introduce new data structures
• Modify JIT compiler to count CIL instructions
• Sort and output profiling statistics on program
  completion
• Code in an unmanaged language (unfamiliar to many)

28
Debugging Exercise

• Optional exercise for students wanting to improve
  their debugging skills which is designed to
• Get the students thinking about all aspects of
  the compilation and execution cycle
• Provide opportunities to develop debugging and
  testing skills
• Familiarise the student with the SSCLI code and
  the techniques used to implement it
• Introduce problem solving and hypothesis testing
  skills relevant to debugging
• Set the worst case scenario
• Deliberate bugs introduced in the underlying
  language and VM implementation students code is
  fine, but it still doesnt work!

29
Our Experience

• Feedback suggests that the students
• Find the exercises initially challenging
  (particularly the JScript exercise)
• Value what these exercises have taught them as
  they reach the latter stages of our Masters
  programme
• Observe an improvement in their own coding, code
  comprehension and debugging skills as a direct
  consequence of being exposed to commercially
  produced code

30
Deploying the SSCLI
31
SSCLI in the Teaching Lab

• The SSCLI is really designed for a single
  developer working exclusively on a single machine
  - i.e. not the normal lab environment
• Students use a different machine each session
• Modifications made to the SSCLI by one student
  must not be visible to other students
• Complete built installation requires 1GB disk
  space
• Building on a network drive gt high network
  traffic
• Managed code executed from a network drive is
  less trusted than code run from the local machine
  and may generate security exceptions

32
Our Solutions

• Provide two local copies of the SSCLI
  distribution
• Students have read/write permissions on one copy,
  read permissions on the other
• Students use local modifiable copy for their work
• Read-only copy manually copied over modifiable
  copy at start of each session and automatically
  each evening
• Use source control system
• Students use a source control system to store
  their changes to SSCLI components (not entire
  source tree)
• Changes checked out over clean local copy at
  start of session
• Changes checked back in at start of sessions
• Promotes good development practice but
  significant overhead in administering source
  control system
• Virtual PC (with Undo Disks) ?

33
Windows Deployment Issues

• SSCLI requires Visual C to build under Windows
• Will not build against Visual C 8 / Visual
  Studio 2005
• Vanilla distribution does not execute correctly
  on Windows XP Service Pack 2
• Changes made to XP in SP2 break the exception
  handling mechanism
• CDs issued today have fixed copy of SSCLI source
• Build harness attempts to create file association
  for Perl files
• Our configuration does not permit students to
  create file associations
• Easily fixed by removing a line from the build
  scripts

34
Summary

• In this presentation we have introduced
• The SSCLI and its source code distribution
• SSCLI organization and operation
• Building and using the SSCLI
• Opportunities for teaching using the SSCLI
• Teaching lab deployment issues

35
Reading and Resources

• Reading
• Stutz, Neward Shilling, Shared Source CLI
  Essentials, OReilly Associates Inc, 2003
• Resources
• Shared Source CLI
• http//msdn.microsoft.com/msdnmag/issues/02/07/sha
  redsourcecli/
• ECMA Specifications
• http//msdn.microsoft.com/net/ecma
• David Stutzs MSDN announcement of Rotor
• http//msdn.microsoft.com/library/default.asp?url
  /library/en-us/Dndotnet/html/mssharsourcecli.asp
• Jason Whittingtons MSDN Rotor article
• http//msdn.microsoft.com/msdnmag/issues/02/07/sha
  redsourcecli/