snowplow-media-player v1.0.2

Summary

This release fixes two issues in snowplow_media_player_base. duration_secs was being overwritten by the maximum duration reported for the same media_identifier, even where a play had already reported its own duration; the media level value is now only used as a fallback for plays that reported none, matching the duration_fix CTE's documented intent, and is_complete_play/content_watched_percent now divide by the same resolved duration instead of disagreeing on which value to use. Separately, a play could be silently split into multiple rows in prep if page_url changed mid-play due to an in-page anchor (e.g. #comments), causing data loss for the discarded row(s); page_url is now normalized to drop the fragment before grouping. It also addresses a missing on_schema_change='append_new_columns' config which is needed for passthrough fields.

Fixes

• Only apply the media level duration where the play reported none, and use the same resolved duration for is_complete_play and content_watched_percent
• Stop URL fragment drift from splitting media plays
• Add append on_schema_change to incremental tables

Upgrading

Update the snowplow-media-player version in your packages.yml file.

Neither fix rewrites existing historical rows on a normal incremental run — new data
is correct going forward regardless. Both issues depend on specific conditions and
may not apply to your historical data at all:

• The duration fix affects duration_secs, is_complete_play, and
  content_watched_percent where a play's own duration was missing or inconsistent
  with other plays of the same media.
• The URL fragment fix affects plays on pages where the URL can change mid-playback,
  e.g. via an in-page anchor link.

If you want to correct already-processed historical data for either issue, see the
full or partial refreshes guide.