This release does not introduce any major change but focuses instead on two major goals:

For Consolidation Server, compared to CloudView 2015x.SP3, the Transformation step can be up to 10 times faster and the Aggregation step can be up to 3 times faster.

Managed Documents are created much quicker.

We have also extended our test coverage on industry related use cases.

Please read carefully the Other Improvements section below.

• Avoid loading payload for each document matched when not necessary (#24774)

• Load virtual vertex arcs from disk only if we there are no arcs pointing to/from it in memory (#24565)

• isFetchOperation is exposed for Java Aggregation processor (in the ICreateAggregationHandler interface) (#24589)

• Methods added to delete parts or parts metas for Groovy and Java processors (#24402)

• The PushAPITransformationHelpers provide a method to add an arc between two vertices not bound to the current PAPI document (#24657)
Beware, in that case, the ownership of the arc is taken by the first vertex, not the PAPI document, meaning that if this particular vertex is modified, the arc will be deleted automatically.

The same operation should be pushed once again to recreate the arc.

• Better highlighting of the current node in the introspection graph (#24396)

• Text copy/selection on payload information is enabled (#24525)

• When dealing with a large number of properties in data model, pagination is enabled on related pages(#24530)

• It is possible to delete a document via the Push API using delete by "Prefix" instead of "RootPath" (#24745). This can be useful if children document URIs are not built using a "/" as separator.

This method has to be used with care by the Push API client because it has to handle the separator on its side.

 

Sending deleteFromPrefix("eno:bo:master") will delete all above documents even if the intention was to only delete eno:bo:master and eno:bo:master:subpart2 .

• New condition behavior for "Copy Context Chunk" document processor to select only a subset of values from a multi-valued field (#24201)

 

• Adding {hl=0} to an arbitrary node in ELLQL or UQL prevents the associated subquery from being used for summary/highlight (#24402).
This option can be useful when query restrictions are not supposed to be visible from the user side (which can occur in case of unwanted highlighting)

• Template prefix handler can be used inside a query template to execute a query rewriting just after parsing (#24532)
For example, with the following definition:
<TemplatePrefixHandler name="myTemplatePH" aliases="myTPh,my_tph" template="categories:\"path/here\" AND title:(__QUERY__)">

the query:

will be rewritten as follows:

• As language detector processor may output several relevant languages, it is possible to limit the number of detected languages that are kept (#22853)

• Alerts (id or name) producing error when compiling alert rules are written in log file (#24486)

• Alerts having a syntax not supported by the FastRuleBuilder are explicitly logged (#24387)

Look for "Unable to convert ELLQL to FastRule for alert(s)" in log files

• From the cvconsole, it is possible to dump every used dictionaries for any field type (#24356)

• The CloudView server now supports DSLS (Dassault Systemes Licensing Server) for licensing mechanism (#24513).
Check the "Installation procedures" and "Monitoring" sections in the Installation & Administration guide.

If you use the Consolidation Server, you MUST run all aggregations before migrating.

During the migration, the graph data structure used by the Consolidation Server is rewritten to improve performances. This operation can take a few minutes but may take a lot of RAM.
If you encounter an OutOfMemoryError, you should edit the install.sh script to increase the Xmx and relaunch the operation.