My task is to create full-text index for a cluster of interconnected vertices with one root element. The only way a found to solve my problem is manual indexes.
But according to the documentation, manual indexes are marked as deprecated.
Why?! Сan i rely on this functionality in the future?
yes manual indexes are going to be removed in the future.
Why do you need manual indexes?
Can’t you just index records with lucene indexes?
I use orientdb as a back-end store for EMF objects (Eclipse Modeling Framework).
Usually, one object is decomposed into a whole graph, vertices and edges. To find an object by keyword, I have to search the entire graph.
For example, you need to find all ETL transformations that mention the DM_BALANCE table.
Examples of possible options:
Transformation-sources->SQLSource.query=“select * from DM_BALANCE”
My idea is that at the time of creating or changing the Transformation object, I combine all the text attributes of all nested objects in one line and use this pseudo-attribute to build a manual index on the Transformation object. Of course, I do not want to store such a pseudo-attribute.
Can you suggest other approaches?