CompSci497
Advanced XML Programming with XmlPL
Lecture Notes
Friday, January 26th, 2007

Contents

1. An Introduction to SAX, DOM, XSLT, & XQuery.
   1. SAX
   2. DOM
   3. XSLT
   4. XQuery
2. A justification for XmlPL
   1. SAX
   2. DOM
   3. XSLT
   4. XQuery
   5. XmlPL
      1. Comparisons
3. XmlPL Architecture
   1. Compiler
      1. xmlplcc: command line
   2. Runtime
   3. Standard Lib

• Simple API for XML
• Event driven
• Defines a set of function callback
• startElement, endElement, startDocument, endDocument, text, processingInstruction, etc.

• Document Object Model
• Language independent
• Defines XML as a tree of objects
• Defines valid operations (methods) on those objects

• Extensible Stylesheet Language Transformations
• Template processor
• Designed to transform XML documents
• Predecessor of: Document Style Semantics and Specification Language (DSSSL)
• http://en.wikipedia.org/wiki/XSLT for examples

• XML Query Language
• semantically similar to SQL
• FLWOR: FOR, LET, WHERE, ORDER BY, RETURN
• http://en.wikipedia.org/wiki/XQuery for example

• Fast and efficient for small tasks.
• Does not allow linear code execution.
• Difficult to relate different parts of an XML document
• Must maintain stacks of context information
• No output processing

• Resource hog.
• XML DOM is often 6 or more times larger in memory than on disk
• APIs generally don't allow for partial loading
• DOM APIs are clumsy
• No output processing

• Verbose: XML is not good at describing computation
• Slow
• Designed mainly for document transformation
• Unable to update XML trees

• Designed for querying XML documents as databases
• Messy SQL like syntax
• Unable to update XML trees
• Not widely supported

• Efficient single pass processing
• Linear code execution
• Allows user to control DOM memory usage
• Internal DOM representation used less memory
• Output processing guarantees well formed XML output.
• Clean C-like syntax
• Allows in memory updating as well as on-the-fly processing
• Designed for general purpose XML processing

• XmlPL vs. SAX: Flour (source)
• XmlPL vs. DOM: FirstRecipeDOM (source)
• XmlPL vs. XSLT: persons (source)
• XmlPL vs. XQuery: hamlet (source) (input source)

• Written in C++
• Antlr parser generator
• AST: Abstract Syntax Tree
• Environment
• Library importer
• C++ code generator
• Should either report an error and fail or produce compilable C++ code
• Known bugs exist (and likely unknown)

• Parses code, generates C++, calls g++ to build
• Able to build standalone or library
• Can dump a stack trace on errors (-t option)
• Able to dump an XML parse tree

• Also written in C++
• Boehm-Demers-Weiser conservative garbage collector
• libxml2 parser and output processing
• Custom low-memory DOM
• Pull parser on top of SAX
• Efficient Sequence implementation
• libcurl network access

• currently fairly limited
• curl, gen, math, process, stdio, stdlib, string, unistd, xml
• Easy to expand