Cloud CMS Release 3.2.90

The following are the new features, enhancements and fixed issues for Cloud CMS 3.2.90 release.

Upgrade Considerations

Please review the upgrade considerations provided here:
https://gitana.io/documentation/gitana/3.2/guide/guide/docker/upgrades/3-2-90.html

With this upgrade:

• MongoDB 4.4 or above is required.
• You will need to perform a cold restart of your API cluster.

Enhancements

File Folder Service

• Faster and more accurate paths-based lookups that further utilize MongoDB aggregation pipelines.

Handlebars

• Added two helper functions (urlEncode and urlDecode) to offer assistance while generating URLs or paths within templated outputs (such as deployment config paths).

Job Queue

• Improved performance of the job queue with faster job dispatching and lower memory overhead.

Repository

• Added repository config field enableStrictDictionaryRefs to allow for separate control of strict dictionary checks on $ref types (defaults to true).

Search

• Faster implementation of paginated calls that leverage Elastic Search search_after mechanics instead of scroller IDs.

Transfer Import

• Faster post-import assignments for propagated authorities.

User Interface

• The All page size option has been relabeled to 1000 to more accurately reflect the max API pagination page size.

Bug Fixes

Associations

• Fix so that paginated handlers use faster pagination options (to avoid counting rows on inner queries).

Concurrency

• Fix so as to prevent datastore configurations from writing incorrectly under high concurrency.

Dictionary

• Fix so that dictionary recompilation works for certain cases where multiple definitions are recompiled at once and some of those definitions are internally dependent on one another.

General

• Fix so that thread locals clean up more thoroughly and with better logging on the termination of each request.
• Fix so that HTML parsing of inline links works faster and more accurately while retaining the original reconstruction of output HTML.

Search

• Fix so that indexing of content properties will use type namespacing. As a result, two properties that share a common path and have different types will no longer result in collisions (or unindexed values). This fix requires branches to have searchSchemaVersion set to 2. This feature is disabled by default.

User Interface

• Fix to content type builder page so that form selector bypasses properly for cases where no default form is configured.
• Fix so that the awareness channel clears upon navigation (preventing some errant page refreshes).
• Fix so that the invite project modal’s from address is optional (as this is generally provided by the email provider).
• Fix so that pagination works consistently across the document library and content instances pages.
• Fix so that field projections are applied consisntently within node and file pickers.
• Fix so that pagination state is retained when adjusting table within documents lists.