> For the complete documentation index, see [llms.txt](https://resources.unizin.org/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://resources.unizin.org/products/data-and-analytics/unizin-data-platform/release-notes/udp-marts-release-notes/1.0.105.md).

# 1.0.105

## File Interaction Mart Logic Change and Monthly Partitions

| Version      | 1.0.105         |
| ------------ | --------------- |
| Release Date | October 5, 2026 |

## Key Features

### File Interaction Mart Logic Change

 We removed a logic filter in the [file interaction](/products/data-and-analytics/unizin-data-platform/data-stores/data-marts/file-interaction.md) mart to identify only files uploaded by instructors. Originally, we filtered based on `uploader_id` and `role` to identify files uploaded by instructor(s) in courses. However, we received feedback that files that are copied from one course template to another do not retain `uploader_id` information, and this field is `null` for those files. This resulted in the mart underrepresenting files.

So, we removed the `uploader_id` and `role` filters in the SQL logic. This now shifts the mart's purpose a bit, but existing use cases are still supported. Previously, the mart was solely instructor-focused. Now, the mart includes all files that reside within course contexts. 

### Monthly Partitions for Two Marts

This update is purely "under the hood" and focuses on optimizing BigQuery refreshes for Caliper event portions of two marts: publish timestamp in the [course status mart](/products/data-and-analytics/unizin-data-platform/data-stores/data-marts/course-status.md) and last activity timestamps in the [last activity mart](/products/data-and-analytics/unizin-data-platform/data-stores/data-marts/last-activity.md).

Previously, we partitioned the `event` helper tables for these marts based on the date of the `event_time` field. For years, this worked seamlessly; however, BigQuery has a limit of 2000 partition updates per query execution, and 2000 partitions (i.e. 2000 event days) based on the existing logic was recently surpassed for these marts with historic data in scope. 

Moving forward, these marts will be partitioned on a monthly basis rather than a daily basis. This will allow BigQuery updates of these marts to happen seamlessly again.