Stm daq develop 2026

[Etho-b02]

An update version of the offline unpacking module which reads in container fragments. Container fragment represents corresponding detector data. Also includes an event and job summary of what the module reads.

[FNALbuild]

Hi @Etho-b02,
You have proposed changes to files in these packages:

• STMReco
• STMMC
• DAQ
• RecoDataProducts

which require these tests: build.

@Mu2e/fnalbuild-users, @Mu2e/write have access to CI actions on main.

📝 The author of this pull request is not a member of the Mu2e github organisation.

About FNALbuild. Code review on Mu2e/Offline.

[AndrewEdmonds11]

Thanks for opening the PR, Bryan. This looks good. Just a few in-line comments we should work through, Let me know if you have any questions. Thanks

[AndrewEdmonds11]

I think since this module is just for writing the binary file for debugging, we don't need to create the Offline data products. You can remove these lines and others in this module

[AndrewEdmonds11]

I think we don't want to hardcode the numbers 103 and 203. Is it possible to define them in the STMFragment.hh file?

[AndrewEdmonds11]

You can remove commented out lines like this to clean up the code

[AndrewEdmonds11]

Can we remove these parameters now, or are they being used somewhere else in the code?

[AndrewEdmonds11]

I think we can remove that // TODO comment because I think we want to keep this function

[AndrewEdmonds11]

Let's put this FIXME on our to-do list to fix ASAP. Maybe we can even get it in before this PR completes review...

[oksuzian]

Bugs / Blockers

1. Instance-name mismatch between producer and downstream fcls — multiple files will fail at runtime. STMDigisFromFragments only ever calls produces<> with HPGe/LaBr suffixes (rawHPGe, zsHPGe, phHPGe, rawLaBr,
   zsLaBr, phLaBr). But the downstream fcls reference unsuffixed instance names:

• STMReco/fcl/mwd.fcl — stmWaveformDigisTag : "makeSTMDigis:zs" for both phHPGe and phLaBr producers.
• STMReco/fcl/plotSTMPHSpectrum.fcl — stmPHDigisTag : "makeSTMDigis:ph".
• STMReco/fcl/makeSTMHits.fcl — stmPHDigisTag : "makeSTMDigis:ph" (and the other instance "phHPGe" references a module label phHPGe, which doesn't exist in this fcl either).
• STMReco/fcl/plotSTMWaveformDigis.fcl — "makeSTMDigis:raw", "makeSTMDigis:zs", "makeSTMDigis:ph".

These will throw "no product found" at runtime. The reason CI may be green is that each fcl has one of HPGe/LaBr commented out of trigger_paths (see #15), so end-to-end runs aren't actually exercising these
tags.

2. MakeSTMHits constructor hardcodes STMChannel::LaBr for both instances. STMReco/src/MakeSTMHits_module.cc:
   ,_channel(STMChannel::LaBr)
   replaces the previous STMUtils::getChannel(config().stmMWDDigisTag()). Same root cause as Andrew's FIXME on PlotSTMWaveformDigis_module.cc:68, but unmarked here. Currently masked because makeSTMHits.fcl
   comments out the HPGe path — re-enabling it will silently apply LaBr energy calibrations to HPGe data.

3. STMPrintFragments::analyze prints the outer container block_count times instead of iterating into inner frags.
   for (size_t ii = 0; ii < contf.block_count(); ++ii){
   const auto dataBegin = frag.dataBegin(); // outer frag, ii unused
   const auto dataEnd = frag.dataEnd();
   ...
   }
   The loop variable ii is never used; frag.dataBegin() always returns the outer container's bytes. Should be auto inner = contf.at(ii); auto dataBegin = inner->dataBegin();.

4. _unreadInnerFrags is gated by verbosity — counters are zero at default verbosity. STMDigisFromFragments_module.cc:
   else {
   if (_verbosityLevel >= 3) {
   ++_unreadInnerFrags; // both job and event counters live inside the if
   ++unread_InnerFrags;
   }
   }
   The endJob "Unread frags" line will always read 0 unless verbosity is ≥ 3. Move the increments outside the if.

5. xAxis condition flipped — looks like an intended pre-existing bug fix; please call it out in the PR body. STMMovingWindowDeconvolution_module.cc:~166 changed if (_xAxis != "") to if (_xAxis == "") around the
   throw. The new condition matches the message ("No xAxis scale defined despite requesting verbosity level >= 5") and is correct, but it's a behavior change that future spelunkers will want to see noted.

Risks / Robustness

6. Unknown container fragment IDs are silently dropped. Beyond Andrew's hardcoding comment, the bigger issue is the silent path: when ContainerFragID is neither 103 nor 203, both isHPGe and isLaBr stay false,
   the nested if(isHPGe) ... else if(isLaBr) ... skips every branch, and no log line records that an inner frag was discarded. Add a default log at verbosity ≥ 1.

7. ZSfromRaw != totalLen cross-check has lifetime issues. readRawZSinfo is set true after a Raw block and never reset, so it then checks every subsequent ZS block in the same container against the same
   ZSfromRaw. If only one Raw–ZS pair per container is expected, document it; if multiples are possible, reset readRawZSinfo/ZSfromRaw after each ZS check.

8. STMBinaryDigisFromFragments::produce() allocates four unique_ptr<...Collection>s that are never event.put()'d. Andrew flagged removing them in spirit; concretely they're per-event allocations of
   STMWaveformDigiCollection × 3 + STMPHDigiCollection × 1 that immediately get destroyed.

9. STMPrintFragments declares BTrk_difAlgebra as a build dep but never uses BTrk. DAQ/CMakeLists.txt:138. Looks copy-pasted from a CRV module — drop it.

Code quality

10. Five debug std::couts not gated by verbosity.

• STMMovingWindowDeconvolution_module.cc: std::cout<<"BG"<<_xAxis<<std::endl; (author-initials marker), "event id = ", "waveformDigisHandle size = ", "waveform id =" (last three printed every waveform of every
  event).
• PlotSTMWaveformDigis_module.cc: std::cout<<"size = "<size()<<std::endl;

11. STMPrintFragments header comment is wrong: "STMPrintFragments_plugin: Add CRV data products to the event". Copy-paste from a CRV module.

12. _totalRaw is ambiguous — it includes the Empty + Zero counts. A reader of the job summary doesn't know whether to subtract. Either rename to _totalRawSeen or add a _totalGoodRaw line.

13. Top-level non-container fragments silently no-op. The else { /* fallback non-container case */ ... } at the outer loop never logs anything — same family as Adjust parameters for new production #6.

Nits

14. Commented-out FragmentWatcher block in inspectSTMFile.fcl has three layers of # stacked (# frag: / blank / # { / # module_type ...). Either delete or reformat as a single comment.

15. Test fcls ship with one detector path commented out and the choice is inconsistent across files. mwd.fcl disables LaBr; makeSTMHits.fcl disables HPGe. The PR does not say why.

16. STMBinaryDigisFromFragments and STMDigisFromFragments include the entire ROOT graphics stack (TROOT.h, TApplication.h, TCanvas.h, TF1.h, TString.h, TLine.h, TGraph.h) but don't use any ROOT graphics. Drop.

17. , , included in both new modules, never used. Lint cleanup.

---