tigerconnect - npm Package Versions

[13.7.0] - 2021-03-18

Fixes

• groups.find(): Errors like PermissionDeniedError and NotFoundError will now be properly thrown as different kinds of errors instead of always being NotFoundError.

[12.10.0] - 2021-01-05

Performance Improvements

• conversations.findAll(): The conversation target and group members will no longer be downloaded for every conversation. This can greatly improve performance for users with large rosters.

[12.6.0] - 2020-11-30

General Changes

• New client.calls service for VoIP Calling. Demo app available here. Documentation to be added soon.
• New function messages.downloadAttachmentBlobAndUrl to efficiently retrieve both the attachment blob and its URL.

Fixes

• Fix build issue that was causing the web bundle to not load.
• Fix an issue where the global instance of lodash was being modified after loading the Web bundle.
• Fix an issue where conversations.fetchTimeline() and conversations.selectConversation() were causing the timeline of previously downloaded conversations to be cleared unnecessarily.
• Miscellaneous optimizations and improvements.

[11.7.0] - 2020-09-18

General Changes

• Improved efficiency when creating groups with metadata using the SDK.
• Improved responsiveness of conversations.findAll() when CStore is on and the user has a large roster with more than 200 items.
• messages.findAttachment(): If an attachment cannot be located, return null instead of throwing an error.

Fixes

• Fix an issue where a group that was created with metadata did not surface the metadata right away.

[10.8.0] - 2020-06-18

General Changes

• The User model now has a removedFromOrg field, that if set to true, indicates that the user has been removed from all organizations.
• Removed users whose names are not known will be given a displayName of Removed User.

Fixes

• Fix an issue where conversation.messages would not be updated after a message was recalled or expired. As a sidenote: for most use-cases, we recommend using conversation.timeline instead, since its contents will be sorted, up-to-date on recipient status flags, and will better reflect the content fetched during client.conversations.fetchTimeline() calls.
• Fix an issue where conversation.lastMessage was sometimes not set to the most up-to-date value.
• Improve performance of client.conversations.clearMessageStatuses, which is especially noticeable when viewing conversations for groups that have a very large number of messages.
• Various other bugfixes and performance improvements.

[10.3.0] - 2020-05-13

General Changes

• The generic client.messages.forward() function has been deprecated in favor of forwardToUser(), forwardToGroup(), etc. and will throw an error if an attempt is made to use it
• The generic client.messages.send() function has been deprecated in favor of sendToUser(), sendToGroup(), etc. and will throw an error if an attempt is made to use it
• Various bugfixes and performance improvements

Fixes

• Fix issue where sending a message immediately after receiving other messages could cause all subsequent messages to be missing for up to 7 minutes
• In conversations.fetchTimeline() and conversations.selectConversation():
  • Fix issue where the conversation content was being fetched more than what was strictly necessary
  • NOTE: If you would like to display a loading indicator in a particular direction if more content is available, please read the explanation of the itemsEstimate continuation field in the introduction of the fetchTimeline docs for guidance on how to implement that correctly.

[8.4.1] - 2020-01-31

General Changes

• typing:change
  • Change contract to include contextId and contextType parameters

Fixes

• In conversations.fetchTimeline() and conversations.selectConversation():
  • Fix issue where the user joins a group and then receives more than a certain number unread bangs and messages, when they next select the group, sometimes not all messages will be fetched, if the first fetched item happens to be a bang.
  • Fix issue where the user enters a forum or a group with replayHistory = true, reads all of the messages, leaves, and then rejoins, then only the very top (lowest sort_number) of the conversation will be fetched.
• conversations.markAsRead():
  • Messages will now consistently get marked as Read.
• conversations.find():
  • Allow this to work without a preceding conversations.findAll()
• messages.sendToUser():
  • Fix an issue where sending a message to a new user resulted in an incorrect conversation.lastMessage

[8.0.2] - 2019-12-04

General Changes

• The contract for typingStatus.startTyping() has changed, and it is now required to pass organizationId as part of the 2nd options parameter. See https://tigerconnect.github.io/js-sdk/api/typingStatus/startTyping for details.
• The contract for typingStatus.stopTyping() has changed, and it is now required to pass organizationId as part of the 2nd options parameter. See https://tigerconnect.github.io/js-sdk/api/typingStatus/stopTyping for details.

Fixes

• Fix a build issue that was causing incorrect minified code to be generated for the browser release, which was causing a <random variable name> is not defined error when instantiating the client.

Known Issues

• In conversations.fetchTimeline() and conversations.selectConversation():
  • If the user joins a group and then receives more than a certain number unread bangs and messages, when they next select the group, sometimes not all messages will be fetched, if the first fetched item happens to be a bang.
  • If the user enters a forum or a group with replayHistory = true, reads all of the messages, leaves, and then rejoins, then only the very top (lowest sort_number) of the conversation will be fetched.
  • If the user is at the very top of the conversation, and the last fetch of the timeline continuation happened to end on the very first item of the conversation, itemsEstimate might not be 0, which can trigger a one-time fetch in that direction to make sure that there aren't any remaining items. This case should be relatively rare compared to past SDK releases.
• conversations.markAsRead():
  • In some cases, messages in certain conversations might not get marked as read. When this happens, messages might appear Read for a few seconds and then switch back after syncing the real values from the server. This is a platform issue and might potentially be fixed without need for an SDK release.

[8.0.0] - 2019-12-02

General Changes

• conversations.selectConversation():
  • The contract has been changed to return Promise<Object>, where Object contains these two fields:
    • addedMessages: Number of items that were added to the Timeline
    • previousFirstUnreadMessage: A message model representing the firstUnreadMessage before all messages in this conversation are marked as 'Read'.
• If a message attachment without a name is downloaded, a default name of "attachment" will be provided.
• If an attempt is made to upload metadata with non-string values, an error will be thrown. This is an improvement to the previous behavior of allowing the operation to happen, but having non-string values silently discarded by the platform.

conversation:change event removal

• The client.on('conversation:change', ...) event has been removed.
• The conversation.on('change', ...) event has also been removed.
• Please use the following pattern instead: client.models.Conversation.on('afterInject', (resource, conversation) => { /* code that uses conversation */ })

Fixes

• In conversations.fetchTimeline() and conversations.selectConversation():
  • Significant performance improvements have been made. In most cases, unnecessary fetches of timeline when at the end of the conversation have been removed, in which case itemsEstimate will already be set to 0.
  • Fix issue where messages sent by the current user sometimes do not appear in conversation.timeline.
• In conversations.markAsRead():
  • Improvements have been made to the speed at which messages will appear Read.
• In the Conversation model:
  • Fix issue where firstUnreadMessage will have an incorrect value in some situations.
  • Fix issue where isUnread will have an incorrect value in some situations.
  • Fix issue where highestSortNumber will have an incorrect value in some situations.
  • Fix issue where unreadMessageCount will have an incorrect value in some situations.
• In the Organization model:
  • Fix issue where lastIncomingMessageSortNumber had an incorrect value in some situations.
• events.connect():
  • Significant improvements have been made to stability of the EventSource connection.
• Fix issue where message.shouldEnsureRecipientStatus would sometimes have an incorrect value of false after receiving an SSE event or downloading the timeline, which could cause message statuses to not be fetched when calling ensureRecipientStatus().

Known Issues

• In conversations.fetchTimeline() and conversations.selectConversation():
  • If the user is at the very top of the conversation, and the last fetch of the timeline continuation happened to end on the very first item of the conversation, itemsEstimate might not be 0, which can trigger a one-time fetch in that direction to make sure that there aren't any remaining items. This case should be relatively rare compared to past SDK releases.
• conversations.markAsRead():
  • In some cases, messages in certain conversations might not get marked as read. When this happens, messages might appear Read for a few seconds and then switch back after syncing the real values from the server. This is a platform issue and might potentially be fixed without need for an SDK release.