Release DBMS InterSystems Caché 2016.2

Hello! The next release of the new version of Caché was number 2016.2. There are not many changes, but they are all important. As always, first we publish a link to the full list of changes (in English).

So.

Atelier support - new Eclipse-based IDE

This is the first release of Caché, which adds server support for the new Atelier integrated development environment.

Atelier combines the strengths of the popular IDE Eclipse and Caché DBMS. It supports exporting code to UDL, out-of-the-box version control, debugging, auto-update and is compatible with most Studio Add-Ins. Atelier is distributed separately from Caché and Ensemble distributions.

There are currently two courses on Atelier available on our learning portal:

• A Tour of Atelier, Our New Eclipse-based IDE
• Moving from Studio to Atelier

Recommended for review.

OAuth 2.0 and OpenID Connect support

Starting with this release, Caché supports OAuth 2.0 and OpenId Connect Core. Caché can act as a client, resource server or authorization server. Read more in the section Using OAuth 2.0 and OpenID Connect with Caché documentation, as well as in an article on our Developer Community.

SQL / JSON support

The SQL syntax has been extended to partially support the SQL / JSON standard, which allows you to get query results in JSON format. Supported features:

• JSON_OBJECT - builds a JSON object for each received string.
• JSON_ARRAY - builds a JSON array for each received string.

SQL enhancements

This release introduces the following SQL performance improvements:

• % Parallel - Improved performance by partitioning requests into chunks that can be executed in parallel on multiprocessor systems. In this release,% Parallel received the following improvements:

  
  
  • Reduction of resource consumption when preparing a request for parallel execution.
  • Dynamic querying of requests based on the number of available physical flows.
  • Increasing the number of types of SELECT queries that can be parallelized. For example, now Caché can break up requests with IN (?,?,?) Or% INLIST () constructs.
  
  

As a result of the above changes, it became possible to use% Parallel with the majority of SELECT queries. If the query is parallelizable, and Caché has several physical threads available, then the performance of the query will be improved. If the request cannot be divided into portions, or only one physical flow is available, then Caché will recognize this situation and will not break the request, saving resources.

• Improved performance bypassing list collections.
• Fixed plans - the ability to “freeze” the query plan while changing indexes or other metadata.

Other changes

Of the other important changes:

• New REST API for accessing source code files stored in Caché. This API was developed to support the Atelier IDE, but it can also be used to create third-party IDEs similar to Atelier. For more information, see the Accessing Caché chapter of the Source Code Files Using REST documentation.
• The DataCheck utility now works more efficiently thanks to parallel processing. The ^ DATACHECK interface has become more convenient.

• The current release of Caché uses faster memory synchronization mechanisms in some critical usage scenarios. Tests of batch loading of data show a performance gain of more than 10% on Linux x86-64. These improvements are relevant to the following platforms:

  
  
  • Apple Mac OS X for x86–64
  • IBM AIX for Power System-64
  • Microsoft Windows for x86–32, x86–64
  • Linux for x86–32, x86–64
  • Solaris for x86–64, SPARC-64
  
  

Thank you for new releases!