CHANGELOG.md

Splunk SDK for Ruby Changelog

Version 1.0.5

• Added support for basic authentication.
• Added an example for load balanced search heads.

Version 1.0.4

• Added support for client certificates and path prefix.
• Added an option to specify a proxy server to use. Includes connect_via_proxy.rb example.
• Minor Fixes

Version 1.0.3

• Splunk SDK for Ruby now works with Splunk 6.
• Fixed URL escaping of owners and apps in namespaces that contain special characters.

Version 1.0.2

• Cosmetic changes to the repository only.

Version 1.0.1

• Fixed Job#results to properly handle arguments.

Version 1.0

• No bugs found during beta period.

Version 0.8.1 (beta)

Bugs fixed

• Fixed wrong version number in a few documentation files.

Version 0.8.0 (beta)

Breaking changes

• The _raw field in events is now returned as text, not XML. That is, all tags such as the sg elements, are removed, and all characters are unescaped. The XML is available from the segmented_raw method on the event, which returns a string containing the raw XML of the _raw field returned by the server.
• The severities in messages in Atom feeds are now strings instead of symbols.

New features

• Added support for inputs via the Service#inputs method, and for modular input kinds via the Service#modular_input_kinds method.
• Added segmented_raw method to events returned by ResultsReader.

Bugs fixed

• Added missing Splunk:: prefix in example in the docs for Service.
• Moved default "segmentation=none" option for asynchronous searches from Job#initialize to Job#events, Job#preview, and Job#results.

Version 0.1.0 (preview)

Breaking changes

• The Splunk SDK for Ruby now uses the same .splunkrc format as the other SDKs.

• Moved Service#connect to being a static method of the Splunk module, so you now call connect as

  service = Splunk::connect(...)
  

  instead of

  service = Splunk::Service.connect(...)
  

• aloader.rb and the AtomResponseLoader it contained have been replaced in their entirety by atomfeed.rb and the AtomFeed class it contains.

• Changes to Context:

  • Removed Context#post, Context#delete, and Context#get. Replaced them all with Context#request (which takes a structured specification of the resource) and Context#request_by_url (which takes an already constructed URL to the resource).
  • Removed the methods Context#fullpath, Context#init_default, Context#url, Context#create_resource, Context#check_for_error_return, Context#build_stream, Context#handle_key, Context#flatten_params, and Context#flatten_params_array since simplification of the request infrastructure made them unnecessary.
  • Renamed Context#protocol to Context#scheme to match correct naming in RFC.

• Removed the parse method from Service, since it didn't actually parse the response in any useful way.

• Changes to naming of the configuration endpoints:

  • ConfCollection renamed to ConfigurationFile (and conf_collection.rb renamed to configuration_file.rb).
  • Conf renamed to Stanza (and conf.rb renamed to stanza.rb).

• Index#submit and Index#attach now take hash arguments instead of positional arguments.

• Job#setttl renamed to Job#set_ttl.

• Job#setpriority renamed to Job#set_priority.

• Job#disable_preview removed, since a working test scenario couldn't be found for it.

• Collection's constructor arguments have changed.

• Entity's constructor arguments have changed.

• SplunkError has been removed, and SplunkHTTPError has been made a direct subclass of StandardError.

• Deleted SearchResults, since it was superseded by ResultsReader.

• The metaprogramming to allow fetching entries of a Hash via dot notation has been removed. That is, hash.key is no longer a synonym for hash["key"].

• The urlencode method previously monkeypatched onto Hash has been removed.

• Service#create_collection was removed.

New features and APIs

• New examples in the /examples directory, showing how to:
  • connect to Splunk
  • manage entities in Splunk
  • run searches and fetch their results from Splunk
  • write data into Splunk
• The whole SDK now handles namespaces properly, and namespaces are first class objects. See the header comments of namespace.rb for a detailed description of Splunk namespaces and their representation in this SDK.
• All XML handling is shimmed to work with either Nokogiri or REXML. It tries Nokogiri first, which can process XML vastly faster, and falls back to REXML, which will be present in all Ruby installs. The user may all call require_xml_library to force the SDK to use a Nokogiri or REXML. See xml_shim.rb for all this behavior.
• Added documentation throughout the SDK.
• The unit testing suite has been completely rewritten and expanded.
• Added a ResultsReader class to parse and iterate over sets of XML results from search jobs.
• Added methods to Context:
  • host
  • port
  • token
  • username
  • password
  • namespace
  • server_accepting_connections?
• Context#restart takes an optional timeout, and properly handles waiting for Splunk to restart.
• Added convenience methods to Service to create searches:
  • create_oneshot
  • create_search
  • create_stream
• Added methods to Service:
  • splunk_version
  • saved_searches
• New features added to Collection:
  • Collection#each, when called without a block, now returns an enumerator over all elements in the collection.
  • Added an optional namespace argument to Collection#delete and Collection#fetch so entities can be specified uniquely even in the presence of name collisions.
  • Added new methods to Collection, mostly designed to recapitulate the relevant methods of Hash. In the notation below, synonym sets are separated by slashes, and names in a set of synonyms that already existed are surrounded by square brackets:
    • [contains?]/has_key?/include?/key?/member?
    • [list]/values/to_a
    • each_key
    • each_pair
    • empty?
    • entity_class
    • fetch/[ [] ]
    • keys
    • length
    • resource
    • service
• Added methods to Entity. In the notation below, synonym sets are separated by slashes, and names in a set of synonyms that already existed are surrounded by square brackets:
  • content
  • delete
  • fetch/[ [] ]
  • links
  • namespace
  • refresh
  • resource
  • service
• Entities now cache their state. To update the cache, call the entity's refresh method.
• Index#clean takes a timeout instead of waiting forever. If no timeout is specified, a default value will be used.
• Added new methods to Job. In the notation below, synonym sets are separated by slashes, and names in a set of synonyms that already existed are surrounded by square brackets:
  • control
  • is_done?
  • is_ready?
  • sid
• Added saved search support, accessible via Service#saved_searches. SavedSearch entities have the following additional methods beyond Entity:
  • dispatch
  • history
• Added the following methods to Stanza (was: Conf). In the notation below, synonym sets are separated by slashes, and names in a set of synonyms that already existed are surrounded by square brackets:
  • length
  • update/[submit]
• Added new custom exceptions:
  • AmbiguousEntityReference, for when a fetch of an entity by name would return multiple entities due to name collisions.
  • EntityNotReady, raised when a program tries to fetch the state of an entity which is queued for creation, but not yet created, such as a search job that is not yet ready.
  • IllegalOperation, raised by a program when an operation known statically not to work, such as deleting an index on Splunk 4.x, is called.

Architectural changes

• Job is now a subclass of Entity.
• Service is now a subclass of Context instead of including the Context as an instance variable.
• Removed dependence of Context (and the SDK) on the libraries netrc, pathname, and rest-client. Now it uses only the standard library plus (optionally) Nokogiri.
• Reorganized directory structure of the SDK.
  • The contents of lib/splunk-sdk-ruby/client/ have been moved to lib/splunk-sdk-ruby, and client.rb has been deleted.
  • jobs.rb was moved to lib/splunk-sdk-ruby/collection/
  • job.rb was moved to lib/splunk-sdk-ruby/entity/
• Made all unit test filenames start with test_ instead of tc_.