[ggmarshall]

• I have reviewed all the code in this pull request and deem it of the standard expected for this project.

Problem

Since #213, LGDOVectorOfVectorsIOManager creates its buffer lazily in the first write(). If the output's unit differs from the variable's (e.g. a timepoint found in sample index and written in ns), get_buffer adds a UnitConversionManager to the chain at that point. By then the first block's processors have already run, so the first block is written from an unfilled buffer.

Result: the first block_width entries of every chain come out as 0, while the vector lengths are still correct. In LEGEND's SiPM chain this hits trigger_pos and trigger_pos_dplms. 2.2.0 got the buffer in __init__ and is not affected.

Fix

In write(), after the lazy get_buffer, run any processor managers it appended, so the block in flight is converted too. Later blocks are unchanged: the conversion is already in _proc_managers.

Tests

• New test_vov_output_unit_conversion_first_block: 10 entries with block_width=4 and tp_max in ns. It fails on main (first 4 entries are 0) and passes with the fix.
• Full suite passes locally: 104 tests.
• I also ran 300 real LEGEND p18 SiPM traces through the SiPM chain with this fix. trigger_pos and trigger_pos_dplms match dspeed 2.2.0 output exactly; without the fix, 15 and 7 traces differ.

This PR was prepared with AI assistance (Claude Code) and reviewed by me.

🤖 Generated with Claude Code

[Copilot]

Note

Copilot was unable to run its full agentic suite in this review.

Copilot review overview

Review effort: Lite
Findings: 1 · 1

Open (2)

• proc_managers is captured before calling get_buffer(), and the subsequent slice assumes… · New
• This assertion line is quite long and may violate common lint/formatting limits (e.g., 88/99… · New

What changed in this PR

Adds a regression test and a processing-chain fix to ensure VectorOfVectors (VoV) outputs are unit-converted for the first processed block as well as subsequent blocks.

Changes:

• Add a test that reproduces missing unit conversion for the first VoV block output.
• Ensure any unit-conversion processors appended by get_buffer() are executed for the first block.

File	Description
tests/​test_processing_chain.py	Adds a regression test validating first-block VoV unit conversion.
src/​dspeed/​processing_chain.py	Executes conversion processors appended by get_buffer() so the first block is converted.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 69.13%. Comparing base (ae7b4e1) to head (72ad10e).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #227      +/-   ##
==========================================
+ Coverage   69.07%   69.13%   +0.05%     
==========================================
  Files          62       62              
  Lines        4188     4192       +4     
==========================================
+ Hits         2893     2898       +5     
+ Misses       1295     1294       -1     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[iguinn]

Better solution to bug in #228