-
Notifications
You must be signed in to change notification settings - Fork 668
Change Log
Anwesha edited this page Dec 11, 2017
·
56 revisions
- Added support for customization_weight or acoustic_customization_id in speech-to-text/v1
2017-11-29
This version contains several breaking changes.
Version 3.0 introduces automatically generated client code. The client code is generated from Swagger. Several deprecated services have been removed from this release. Furthermore, language has been changed from NodeJS to TypeScript, and file extensions have been changed from *.js
to *.ts
Services that are not affected in this release:
- Authorization
- Dialog
Services that are affected in this release:
- Conversation
- Discovery
- Language Translator
- Natural Language Classifier
- Natural Language Understanding
- Personality Insights
- Text to Speech
- Tone Analyzer
- Speech to Text
- Visual Recognition
Services that are removed in this release:
- Alchemy Language
- Alchemy Data News
- Document Conversion
- Retrieve and Rank
- Tradeoff Analytics
Upon upgrading the SDK, follow the migration guide here.
- Added support for
highlight
param inDiscoveryV1.query()
- DiscoveryV1.VERSION_DATE_2017_08_01 = '2017-08-01';
-
ToneAnalyzerV3.tone()
now acceptslanguage
parameter to specifycontent-language
header for input.
- Improved formatting of Alchemy Language error messages
- Added
ConversationV1.VERSION_DATE_2017_05_26
- Added
DiscoveryV1.updateJsonDocument
to complimentaddJsonDocument()
- Added
DiscoveryV1.addJsonDocument()
method to make uploading in-memory JSON structures easier - Fixed bug in
DiscoveryV1.addDocument()
that prevented setting custom filenames #474
- Added support for
fuzzy_match
param in Conversation entities methods - Added support for JSON metadata when adding documents to Discovery #474
- Discovery: fix createEnv with size 0
- Add natural_language_query and passages parameters to Discovery.query()
- Fix collection name param for Discovery.createCollection (#https://github.com/watson-developer-cloud/node-sdk/issues/457)
- Bumped
solr-client
dependency to latest version
- Added VisualRecognitionV3.VERSION_DATE_2016_05_20 constant
- Added support for sort param on DiscoveryV1.query() (#446)
- Added DiscoveryV1.VERSION_DATE_2017_04_27
- Made utterances a top-level param for ToneAnalyzerV3.tone_chat() (#439)
- Added support for Conversation entities (w/ values and synonyms) and logs
- Added support for updating Discovery environments
- Added tone_chat endpoint for Tone Analyzer
- Added support for WebM and Ogg/Vorbis formats to Speech to Text
- Fixed Visual Recognition credentials in dedicated environments (#436)
- bumped vcap_services library version
- Fixed bug with STT async recognition & multiple callback events
- Moved changelog to wiki
- Fixed issue when creating STT Async recognize job with multiple callback events (#415)
- Add support for Conversation Counter Examples API
- Correct name of
events
parameter in STT async recognize.
- Conversation sorting and pagination for workspaces, intents, and examples
- Natural Language Understanding: fixed credentials pulling from bluemix
- Natural Language Understanding: new version_date and addition of listModels() and deleteModel() methods
- STT RecognizeStream now exposes Transaction ID
- Restored support for Node.js 4.0-4.4
- Added support for Conversation intents and examples
- Speech to Text createRecognitionJob() now accepts all params from .recognize()
- Speech to Text getRecognitionJobs() accepts an optional params object in order to match the signature of the rest of the API
- Make callback_url optional for Speech to Text createRecognitionJob()
- Speech to Text Asychronous API support
- Added support for sort paramater in SpeechToTextV1.getWords()
- Added updateDocument() method to DiscoveryV1
- Fixed up internal code to avoid using the deprecated
new Buffer(...)
API - Improved documentation
- Conversation VersionDate 2017-02-03
- Added support for find_preferable_options flag in Tradeoff Analytics
- Added various methods to Conversation service to create and manage workspaces
- Prevent docs site files from being included in npm releases
- Fixed bug in STT getWords method (#390)
- No API changes, but significant internal changes. Should behave exactly the same, but releasing as a standalone version out of caution.
- Added NaturalLanguageUnderstandingV1
- Added support for pulling SERVICE_NAME_URL from enviroment properties along with username and password (or api key)
- stt.whenCustomizationReady() no longer incorrectly requires that a corpus be added. (#382)
- various JSDoc corrections
- Fix slightly-incorrect URL in Language Translator V2 example and error message
- SDK now emits missing parameter errors on returned stream if no callback is supplied (#368 / #377)
- Fix DocumentConversion#convert() to accept config params as documented in api ref
- Fix param checking on various TTS customization methods
- Fix incorrect error messages (#373)
- Fix issue where adding a document as buffer/string fails in Discovery v1 (#370)
- Fix issue where STT RecognizeStream could fail to emit speaker_labels event in rare circumstances
- Update jsdoc for Speech to text
- Expose discovery v1 in index
- Same fix for language translation
- Fixed DocumentConversionV1.convert() to accept config params as a seperate object, matching api ref documentation (#375)
- Fix regression in error handler that would incorrectly overwrite error message in some cases
- Fix isue where STT RecognizeStream could fail to emit a speaker_labels events in certain rare circumstances
- Added Discovery add createEnvironment and deleteEnvironment methods
- Fix Discovery addDocument when adding a Buffer
- Update JSDOc for speech to Text
- Expose discovery v1 in index.js
- Fix content type for language translation (#362)
- Fix content type for language translator (#362)
- Added Discovery V1 (general availability release)
- Added STT whenCorporaAnalyzed() helper, #353
- Bumped dependencies
- Added Speech to Text speaker_labels support, added new speaker_labels event to RecognizeStream
- Fixed main file path in package.json
- added a conversation example
- Added several methods and a version_date constant to DiscoveryV1Experimental
- Added version_date constants to ConversationV1, updated docs, examples, etc to latest version_date.
- Added customization_id support to STT RecognizeStream
- Allow VisualRecognitionV3.classify() to accept Buffers (with automatic content-type detection) or Objects with specified filename or content-type
- Improve Visual Recognition error formatting
- Document Conversion: Remove forced utf-8 charset header for html files
- Fixed issue with formatting multiple keywords for STT recognize() - #261
- Added
getCredentials()
method to base service, primarily for use with AuthorizationV1()
- dependency fix
- Speech to Text customization support
- Ensure errors are always instanceof Error
- Doc improvements
- Added support for Personality Insights V3
- Split LanguageTranslationV2 into a seperate service from LanguageTranslatorV2, added a warning when translator service is used without specifying a url since the default is currently incorrect.
- Fixed bug with setting user-agent header
- Fixed issue with send data on non-open WebSockets for STT RecognizeStream - see #322
- Fixed credential bugs in R&R and STT
- Fixed bug where credentials could be lost when calling certain methods in Language Translator and Dialog services
- Doc-only changes, focused on Language Translation/Translator differences and temporary workaround
- Fix issue with automatically loading Alchemy* credentials from the Bluemix environment
- Revert part of credential change: Alchemy* services again use
apikey
while visual recognition usesapi_key
(Most endpoints accept either, but this split follows the documented API for each service.)
- Added support for Visual Recognition similarity search beta
- Refactored handling of credentials to support constraints in similarity search
- Add support for RetrieveAndRankV1.rank() answers param
- Added support for creating and managing TTS Customizations
- Deprecation warning for Dialog
- Added TTS .voice() and .pronunciation() methods
- Added support for
customization_id
to existing TTS methods - Typo and JSDoc fixes
- Fixed bug with pulling api key from env properties for alchemy services
- Added profanity_filter support to STT RecognizeStream
- Corrected some examples to use - instead of _
- Renamed NLC folder to use - instead of _
- Updated AlchemyLanguageV1.emotion() to support new targeted_emotion parameter & endpoint
- Added Conversation/Tone Analyzer integration example
- Updated Readme & Examples to use newer constructor style
- Use actual Error instances for errors (#298)
- Added support for the
intents
,entities
andoutput
parameters in ConveersationV1.message() - Removed sunset services: Concept Insights and Relationship Extraction
- Dependency bump
- Added VisualRecognitionV3.retrainClassifier() to facilitate updating of existing custom classifiers
- Added support for
alternate_intents
parameter in ConversationV1.message()
- Breaking: prefer programatic (user-supplied) credentials over bluemix-provided ones (VCAP_SERVICES)
- New preferred method of instantiating services:
new watson.PersonalityInsightsV2({/*...*/});
instead ofwatson.personality_insights({/*...*/});
. Older method still works - Restructured code to support client-side usage via tools such as Browserify and Webpack. Most services support CORS; this will be documented and the remaining service teams will be nagged.
- Added a changelog to capture both major breaking changes and smaller
Several breaking changes were introduced with the v1.0.0 release:
- Experimental and Beta services now require the appropriate tag to be added to their version:
- Concept Expansion
v1
is nowv1-beta
- Question and Answer
v1
is nowv1-beta
- Relationship Extraction
v1
is nowv1-beta
- Tone Analyzer
v3
is nowv3
(latest) orv3-beta
(compatibility with old Beta plan) - Visual Insights
v1
is nowv1-experimental
- Visual Recognition
v1
is nowv1-beta
- Concept Expansion
- Speech to Text gained a new
createRecognizeStream()
method replacing the existing live streaming methods with a simpler Read/Write stream. The older methods are still available in v1.0 but each log a deprecation warning (unless{silent: true}
is passed in) and will be removed from a future release. The affected methods are:recognizeLive()
observeResult()
getRecognizeStatus()
- The Document Conversion API has been reduced to a single
convert()
method; it no longer offers batch conversion or cloud storage of files. - Several deprecated services have been removed:
- Message Resonance
- Tone Analyzer v1 and v2 (replaced by v3)
- Search (replaced by Retrieve and Rank)
- Dropped support for node.js v0.10.x (For reference: the WDC Node.js SDK now officially support the latest 0.12, LTS, and Stable releases of Node.js.)