documentation
.HELP!
Bazaar Documentation
Miscellaneous notes
Bazaar
previous page
next page
Navigation
next
previous
|
Home
|
Documentation
|
Developer Document Catalog (2.2b1)
»
Miscellaneous notes
dirstate
— An observation re. the dirstate file
“bzr update” performance analysis
— “bzr update” performance analysis
Navigation
next
previous
|
Home
|
Documentation
|
Developer Document Catalog (2.2b1)
»
previous page
start
next page
Menu
Homepage
Table of contents
Developer Document Catalog (2.2b1)
Contributing to Bazaar
Talk to us
Starting
Making a branch
Writing tests
Proposing a merge
Bazaar Release Cycles
The Process
Schedule
Regularity
Numbering
Terminology
Bug Work
Feature and Performance Work
Plugins
Data and Network Formats
Hosting Services
Simultaneous Installation
OS Distributions
Packaging
Code Freeze vs Announcement
Weekly Metronome Mail
Questions
Expected Benefits
Reviewing for the Stable Branch
References
Profiling
Using profilers
Profiling locks
Profiling HPSS Requests
Tracking Bugs in Bazaar
Links
Generalities
Priorities
Clear Bugs
Bug Status
Bug Importance
Assignment
Targetting Bugs
Backports
The News File
Tags
Bazaar Developer Guide
Getting Started
Exploring the Bazaar Platform
Finding Something To Do
Planning and Discussing Changes
Bazaar Development in a Nutshell
Making the change
Making a Merge Proposal
Why make a local copy of bzr.dev?
Understanding the Development Process
Preparing a Sandbox for Making Changes to Bazaar
Navigating the Code Base
The Code Review Process
Review cover letters
Reviewing proposed changes
Checklist for reviewers
Reviews on Launchpad
Proposing a merge through the web
Proposing a merge by mail
Reviewing changes
Reviews through Bundle Buggy
Sending patches for review
Coding Style Guidelines
hasattr and getattr
Code layout
Module Imports
Naming
Standard Names
Destructors
Cleanup methods
Factories
Registries
InterObject and multiple dispatch
Lazy Imports
The Null revision
Object string representations
Exception handling
Test coverage
Core Topics
Evolving Interfaces
Deprecation decorators
Getting Input
Processing Command Lines
Standard Parameter Types
Writing Output
Progress and Activity Indications
Displaying help
Handling Errors and Exceptions
Assertions
Documenting Changes
NEWS File
Commands
API Documentation
General Guidelines
Copyright
Miscellaneous Topics
Debugging
Debug Flags
Jargon
Unicode and Encoding Support
Command.outf
bzrlib.urlutils.unescape_for_display
Portability Tips
C Extension Modules
Making Installers for OS Windows
Core Developer Tasks
Overview
What is a Core Developer?
Communicating and Coordinating
Submitting Changes
An Overview of PQM
Selecting a Public Branch Location
Configuring the PQM Plug-In
Submitting a Change
Tracking Change Acceptance
Reviewing Blueprints
Blueprint Tracking Using Launchpad
Recording Blueprint Review Feedback
Planning Releases
Using Releases and Milestones in Launchpad
Bug Triage
Bazaar Testing Guide
The Importance of Testing
Running the Test Suite
Running particular tests
Disabling plugins
Disabling crash reporting
Test suite debug flags
Using subunit
Using testrepository
Writing Tests
Where should I put a new test?
Unit tests
Blackbox (UI) tests
Per-implementation tests
Doctests
Shell-like tests
Import tariff tests
Testing locking behaviour
Skipping tests
Test feature dependencies
Writing tests that require a feature
Defining a new feature that tests can require
Testing exceptions and errors
Testing warnings
Interface implementation testing and test scenarios
Test scenarios
Test support
TestCase and its subclasses
BranchBuilder
TreeBuilder
Releasing Bazaar
Preconditions
Starting a cycle
Starting the release phase
Making the source tarball
Publishing the source tarball
Announcing the source freeze
Publishing the release
Announcing the release
Merging the released code back to trunk
Releases until the final one
See also
Managing the Bazaar PPA
Background
Preconditions
Packaging Bazaar
Short form
Long Form
Packaging bzr-svn
Monitoring the contents of PPAs
Bazaar Windows EC2 Server
Goals
Approach
Procedures
Preparation
To start up an instance
To save a system snapshot as an image
Management console (useful!)
Bazaar Architectural Overview
Essential Domain Classes
Transport
Filenames vs URLs
Repository
Stacked Repositories
Integrating with Bazaar
Manipulating the Working Tree
Compare trees
Adding Files
Removing Files
Renaming a File
Moving Files
Committing Changes
Generating a Log for a File
Annotating a File
Working with branches
Branching from an existing branch
Pushing and pulling branches
Checkout from an existing branch
History Operations
Finding the last revision number or id
Getting the list of revision ids that make up a branch
Getting a Revision object from a revision id
Accessing the files from a revision
Bazaar Design Principles
Testing
Data formats
Plans
1 Bazaar Performance Roadmap
1.1 About the performance roadmap
1.1.1 What should be in the roadmap?
1.1.2 What should the final system look like, how is it different to what we have today?
1.1.3 What use cases should be covered?
1.1.4 How is development on the roadmap coordinated?
1.1.5 Planned changes to the bzr core
1.1.5.1 Library changes
1.1.5.2 Interoperable disk changes
1.1.5.3 Possibly non-interoperable disk changes
1.1.5.4 Non-interoperable disk changes
1.1.6 Integration of performance changes
1.2 Analysis of use cases
1.2.1 Analysing a specific use case
1.2.2 Performing the analysis
1.2.3 What factors should be considered?
1.3 Use cases
1.3.1 Initial push / pull
1.3.1.1 Optimal case
1.3.1.2 Disk case
1.3.1.3 Smart Network Case
1.3.1.3.1 Phase 1
1.3.1.3.2 Phase 2
1.3.1.4 Dumb Network Case
1.3.1.5 Wants
1.3.2 Incremental push/pull
1.3.2.1 Functional Requirements
1.3.2.2 Factors which should add work for push/pull
1.3.2.3 Push/pull overview
1.3.2.3.1 New data identification
1.3.2.3.1.1 Set synchronisation approaches
1.3.2.3.1.2 DAG synchronisation approaches
1.3.2.3.1.3 File level scaling
1.3.2.3.1.4 API scaling
1.3.2.3.2 Data reading
1.3.2.3.2.1 File level scaling
1.3.2.3.2.2 API scaling
1.3.2.3.3 Data Verification and writing
1.3.2.3.3.1 Overview summary
1.3.2.3.3.2 File level scaling
1.3.2.3.3.3 API scaling
1.3.2.4 Notes from London
1.3.3 Add
1.3.3.1 Least work we can hope to perform
1.3.3.2 Per file algorithm
1.3.4 Commit Performance Notes
1.3.4.1 Changes to commit
1.3.4.2 Commit: The Minimum Work Required
1.3.4.3 Commit vs Status
1.3.4.4 Avoiding Work: Smarter Change Detection
1.3.4.5 Avoiding Work: Better Layering
1.3.4.6 Avoiding work: avoiding reading parent data
1.3.4.7 Code structure
1.3.4.8 Complications of commit
1.3.4.9 Interface stack
1.3.4.10 Branch->Tree interface
1.3.4.11 Information from the tree to repository
1.3.4.12 Information from the repository to the tree
1.3.4.13 Selective commit
1.3.4.14 Common commit code
1.3.4.15 Order of traversal
1.3.4.16 Open question: per-file graphs
1.3.5 diff Performance Analysis
1.3.5.1 Minimal Work
1.3.5.1.1 Reuse of historical comparisons
1.3.5.1.2 Historical Tree Against Historical Tree
1.3.5.1.3 Basis Against Historical Tree
1.3.5.1.4 Basis Against Basis
1.3.5.1.5 Working Tree Against Basis
1.3.5.1.6 Working Tree Against Historical Tree
1.3.5.1.7 Working Tree Against Working Tree
1.3.5.2 API Changes
1.3.5.3 Storage considerations
1.3.6 Garbage Collection
1.3.6.1 Least work we can hope to perform
1.3.7 Revert
1.3.7.1 Least work we can hope to perform
1.3.8 The status command
1.3.8.1 UI Overview
1.3.8.2 Ideal work for working tree to historical status
1.3.8.3 Locality of reference
1.3.8.4 Scaling observations
1.3.9 Annotate
1.3.10 Scaling analysys of Merge
1.3.10.1 Needs
1.3.10.2 Notes
1.3.11 Bundle Creation
1.3.11.1 Needs
1.3.12 Uncommit Performance Notes
1.3.12.1 Specification of uncommit
1.3.13 Missing
1.4 Subsystem designs
1.4.1 Directory fingerprints
1.4.1.1 Introduction
1.4.1.2 Use-case oriented APIs
1.4.1.2.1 commit
1.4.1.2.2 log
1.4.1.3 Open questions
1.4.1.4 Conclusions
1.4.1.5 Design changes
1.4.1.6 API changes
co-located branches
Rationale
Use Cases
Implementation
UI Changes
Code Changes
Schema Changes
Unresolved Issues
Bazaar Windows Shell Extension Options
Introduction
Background Information
The facts about shell extensions
Analysis of TortoiseSVN code
Analysis of existing TortoiseBzr code
Detailed Implementation Strategy
External Command Processor
Performance considerations
RPC options
Vista versus XP
Reuse of TSVNCache?
Reuse of this code by Mercurial or other Python based VCS systems?
Implementation plan
Alternative Implementation Strategies
Implement Completely in Python
CHK Optimized index
Required information
Partial Hash
Scaling Sizes
Scaling up
Scaling down
Other discussion
group encoding
variable length index entries
storing uncompressed offset + length
null content
VF.keys()
More than 64k groups
Specifications
Revision Properties
Usage
Well-known properties
API Versioning
Status
Motivation
Terminology
API versions
Managing API versions
Exported API's
Use Cases
Requiring bzrlib 0.18 in a plugin
Exporting an API from a plugin
Bazaar Apport Integration
Consequences for users
Implementation notes
Future ideas
Authentication ring
Rationale
Authentication definitions
File format
Use Cases
UI Changes
Implementation constraints
Questions and Answers
Bundles
Status
Motivation
Desired features
Historical Design
June 2007 Design
Future Plans
Physical encoding
Code reuse
Container format
Status
Motivation
Terminology
Use Cases
Streaming data between a smart server and client
Incremental push or pull
Persistent storage on disk
Usable before deep model changes to Bazaar
Examples of possible record content
Characteristics
No length-prefixing of entire container
Structured as a self-contained series of records
Addressing records
Reasonably cheap for small records
Specification
Record types
End Marker
Bytes
Names
Overview
Current approach
Things to keep
Things to remove
Thoughts
Delta strategy
Implementation Strategy
Indices
Status
Motivation
Terminology
Overview
General Index API
Services
Build index
Retrieve entries from the index
Merging of indices
Index implementations
GraphIndex
Inventories
Overview
In memory inventories
Serialization
dirstate
xml
Serialization scaling and future designs
Current situation
Long term work
Layering
Design elements to achieve the goals in a future inventory implementation
Hash bucket based inventories
Overview
Goal satisfaction
Issues
Canonical form
Apply
Delta
Radix tree based inventories
Overview
Goal satisfaction
Issues
Canonical form
Apply
Delta
Hash Trie details
Insertion
Inventory deltas
Delta consistency
Avoiding inconsistent deltas
LCA Merge
Essential characteristics
Algorithm
LCA merge and Three-way merge
Exception to three-way behavior
Why a new name
Performance
Possible flaws
Acknowledgements
Network Protocol
Overview
Layering
Medium
Protocol
Request/Response processing
Server-side
Client-side
Protocol description
Version one
Version two
Version two with streamed bodies
Version three
Headers
Conventional requests and responses
Early error returns
Full-duplex operation
APIs
Paths
Requests
Recognised errors
Plugin API
Introduction
See also
Structure of a plugin
Plugin metadata before installation
Metadata protocol
Control Formats
Example
Plugin metadata after installation
Help and documentation
API version
Plugin version
Detecting whether code's being loaded as a plugin
Plugin performance
Plugin registrations
Publishing your plugin
Repositories
Status
Motivation
Terminology
Command Requirements
Data access patterns
Patterns used
Facilities to scale well
Indices
Index size
Index ordering
Changing our current indexes
Replace .kndx
Data
Moving to pack based repositories
Naming of files
Discovery of files
Housing files
Combining indices on demand
Merging data on push
Choosing compression/delta support
Caching and writeing of data
Locks
Write Groups
Repository Streams
Status
Motivation
Use Cases
Fetch operations
Smart server operations
Bundles
Data conversion
Characteristics
Single round trip
Forward-only reads
Serialisation
Weaves
Bundles
Specification
Requesting a stream
Structure of a stream
Consuming a stream
factory metadata
Case Insensitive File Systems
Handling these file-systems
Implementation of support for these file-systems
Development repository formats
Using development repository formats
Motivation
Format names
Support for upgrade and migration
Before converting to a development format
Creating a new development format branch
Creating a new development format repository
Upgrading an existing branch or repository to development
Starting a new development format branch from one in an older format
Develoment formats for bzr-svn users
Reporting problems
Technical notes
When to create a new development format
How to create a new development format
Format Details
development
development-rich-root
development-subtree
Development6RichRoot[Subtree]
KnitPack repository format
Using KnitPack repositories
Motivation
Preparation
Creating a new knitpack branch
Creating a new knitpack repository
Upgrading an existing branch or repository to knitpack format
Starting a new knitpack branch from one in an older format
Testing packs for bzr-svn users
Reporting problems
Technical notes
Implementation notes
BTree Index Prefetch
Rationale
Example
BTree Structure
Example 1 layer
Example 2 layer
Example 3 layer
Current Info
Data and Request
Thoughts on expansion
Algorithm
Computing last_modified values
Introduction
Definition
Generation in commit
Generation during conversion
Avoiding last_modified
Compared to annotate
Use cases
Cases to test
Performance considerations
Open questions
Possible structure
Content Filtering
Filters
Conventions
Dirstate interactions
User interface
Performance considerations
Future ideas and open issues
See also
LCA Tree Merging
Scalars
Definition
Resolution Algorithm
InventoryEntry.revision
Miscellaneous notes
Dirstate
"bzr update" performance analysis
No new revisions
No local changes, only new revisions
out of date heavyweight checkout, out of date w/master
Get in touch
Submit feedback about this site to:
[email protected]
Home