You are previewing Beautiful Code.

Beautiful Code

Cover of Beautiful Code by Greg Wilson... Published by O'Reilly Media, Inc.
  1. Beautiful Code
    1. SPECIAL OFFER: Upgrade this ebook with O’Reilly
    2. A Note Regarding Supplemental Files
    3. Foreword
    4. Preface
      1. How This Book Is Organized
      2. Conventions Used in This Book
      3. Using Code Examples
      4. How to Contact Us
      5. Safari® Enabled
    5. 1. A Regular Expression Matcher
      1. 1.1. The Practice of Programming
      2. 1.2. Implementation
      3. 1.3. Discussion
      4. 1.4. Alternatives
      5. 1.5. Building on It
      6. 1.6. Conclusion
    6. 2. Subversion's Delta Editor: Interface As Ontology
      1. 2.1. Version Control and Tree Transformation
      2. 2.2. Expressing Tree Differences
      3. 2.3. The Delta Editor Interface
      4. 2.4. But Is It Art?
      5. 2.5. Abstraction As a Spectator Sport
      6. 2.6. Conclusions
    7. 3. The Most Beautiful Code I Never Wrote
      1. 3.1. The Most Beautiful Code I Ever Wrote
      2. 3.2. More and More with Less and Less
      3. 3.3. Perspective
      4. 3.4. What Is Writing?
      5. 3.5. Conclusion
      6. 3.6. Acknowledgments
    8. 4. Finding Things
      1. 4.1. On Time
      2. 4.2. Problem: Weblog Data
      3. 4.3. Problem: Who Fetched What, When?
      4. 4.4. Search in the Large
      5. 4.5. Conclusion
    9. 5. Correct, Beautiful, Fast (in That Order): Lessons from Designing XML Verifiers
      1. 5.1. The Role of XML Validation
      2. 5.2. The Problem
      3. 5.3. Version 1: The Naïve Implementation
      4. 5.4. Version 2: Imitating the BNF Grammar O(N)
      5. 5.5. Version 3: First Optimization O(log N)
      6. 5.6. Version 4: Second Optimization: Don't Check Twice
      7. 5.7. Version 5: Third Optimization O(1)
      8. 5.8. Version 6: Fourth Optimization: Caching
      9. 5.9. The Moral of the Story
    10. 6. Framework for Integrated Test: Beauty Through Fragility
      1. 6.1. An Acceptance Testing Framework in Three Classes
      2. 6.2. The Challenge of Framework Design
      3. 6.3. An Open Framework
      4. 6.4. How Simple Can an HTML Parser Be?
      5. 6.5. Conclusion
    11. 7. Beautiful Tests
      1. 7.1. That Pesky Binary Search
      2. 7.2. Introducing JUnit
      3. 7.3. Nailing Binary Search
      4. 7.4. Conclusion
    12. 8. On-the-Fly Code Generation for Image Processing
    13. 9. Top Down Operator Precedence
      1. 9.1. JavaScript
      2. 9.2. Symbol Table
      3. 9.3. Tokens
      4. 9.4. Precedence
      5. 9.5. Expressions
      6. 9.6. Infix Operators
      7. 9.7. Prefix Operators
      8. 9.8. Assignment Operators
      9. 9.9. Constants
      10. 9.10. Scope
      11. 9.11. Statements
      12. 9.12. Functions
      13. 9.13. Array and Object Literals
      14. 9.14. Things to Do and Think About
    14. 10. The Quest for an Accelerated Population Count
      1. 10.1. Basic Methods
      2. 10.2. Divide and Conquer
      3. 10.3. Other Methods
      4. 10.4. Sum and Difference of Population Counts of Two Words
      5. 10.5. Comparing the Population Counts of Two Words
      6. 10.6. Counting the 1-Bits in an Array
      7. 10.7. Applications
    15. 11. Secure Communication: The Technology Of Freedom
      1. 11.1. The Heart of the Start
      2. 11.2. Untangling the Complexity of Secure Messaging
      3. 11.3. Usability Is the Key
      4. 11.4. The Foundation
      5. 11.5. The Test Suite
      6. 11.6. The Functioning Prototype
      7. 11.7. Clean Up, Plug In, Rock On…
      8. 11.8. Hacking in the Himalayas
      9. 11.9. The Invisible Hand Moves
      10. 11.10. Speed Does Matter
      11. 11.11. Communications Privacy for Individual Rights
      12. 11.12. Hacking the Civilization
    16. 12. Growing Beautiful Code in BioPerl
      1. 12.1. BioPerl and the Bio::Graphics Module
      2. 12.2. The Bio::Graphics Design Process
      3. 12.3. Extending Bio::Graphics
      4. 12.4. Conclusions and Lessons Learned
    17. 13. The Design of the Gene Sorte
      1. 13.1. The User Interface of the Gene Sorter
      2. 13.2. Maintaining a Dialog with the User over the Web
      3. 13.3. A Little Polymorphism Can Go a Long Way
      4. 13.4. Filtering Down to Just the Relevant Genes
      5. 13.5. Theory of Beautiful Code in the Large
      6. 13.6. Conclusion
    18. 14. How Elegant Code Evolves with Hardware The Case of Gaussian Elimination
      1. 14.1. The Effects of Computer Architectures on Matrix Algorithms
      2. 14.2. A Decompositional Approach
      3. 14.3. A Simple Version
      4. 14.4. LINPACK's DGEFA Subroutine
      5. 14.5. LAPACK DGETRF
      6. 14.6. Recursive LU
      7. 14.7. ScaLAPACK PDGETRF
      8. 14.8. Multithreading for Multi-Core Systems
      9. 14.9. A Word About the Error Analysis and Operation Count
      10. 14.10. Future Directions for Research
      11. 14.11. Further Reading
    19. 15. The Long-Term Benefits of Beautiful Design
      1. 15.1. My Idea of Beautiful Code
      2. 15.2. Introducing the CERN Library
      3. 15.3. Outer Beauty
      4. 15.4. Inner Beauty
      5. 15.5. Conclusion
    20. 16. The Linux Kernel Driver Model: The Benefits of Working Together
      1. 16.1. Humble Beginnings
      2. 16.2. Reduced to Even Smaller Bits
      3. 16.3. Scaling Up to Thousands of Devices
      4. 16.4. Small Objects Loosely Joined
    21. 17. Another Level of Indirection
      1. 17.1. From Code to Pointers
      2. 17.2. From Function Arguments to Argument Pointers
      3. 17.3. From Filesystems to Filesystem Layers
      4. 17.4. From Code to a Domain-Specific Language
      5. 17.5. Multiplexing and Demultiplexing
      6. 17.6. Layers Forever?
    22. 18. Python's Dictionary Implementation: Being All Things to All People
      1. 18.1. Inside the Dictionary
      2. 18.2. Special Accommodations
      3. 18.3. Collisions
      4. 18.4. Resizing
      5. 18.5. Iterations and Dynamic Changes
      6. 18.6. Conclusion
      7. 18.7. Acknowledgments
    23. 19. Multidimensional Iterators in NumPy
      1. 19.1. Key Challenges in N-Dimensional Array Operations
      2. 19.2. Memory Models for an N-Dimensional Array
      3. 19.3. NumPy Iterator Origins
      4. 19.4. Iterator Design
      5. 19.5. Iterator Interface
      6. 19.6. Iterator Use
      7. 19.7. Conclusion
    24. 20. A Highly Reliable Enterprise System for NASA's Mars Rover Mission
      1. 20.1. The Mission and the Collaborative Information Portal
      2. 20.2. Mission Needs
      3. 20.3. System Architecture
      4. 20.4. Case Study: The Streamer Service
      5. 20.5. Reliability
      6. 20.6. Robustness
      7. 20.7. Conclusion
    25. 21. ERP5: Designing for Maximum Adaptability
      1. 21.1. General Goals of ERP
      2. 21.2. ERP5
      3. 21.3. The Underlying Zope Platform
      4. 21.4. ERP5 Project Concepts
      5. 21.5. Coding the ERP5 Project
      6. 21.6. Conclusion
    26. 22. A Spoonful of Sewage
    27. 23. Distributed Programming with MapReduce
      1. 23.1. A Motivating Example
      2. 23.2. The MapReduce Programming Model
      3. 23.3. Other MapReduce Examples
      4. 23.4. A Distributed MapReduce Implementation
      5. 23.5. Extensions to the Model
      6. 23.6. Conclusion
      7. 23.7. Further Reading
      8. 23.8. Acknowledgments
      9. 23.9. Appendix: Word Count Solution
    28. 24. Beautiful Concurrency
      1. 24.1. A Simple Example: Bank Accounts
      2. 24.2. Software Transactional Memory
      3. 24.3. The Santa Claus Problem
      4. 24.4. Reflections on Haskell
      5. 24.5. Conclusion
      6. 24.6. Acknowledgments
    29. 25. Syntactic Abstraction: The syntax-case Expander
      1. 25.1. Brief Introduction to syntax-case
      2. 25.2. Expansion Algorithm
      3. 25.3. Example
      4. 25.4. Conclusion
    30. 26. Labor-Saving Architecture: An Object-Oriented Framework for Networked Software
      1. 26.1. Sample Application: Logging Service
      2. 26.2. Object-Oriented Design of the Logging Server Framework
      3. 26.3. Implementing Sequential Logging Servers
      4. 26.4. Implementing Concurrent Logging Servers
      5. 26.5. Conclusion
    31. 27. Integrating Business Partners the RESTful Way
      1. 27.1. Project Background
      2. 27.2. Exposing Services to External Clients
      3. 27.3. Routing the Service Using the Factory Pattern
      4. 27.4. Exchanging Data Using E-Business Protocols
      5. 27.5. Conclusion
    32. 28. Beautiful Debugging
      1. 28.1. Debugging a Debugger
      2. 28.2. A Systematic Process
      3. 28.3. A Search Problem
      4. 28.4. Finding the Failure Cause Automatically
      5. 28.5. Delta Debugging
      6. 28.6. Minimizing Input
      7. 28.7. Hunting the Defect
      8. 28.8. A Prototype Problem
      9. 28.9. Conclusion
      10. 28.10. Acknowledgments
      11. 28.11. Further Reading
    33. 29. Treating Code As an Essay
    34. 30. When a Button Is All That Connects You to the World
      1. 30.1. Basic Design Model
      2. 30.2. Input Interface
      3. 30.3. Efficiency of the User Interface
      4. 30.4. Download
      5. 30.5. Future Directions
    35. 31. Emacspeak: The Complete Audio Desktop
      1. 31.1. Producing Spoken Output
      2. 31.2. Speech-Enabling Emacs
      3. 31.3. Painless Access to Online Information
      4. 31.4. Summary
      5. 31.5. Acknowledgments
    36. 32. Code in Motion
      1. 32.1. On Being "Bookish"
      2. 32.2. Alike Looking Alike
      3. 32.3. The Perils of Indentation
      4. 32.4. Navigating Code
      5. 32.5. The Tools We Use
      6. 32.6. DiffMerge's Checkered Past
      7. 32.7. Conclusion
      8. 32.8. Acknowledgments
      9. 32.9. Further Reading
    37. 33. Writing Programs for "The Book"
      1. 33.1. The Nonroyal Road
      2. 33.2. Warning to Parenthophobes
      3. 33.3. Three in a Row
      4. 33.4. The Slippery Slope
      5. 33.5. The Triangle Inequality
      6. 33.6. Meandering On
      7. 33.7. "Duh!"—I Mean "Aha!"
      8. 33.8. Conclusion
      9. 33.9. Further Reading
    38. A. Afterword
    39. B. Contributors
    40. Colophon
    41. SPECIAL OFFER: Upgrade this ebook with O’Reilly
O'Reilly logo

Search in the Large

When most people think of search they think of web search, as offered by Yahoo!, Google, and their competitors. While ubiquitous web search is a new thing, the discipline of full-text search upon which it is based is not. Most of the seminal papers were written by Gerald Salton at Cornell as far back as the early 1960s. The basic techniques for indexing and searching large volumes of text have not changed dramatically since then. What has changed is how result ranking is done.[3]

Searching with Postings

The standard approach to full-text search is based on the notion of a posting, which is a small, fixed-size record. To build an index, you read all the documents and, for each word, create a posting that says word x appears in document y at position z. Then you sort all the words together, so that for each unique word you have a list of postings, each a pair of numbers consisting of a document ID and the text's offset in that document.

Because postings are small and fixed in size, and because you tend to have a huge number of them, a natural approach is to use binary search. I have no idea of the details of how Google or Yahoo! do things, but I'd be really unsurprised to hear that those tens of thousands of computers spend a whole lot of their time binary-searching big arrays of postings.

People who are knowledgeable about search shared a collective snicker a few years ago when the number of documents Google advertised as searching, after having been stuck at two billion and change for some years, suddenly became much larger and then kept growing. Presumably they had switched the document ID in all those postings from 32-bit to 64-bit numbers.

Ranking Results

Given a word, searching a list of postings to figure out which documents contain it is not rocket science. A little thought shows that combining the lists to do AND and OR queries and phrase search is also simple, conceptually at least. What's hard is sorting the result list so that the good results show up near the top. Computer science has a subdiscipline called Information Retrieval (IR for short) that focuses almost entirely on this problem. Historically, the results had been very poor, up until recently.

Searching the Web

Google and its competitors have been able to produce good results in the face of unimaginably huge data sets and populations of users. When I say "good," I mean that high-quality results appear near the top of the result list, and that the result list appears quickly.

The promotion of high-quality results is a result of many factors, the most notable of which is what Google calls PageRank, based largely on link counting: pages with lots of hyperlinks pointing at them are deemed to be more popular and thus, by popular vote, winners.

In practice, this seems to work well. A couple of interesting observations follow. First, until the rise of PageRank, the leaders in the search-engine space were offerings such as Yahoo! and DMoz, which worked by categorizing results; so, the evidence seems to suggest that it's more useful to know how popular something is than to know what it's about.

Second, PageRank is applicable only to document collections that are richly populated with links back and forth between the documents. At the moment, two document collections qualify: the World Wide Web and the corpus of peer-reviewed academic publications (which have applied PageRank-like methods for decades).

The ability of large search engines to scale up with the size of data and number of users has been impressive. It is based on the massive application of parallelism: attacking big problems with large numbers of small computers, rather than a few big ones. One of the nice things about postings is that each posting is independent of all the others, so they naturally lend themselves to parallel approaches.

For example, an index based on doing binary search in arrays of postings is fairly straight-forward to partition. In an index containing only English words, you could easily create 26 partitions (the term used in the industry is shards), one for words beginning with each letter. Then you can make as many copies as you need of each shard. Then, a huge volume of word-search queries can be farmed out across an arbitrarily large collection of cooperating search nodes.

This leaves the problem of combining search results for multiword or phrase searches, and this requires some real innovation, but it's easy to see how the basic word-search function could be parallelized.

This discussion is a little unfair in that it glosses over a huge number of important issues, notably including fighting the Internet miscreants who continually try to outsmart search-engine algorithms for commercial gain.



[3] This discussion of full-text search borrows heavily from my 2003 series, On Search, available online at http://www.tbray.org/ongoing/When/200x/2003/07/30/OnSearchTOC. The series covers the topic of search quite broadly, including issues of user experience, quality control, natural language processing, intelligence, internationalization, and so on.

The best content for your career. Discover unlimited learning on demand for around $1/day.