Django 1.5 + More

• Add support for Django 1.5
• Increase minimum version of ModelTree 1.1.5
• Add a Haystack search_sites module for convenience when integrating in a project
• Remove default search results size of 10
  • This is arbitrary and has caused confusion that it was truncating results
• Augment the language text directly to DataContext JSON attributes
  • This reduces the overhead for clients to parse the language dict separately
• Modify context validation to disable invalid nodes
  • This takes a more passive approach to validating context nodes, but prevents the context from getting in an unusable state. Validation errors result in the node being disabled.
• Add template and default fields to DataContext and DataView
  • This enables the use case of differentiating template contexts and views vs. user-defined ones. The default flag enables marking a template as the default one to be used so there is a starting point for clients.
• The dict representing the JSON data can be passed into the DataContext and DataView constructors directly for convenience
  • Other field values can still be defined using keyword arguments DataContext({...}, name='My Context')
• Modify verbose name for 'in' to be 'is either X, Y or Z' and the negated form to 'is neither X, Y, nor Z'
• Change container-based operators to use the Exact and NotExact text for single value lists

Backport

• Backport context node schema from 2.1
  • This is a backwards compatible port of the context node changes in Avocado 2.1 specifically to:
    • Support the field key instead of id as the field identifier
    • Support the concept key which is used to scope the field
    • Support branches without children or one child to act as containers

Bugs

• Fix possible mis-ordering of formatter keys and fields

Transparent DISTINCT/ORDER BY

• Add implementation (and fix) for handling the SELECT DISTINCT/ORDER BY behavior in databases more transparently

2.0.18

• Fix #67, change ObjectSet created and modified to not be auto-updated
• Increase length of DataField app_name, model_name, and field_name
  • For extra long field names...
• Change avocado init to not prepend the model name to the field name
• Fix and ensure unicode throughout
• Fix the admin action for creating a single DataConcept from multiple fields when only one field is selected

Refactor

• Fix performance in Formatter class due to redundant logging
• Add support for Decimal types in Formatter.to_number method

Bugs

• Update ModelTree to version 1.1.3 (critical bug fix)
• Add receiver for change Avocado settings during test execution
• Add missing DataField.coded_values which the Formatter.to_coded relied on

Metadata Migration

• Add backup utilities for performing metadata data migrations
• New command avocado migration for creating a metadata fixture and a corresponding South migration to load the fixture
• Fix bug in R and SAS exporters
• Improve ObjectSet class to enable deleting set objects rather than just flagging as being deleted
  • Pass the flag delete=True when calling a remove-based command, e.g. foo.replace(objs, delete=True)

Bugs

• Fix bug that only checked for NumPy for the for SciPy feature

2.0.13

• Change HTMLFormatter to require the template argument which may be a template name
  or a Template object.
• Allow arbitrary *args and **kwargs to be passed into *Exporter.write and
  *Exporter.read to enable propagation from write to read
• Add short_name and long_name for use downstream by clients
• Add export-specific format as the first preferred format
  • This is a better default which enables specific formatting when needed. The more general format can be reused across formatters, but when very specific formatting is necessary, this default is preferred.
• Refactor DataField.model and DataField.field to take in account Lexicon and ObjectSet models
  • The real field and model instances are now named real_field and real_model, respectively.
• Fix #55, Lexicon.label is now correctly used by the DataView
• Rename sync subcommand to init
• Rename SYNC_ENUMERABLE_MAXIMUM to ENUMERABLE_MAXIMUM
• Remove orphaned command in favor of new check that performs multiple setup
  checks as well as check for invalid datafields.
• Remove searchable model field since this applies only to text-based fields.
  • This has been repurposed as a deprecated computed property
• Fix #45, allow dict-based settings to be updated, but not overridden
• Remove DataField.data_source field
  • There was no functional utility of this field and is (currently) out of the scope for Avocado