What's New: Server and Database (2025 release wave 2)

Overview

The session covers platform, development and reporting changes in 2025 release wave 2. On the AL side it introduces a sequential GUID method for better indexing, a lock timeout method that takes a duration, and a truncate method that deletes large amounts of data faster than delete, with a long list of conditions under which it returns false.

It also covers database optimizations for flow fields and server-side pivot on date hierarchies in analysis mode, semantic similarity search on metadata (with feature toggles, embedding maintenance and a demo of advanced tell me), report column tooltips in the Word addin data picker, PDF/UA and PDF/A support, SQL call information in performance profiles, and app version in copied error details. Semantic search is cloud-only; the other functionality applies to both cloud and on-premises.

Key points

• Truncate AL method uses SQL truncate, supports filtering and keeps auto-increment values. It returns false for non-SQL or system tables, inside try functions, with security filters, media fields, too many marked rows, filters on flow fields, or before/after delete subscribers. Wrap it in an if and fall back to delete all.
• Truncate should only be used when deleting the vast majority of a table, because filtered truncate copies the kept rows out and back.
• The lock timeout duration method sets a custom duration instead of on/off, but is still limited by the command timeout.
• Flow fields that sum over other tables now use a single outer apply instead of several, which speeds up list pages showing many calculated fields.
• Server-side analysis mode (queries, over 100,000 rows, or joined related tables) can now pivot on date hierarchy fields such as year, quarter and month.
• Semantic search is cloud-only: it needs embedding vector maintenance and SQL Server 2025, which is available in Azure SQL but not yet on-premises. Embedding vectors are removed when the database is exported. Preview semantic similarity search and advanced tell me are separate feature management toggles, and results may change as the algorithm is adjusted.
• Report columns can have tooltips inherited from table fields or overridden in the report; they show in the new Word addin data picker. The speakers recommend putting data descriptions as tooltips on tables rather than pages.

Features

AL objects mentioned

As heard in the captions. A name that matches one object page by exact type and name links to it; the others stay as named.

• table “GL account” at 6:13
• table “GL entry” at 6:13
• other “Word addin” at 15:05

Not found in BC28-30: table “GL account”, table “GL entry”.

Quotes

• 0:38 “it helps being able to run some of the business logic the partners have written in a performant matter”
• 1:31 “GUID is great but GUID is horrible for indexing. So that’s why we stole the code from SQL server to be able to do”
• 2:59 “Truncate is a SQL method that is much faster than delete. It’s not instantaneous. It is actually transactional.”
• 3:14 “the difference between trunk truncate and delete is truncate will just lock in the transa transaction log the page ids and not the full”
• 4:19 “if you’re not deleting the vast majority of the table, don’t use truncate because it’s counterproductive”
• 7:50 “not only a lot, it’s actually almost like um a little over 70% of all customers use analysis mode on a monthly basis”
• 8:22 “we take the input, we calculate a semantic vector, and then we compare to the semantic vectors for each of the metadata elements”
• 11:21 “you need to press enter to start the search because word words can change meaning dramatically when you add an extra letter”
• 13:51 “also do expect that results can change. We will adjust the algorithm for how we treat similarity based on measurements of what works best”
• 16:25 “So save will default now save in PDFUA. Eventually we’ll also have um document layouts in word in a certain way that once they”
• 18:23 “all the functionality we talk about apply both to on premises and the cloud that does not apply to the semantic search”
• 18:34 “we are maintaining the embedding vectors etc. um we do not have support for that in the on-remise world”
• 18:45 “if you take um and export your database, then we’re actually going to take out the embedding vectors before”
• 18:56 “SQL Server 2025, which is available for us in Azure SQL and for you in Azure SQL, but it’s not available yet uh for”

Disclaimers in the video

• 9:07 preview: preview semantic similarity search and application metadata
• 13:51 subject-to-change: do expect that results can change. We will adjust the algorithm for how we treat similarity based on measurements of what works best in production
• 14:06 subject-to-change: do expect that results can change. We will adjust the algorithm for how we treat similarity based on measurements of what works best in production.
• 18:23 not-in-this-release: all the functionality we talk about apply both to on premises and the cloud that does not apply to the semantic search

Presenters (as heard): Kenny Pontop, Yens.