Existing art/FHiCL workflows¶
The tables below list FHiCL files and producers from the August 2026 FD workflow presentations. They identify migration candidates; running them requires checking the filenames and dependencies against a specific dunesw release.
Only two of the six workflow presentations list the stages and FHiCL files of art production chains. The remaining presentations either describe non-FHiCL workflows or do not provide enough FHiCL detail to reconstruct the chain, as listed at the bottom of this page.
FD sim/reco: TPC (HD 10kt, long-baseline)¶
As given in Chappell's talk, for the HD 10kt long-baseline path:
| Stage | FHiCL | Producers |
|---|---|---|
| Generation | prodgenie_common_dunefd.fcl |
rns, generator (GENIE via GENIEGen, tuned by genie_dune.fcl) |
| G4 | standard_g4_dune10kt.fcl |
largeant, IonAndScint |
| Detsim | standard_detsim_dune10kt.fcl |
tpcrawdecoder (Wire-Cell, config wirecell_dune.fcl) |
| Reco1 | standard_reco1_dune10kt.fcl |
gaushit, spsolve, hitfd (HD only) |
| Reco2 | standard_detsim_dune10kt.fcl (as given on the deck; same FHiCL name as Detsim, later stage) |
pandora (config pandoramodules_dune.fcl in dunereco) and downstream (pandoraTrack/Shower/calo/pid, cvnmap/eva, emtrkmichelid, energyrec*) |
HD and VD workspaces mostly run the same producers with only configuration changes; the deck's own worked example is HD 10kt.
FD sim/reco: PDS (photon detection side, same chain)¶
As given in Paulucci's talk, the PD-focused view of the same FD sim/reco chain:
| Stage | FHiCL | Modules |
|---|---|---|
| G4 | standard_g4_dune10kt.fcl |
largeant, IonAndScint, PDFastSim (uses LArPropertiesService), PhotonBackTracker |
| Detsim | standard_detsim_dune10kt.fcl |
tpcrawdecoder (TPC side), opdigi (uses OpDetResponseInterface) |
| Reco1 | standard_reco1_dune10kt.fcl |
gaushit, spsolve, hitfd (TPC side), opdec, ophit |
| Reco2 | standard_reco2_dune10kt.fcl |
TPC modules, opflash, opslicer, flashmatch |
The source decks disagree on Reco2: Chappell's talk names standard_detsim_dune10kt.fcl for the Reco2 stage; Paulucci's talk names standard_reco2_dune10kt.fcl for the same stage. Both decks are primary sources from the same working group, dated 2026-08-10 and presenting companion views of the same chain. This should be checked against the actual dunesw FHiCL files before treating either name as authoritative for a migration task.
Candidate first migration¶
Chappell's own prioritization proposal (see the FD TPC page) names a specific entry point: start at gaushit and work outward through spsolve, hitfd, pandora, pandoracalo, pandorapid, energyrec*, using pre-existing art files as input. This would allow early comparison with LArSoft output if Phlex/FORM can read the required inputs.
For the migration itself, follow the Phlex developers' Migrating to Phlex guide. Its worked example is GausHitFinder, the module behind the gaushit producer, so this entry point matches the guide directly.
Backward compatibility is a validation gate, not yet a demonstrated guarantee. The test must specify the Phlex/FORM version, input-file format, product dictionaries, provenance, and associations needed by the chosen modules. See I/O, persistence, and associations.
The alternative path in the same deck, starting from generator and rebuilding the full chain, does not provide that early file-based comparison and was not the presenter's recommendation.
Subsystems without a documented art/FHiCL chain¶
- SAND: explicitly stated on the source deck, "We do not use LArSoft at all." Built on
μfw(repo slugufw), a separate C++ micro-framework, JSON-configured, not FHiCL. See the SAND page. - ND-LAr+TMS and ND prototypes: driven by
install_[step].sh/run_[step].shshell scripts and environment variables across Docker containers and Python virtual environments (GENIE, edep-sim, larnd-sim, ndlar-flow, SPINE), not FHiCL. Pandora andtms-recoin this chain use SL7/UPS/cmake builds, but the deck does not name FHiCL files for them. See the ND-LAr+TMS page. - protoDUNE: the current source deck (Calcutt) establishes that protoDUNE's raw and Reco1 data already live inside art/LArSoft, but does not walk a step-by-step FHiCL chain the way the FD-TPC and FD-PDS decks do. This is the same gap already noted on the protoDUNE page: a full production-chain deck for protoDUNE does not exist yet in this documentation.
Source: Chappell, FD TPC talk, Phlex Adoption WG, 2026-08-10; Paulucci, FD PDS talk, Phlex Adoption WG, 2026-08-10.