ingestr v1.0.38 - SQL Server and SharePoint Sources

ingestr v1.0.38 was published on June 19, 2026, for the ingestr ELT runner. The most useful change for data engineers is the wider source surface: SQL Server Change Tracking and SharePoint are now in the connector set, while the sports data sources get cleaner schema and interval behavior. This release is not marked as a prerelease in the GitHub metadata.

The full release notes and downloads are on the GitHub release page.

The release adds a SQL Server Change Tracking source. That matters when SQL Server is the system of record and the database already tracks row change state. It gives ingestr a more natural path for incremental extraction from tables where full scans are too expensive or too noisy for routine ELT.

SharePoint is also added as a source. The notes do not spell out the exact object surface, so this is not a place to infer file, list, or drive behavior beyond the source addition itself. The practical point is narrower and still useful: teams that have SharePoint in their ingestion estate can now evaluate it inside the same runner rather than wrapping it with a separate staging script.

The changelog also mentions added history tables for existing objects and a case related object update, but that entry is truncated in the release notes. Treat it as a signal to inspect generated schemas before promoting jobs that depend on those objects.

A sizable part of v1.0.38 is connector maintenance for sports data sources. api-football and football-data are refactored to use schema inference, and BallDontLie gets schema inference along with a replace strategy refactor. This is not a headline feature, but it is exactly the kind of change that affects warehouse table shape and test fixtures.

The source scheme names also change. The api-football source moves to the apifootball scheme, football-data moves to footballdata, and BallDontLie uses balldontlie. For existing jobs, scheme names are operational configuration. Review source URIs before rerunning scheduled tasks, especially where the same pipeline definition is shared across environments.

The release notes also mention api-football.md updates and an aligned football-data research document. Those are documentation changes, but they matter when a source has just changed scheme naming and schema behavior.

The football source cleanup removes date_from and date_to URI params in favor of interval. That is a small surface change with direct pipeline impact. Jobs that build source URIs from scheduler windows should pass interval intent consistently instead of mixing date params into the URI.

api-football also changes merge behavior. Merge is used only for tables that can load incrementally, and fixture based tables are merged in a way that honors the interval. This is the right kind of narrow fix: it does not promise a new sync model, it narrows when merge is appropriate so load strategy follows what the table can actually support.

The rate limit burst for api-football is lowered to stay under the free tier cap. That may reduce burst throughput, but it is usually better than discovering the cap through failed jobs in production. Operators should account for that in run timing if several football tables are pulled together.

There are two correctness notes worth calling out. First, stale primary keys after removing flatten helpers are fixed. Primary key drift is one of those defects that can look harmless until it breaks deduplication or merge matching downstream, so this belongs in the upgrade checklist for any connector touched by that path.

Second, match event synthetic keys are protected from API field collisions. That is a data shape concern, not just internal cleanup. If an upstream API adds a field that overlaps with an internal event key, the extractor should not silently corrupt the synthetic identity used later in the pipeline.

Cancellation is also tightened, and the streaming test flush loop gets a type correction. Those are mostly internal maintenance items, but they point at run control and test reliability rather than cosmetic refactor. For operators, the useful takeaway is to rerun cancellation and streaming test cases if they exist in local checks.

• Release page URL: GitHub release page
• Repository URL: GitHub repository
• Tag: v1.0.38