Version 0.6.4
stable
Added a toggleable result timeline for time-based Functions, quick case-insensitive text searches across message headers, keys, and values, consistent found/read counts, and JSON syntax highlighting for schema-decoded message details.
What changed
- New: A toggleable result timeline histogram shows how many displayed messages fall across the scanned time window for time-based Functions (since, from, today, utc/abs ranges). It sits between the result controls and the results table and is enabled with the chart toggle beside Import/Wrap.
- Drag across the timeline to zoom: the Function field updates to the selected UTC range and displayed results filter immediately. Reset zoom restores the previous Function and retained results, Esc cancels a drag, and bars are keyboard selectable.
- Timeline arrows step the window by one bucket or jump a full displayed range, updating the Function and fetching the new range automatically; navigation is disabled while playing or live.
- The timeline is available only for time-based Functions. Offset-based functions such as tail, head, beginning, and offset= do not offer the chart because partition-local offsets, especially on compacted topics, are not a reliable timeline.
- Axis ticks at round time intervals, gridlines, bucket tooltips, and theme-tinted bars make the scanned window easier to read; labels follow the configured message timezone.
- Fix: The message details Message and Key editors now apply JSON syntax highlighting when the format is Schema-Json, Schema-Avro, or Schema-Protobuf, matching the decoded JSON shown for schema-recognized payloads.
- Structural indent guides are now available for Schema-* formats, consistent with JSON payloads.
- Start a message filter with an apostrophe (') to perform a case-insensitive literal text search across header names and values, decoded key text, and decoded message value text, bypassing SL criteria evaluation. No closing quote is needed.
- Quick text searches are one-off filters and are not added to recent filter history; SL autocomplete and formatting are disabled while text-search mode is active.
- Fix: The result status consistently shows messages found versus messages consumed/read, even without a filter or after clearing it. While a scan runs and no read count has arrived, the status shows 'Searching...' instead of a placeholder.
- Fix: Re-playing an unchanged query keeps retained results and counts correctly: totals accumulate across incremental replays instead of resetting, so a second play no longer shows counts like 150/0.
- Fix: Re-playing a rolling time window such as since=1m evicts retained rows that have aged out of the window, and the found/read counts reflect only records in the replayed range.
- Fix: Re-playing a filtered scan resumes each partition from the backend's consumed frontier rather than the last matched row, so records that did not match are not re-read and double-counted.
- Fix: Read counts on incremental replays are bounded by the configured window (for example tail=5 over three partitions stays at 15): rows and counts for records that slide out of the tail are subtracted as new records arrive.
- Sending a message from the details panes invalidates the remembered playback for that topic, so the next Play clears the model and replays as a fresh request instead of a smooth incremental update.