DataGrip Early Access Program

Support for more cutting-edge LLMs

JetBrains AI Assistant is leveling up! We've added support for Claude 3.5 Sonnet and Claude 3.5 Haiku, provisioned in Amazon Bedrock. This means sharper responses, faster insights, and an even smoother experience.

We're also expanding the lineup of LLMs used by AI Assistant to include OpenAI's o1, o1-mini, and o3-mini, providing more versatility and speed. Whether you need deep analysis or quick yet well-considered insight, AI Assistant in DataGrip is up to the task like never before!

Support for loading full cell values in place

DataGrip has a setting that limits the amount of data that's loaded in each cell. This is designed to improve performance when there is too much data in your tables. Partially loaded cells look like this:

But in some cases, you need to load all of the data, and now that's possible! To do this, hover over the cell and click Reload Cell on the quick actions popup toolbar.

[SQL Server] .NET JDBC driver by JetBrains: Shared memory support

This JDBC driver was initially developed by the Rider team. From this release, it's also available in DataGrip.

The main feature of this driver is its support for the shared memory protocol. This protocol is used by clients to connect to the SQL Server instance that is running on the same machine. SQL Server's shared memory protocol is the simplest protocol, as it has no configurable settings that need to be tuned in order to use it.

[Oracle, SQLite] Suggestion to switch to single session mode when using temporary tables

When creating temporary tables, DataGrip now suggests you switch to single session mode. This mode will allow you to see these tables in the database explorer.

Support for tilde in path fields

We have supported the use of the tilde (~) in the path fields for SSL and BigQuery key and certificate files. Now, you can use the character to specify directories in the Data Sources and Drivers dialog. To do that for your SSL connection, open the dialog by pressing Cmd+; (macOS)/Ctrl+;(Windows/Linux), and select the data source you want to change the settings for. Then open the SSH/SSL tab and navigate to the SSL section.

[PostgreSQL, MySQL, MariaDB] Geo type data display

For the PostgreSQL, MySQL, and MariaDB spatial databases, DataGrip can now correctly display geo types raw in the grid.

Fixes

• DBE-20795: [Snowflake] ASOF JOIN syntax is now correctly interpreted.
• DBE-21374: Excel file import for larger files now works as expected.
• DBE-2570: [SQLite] DATETIME field data is now stored in TEXT format by default.
• DBE-9253: [Microsoft SQL Server] Execution of the currently selected query now works as expected.
• DBE-22637: The Shift + Enter shortcut behavior now works as intended.
• DBE-17082, DBE-17969: [Oracle, IBM Db2 LUW] The Enable DBMS_OUTPUT setting now works for existing consoles. The setting is now available in the IDE settings dialog and on the query console toolbar.
• DBE-12894: JOIN autocompletion now works as expected.

Hello! We are starting the EAP program for version 2025.1. This is the perfect opportunity to try new features before the official release and share your feedback with us.

Introspection by levels for MySQL and MariaDB

We're continuing our work to enhance introspection performance. Starting from 2025.1, there will be different introspection levels for MySQL and MariaDB, adjusting the amount of metadata that gets loaded depending on the size of your database. In other words, we won't load all the metadata if your database is large. This will greatly reduce introspection time and allow you to start working with newly connected data sources right away.

What are the introspection levels?

• Icon:
• Metadata loaded: Names only.
• Main missing features: Modify Table, foreign key navigation, Find Usages, the ability to search in the text of views and procedures.

• Icon:
• Metadata loaded: Everything except the source code of functions, procedures, views, events.
• Main missing features: Find Usages, the ability to search in the text of views and procedures.

• Icon:
• Metadata loaded: Everything, just as in previous versions.
• Main missing features: None, all features are working as expected.

DataGrip now prioritizes the user experience and performance over the comprehensiveness of functionality. The default introspection level value is set to Auto select, which means that DataGrip uses an internal heuristic to detect how much metadata needs to be loaded.

For now, this heuristic looks like this:

Current schema:

• Level 1 when the number of schema objects is greater than 3000.
• Level 2 when the number of objects with sources is greater than 500.
• Level 3 in all other cases.

Other schemas:

• Level 1 when the number of schema objects is greater than 300.
• Level 2 when the number of objects with sources is greater than 50.
• Level 3 in all other cases.

We are considering whether to adjust the heuristic or even make it customizable.

By default, DataGrip will automatically set the introspection level. If you want to restore DataGrip's previous behavior, go to Data Source properties | Options | Introspection | Default level and select Level 3.

You can also manually select a level if you think that will maximize DataGrip's performance for your use cases.

To summarize, you should no longer experience any significant performance problems related to MySQL / MariaDB introspection. If you do, please let us know about them via our issue tracking system.

Attached schemas in the Explain with AI chat

Explanations of your queries are more effective with the full context! Now, when you ask AI Assistant to explain a query, the corresponding schema is automatically attached to the chat.

[Oracle] The ability to connect with a specific role

DBE-96 has been fixed. It's not every day we close a ticket with an ID less than 100! Starting with 2025.1, you can connect to Oracle and define the specific role for the connection in the connection dialog.

Simply select the desired role from the Authentication drop-down list:

The ability to download drivers from Maven

Version 2025.1 allows you to add custom repositories for downloading drivers. To do so, add the repositories you need to the mirrors attribute of the HOME_PATH/.m2/settings.xml file.

Important fixes

• DBE-9626: The color used for a data source is now applied to the attached file tab.
• DBE-21934: Value completion in the data editor inserts the correct string.
• DBE-18569: The Cancel button works as intended when downloading drivers.
• DBE-15201, DBE-20350: [SQL Server] Stored procedures and system objects are qualified with the relevant schema names if the corresponding setting is enabled.
• DBE-22250: [Snowflake] It is possible to remove Role and Schema after they've been added to the connection parameters.
• DBE-21514: [Oracle] Aliases are inserted for code completion of synonyms.
• DBE-17959 [PostgreSQL] UNIQUE NULLS NOT DISTINCT is now supported.