.. include:: /includes.rst.txt

.. comments - headings

   # with overline, for parts
   * with overline, for chapters
   = for sections
   - for subsections
   ^ for subsubsections
   " for paragraphs
   * for H5
   + for H6

.. _migration_8_5:

Migration Notes: EDG 8.4 to 8.5
-------------------------------

Data Platform upgrade
^^^^^^^^^^^^^^^^^^^^^

Deployments that use :ref:`topbraid_data_platform` must also :ref:`upgrade the Data Coordinator server <upgrading_data_platform>` to version 2.0.0-tq-2.

SPARQL Changes
^^^^^^^^^^^^^^

Starting with version 8.5, TopBraid made a low-level switch to the way that the Apache Jena API is handling
pre-bound variables in SPARQL queries.  This may affect any parameterized SPARQL query, for example in ADS scripts,
SWP and SPARQL-based SHACL (or SPIN) features.
TopBraid is now using variable substitution instead of so-called initial bindings for pre-bound variables.

While the full impact of this change is not fully understood, we have so far only observed differences w.r.t.
the semantics of the `BIND` operator in SPARQL.  Previously, when a variable on the right side of a BIND was pre-bound,
the engine had treated the BIND like an equals FILTER.  For example, if your SPARQL query contains
`BIND (?someExpr AS $preBound)` then it was interpreted as `FILTER (?someExpr = $preBound)`.
It is no longer valid to use pre-bound variables on the right hand side, and the SPARQL engine will throw an exception
when it runs into such cases.  You need to re-formulate your queries to avoid this.

If this is too difficult, in 8.5 you can use Use Product Configuration > SPARQL to re-activate the old pre-binding algorithm.
However, this option will be removed once TopBraid has been upgraded to a newer Jena version in which the original
pre-binding algorithm will be completely deleted.