revised

was endorsed as the shared reduction lineage to target; and a two-list

planning model (refactor-list + LR-wants-list) was agreed. A second

hack-a-thon / deep-dive week was committed for the back-end work proper,

to land **before the mid-to-late-June beam outage**. See **Consensus &

decisions** below.

to land **before the 2026-06-25 beam outage** (outage runs through

2026-08-04, per the SNS FY2026 Q1–Q3 official schedule). See

**Consensus & decisions** below.

Sources used for this summary:

- Bogdan's handwritten notes, `~/Desktop/Hack-A-Thon-2026/20260424071504.pdf`

  (30 pages; pages 4, 7, 9, 15, 19, 22, 23, 26 scanned inverted and rotated

  for reading).

- The shared Google-Doc notes prepared by Glass and circulated during the

  day (`2026-04-23-Notes-GoogleDoc.md`) — this is the most authoritative

  record of the morning-session tables (Mag / Shared / Liquids) and the Web

- Glass's live Confluence page in the Neutron Data Project Documentation

  space — *"Notes from hackathon discussion on Liquids Reflectometer

  requirements for QuickNXS 2026-04-22"* (downloaded PDF at

  `~/Desktop/Hack-A-Thon-2026/Notes+from+hackathon+discussion+on+Liquids+Reflectometer+requirements+for+QuickNXS.pdf`;

  same content as the local `2026-04-23-Notes-GoogleDoc.md`). This is the

  authoritative record of the Mag / Shared / Liquids table and the Web

  vs. Qt comparison matrix.

- Becky's `LR_reduction_workflow_config_parameters.xlsx` (linked in the

  Day-2 files manifest; not read directly — referenced as the forthcoming

  global-vs-per-run reference for the LR template).

- Becky's parameter inventory

  `~/Desktop/Hack-A-Thon-2026/LR_reduction_workflow_config_parameters.xlsx`

  — the canonical list of `lr_reduction` new-workflow config parameters,

  pre-sorted into seven tiers (per-run required, per-run often-changed,

  per-run-group required / often-changed / optional, per-experiment /

  reduction-method globals, instrument-based defaults, and a "not fully

  used" tier).

- SNS FY2026 Q1–Q3 official schedule

  (`~/Desktop/Hack-A-Thon-2026/sns-fy2026-q1-q3-official-updated-2026-03-01.jpg`),

  source: <https://neutrons.ornl.gov/sns/schedule> — used to pin the

  beam-outage dates.

- Meeting transcripts `2026-04-23-Part1-transcript.txt` and

  `Part2-transcript.txt` (Teams auto-transcripts; noisy — used for

  corroboration).

Scientists and software devs present or named: John Hetrick, Valeria

Lauter, Tim Charlton, Erik Watkins, Marie Backman, Becky Anderson, Jim

Browning, Glass Elsarboukh, Kevin Tactac, Volker Urban, **Maria Patrou**

(guest speaker, MVP tutorial), **Addi Malviya-Thakur** (joined in Session

3). Bogdan chaired. Asmaa Qdemat's presence on Day 2 is not clear from the

notes.

(guest speaker, MVP tutorial), **Addi Malviya-Thakur** (attending ad-hoc;

present for portions of Session 3). Bogdan chaired.

One housekeeping aside in the opening line: *"Bring tea for Becky for

Fri."*

Fri." ( *but she's not here for the rest of the hack-a-thon* )

---

## 10:00 – 12:00 · Session 2 — Qt vs. Web, and the LiqRef requirements

This session ran the length of the slot. Glass maintained a live Google

Doc that tracked the room's consensus (see `2026-04-23-Notes-GoogleDoc.md`

for the authoritative tables — they're reproduced in summary form below).

This session ran the length of the slot. Glass maintained a live

Confluence page in the NDIP Managed-Software space (*"Notes from

hackathon discussion on Liquids Reflectometer requirements for

QuickNXS 2026-04-22"*) that tracked the room's consensus — the

`2026-04-23-Notes-GoogleDoc.md` file is a local mirror of the same

content. Authoritative tables live there; reproduced in summary form

below.

### Web vs. Qt — the morning's architectural debate

  easier to dev and maintain.

- **Erik:** agreed in principle.

- **Marie:** aligned with *institutional* direction — all effort at NSD

  is currently pushed toward NOIP/NOVA apps; the organisation "only way"

  is currently pushed toward NDIP/NOVA apps; the organisation "only way"

  is to deploy webapps long-term. QuickNXSv2 is production today

  (apprehension *and* excitement about touching it).

- **Glass / Tim:** a NOIP/NOVA "toy app" already exists and demonstrates

- **Glass / Tim:** a NDIP/NOVA "toy app" already exists and demonstrates

  pieces in place — deploy/replay are working.

- **Tim:** wants to track the effort point-by-point (Qt lane vs webapp

  lane). Note that **the web-app framing presupposes all business logic

  is already in the BE** — i.e. the refactor we were already doing is a

  precondition for either path.

- **?Nelld demo:** a browser demo got stuck loading a problem during the

  session. Tim: *"not all browsers are treated the same; some are not

  supported well."*

- **refl1d demo:** a browser demo of

  [refl1d](https://github.com/reflectometry/refl1d) (the precedent

  webapp from the reflectometry community) got stuck loading a problem

  during the session. Tim: *"not all browsers are treated the same;

  some are not supported well."*

#### NOIP/NOVA capacity / resource model (John's block)

#### NDIP/NOVA capacity / resource model (John's block)

- LDRD funding runs out ~FY26; operationalization falls under NSD;

  primary development on apps will transition. NSD fully commits to

  NOIP/NOVA but solution-probe testing is still unsure.

  NDIP/NOVA but solution-probe testing is still unsure.

- Spin-up of an app is Docker-based, so resources are allocated on

  demand. (refined uses all cores available today.)

- **Paradigm today:** all one character (node); if a job needs a super-

- Tim's users: give them a weblink and one click — benefit is a direct

  link.

### LiqRef requirements table (from the Google Doc)

### LiqRef requirements table (from Glass's Confluence page)

Glass's live table captured the Mag / Shared / Liquids breakdown. I have

not reproduced it here verbatim; see `2026-04-23-Notes-GoogleDoc.md` for

the authoritative version. Essentials only:

Glass's live table on the Confluence page — *"Notes from hackathon

discussion on Liquids Reflectometer requirements for QuickNXS

2026-04-22"* (NDIP-managed software documentation space) — captured the

Mag / Shared / Liquids breakdown. I have not reproduced it here

verbatim; see the Confluence page (or the locally saved copy and the

`2026-04-23-Notes-GoogleDoc.md` mirror) for the authoritative version.

Essentials only:

| Area | Key point |

|---|---|

### Lingering design questions (morning)

- Q: In `$mode`, do log values not fit? A: no, they exist.

- "Linkme / linear — *semantic triples* (subject-object-predicate)" —

  captured as an aside, context unclear from the notes; likely a

  reference to a metadata model discussion.

- **LinkML / semantic triples** (subject-object-predicate) — captured

  as an aside during Glass's discussion. This is the linked-data

  framing Glass pointed to for naming conventions in the code that

  "make more sense": see

  <https://linkml.io/linkml/data/rdf.html>. The idea is that

  parameter / metadata names should map cleanly onto subject-

  predicate-object triples so downstream tools can consume them

  without bespoke glue.

---

### Addi Malviya-Thakur's monologue — technical-debt framing (p.21)

Addi chimed in (he had joined for Session 3) with the two-axis

debt framing he uses:

Addi chimed in (she was attending ad-hoc, present for this portion)

with the two-axis debt framing she uses:

- **Cognitive debt** — "must chase it in N places."

- **Intent debt** — fix one-by-one after you discover it.

## 15:15 – 16:30 · Session 3 — Technical debt, test debt, and the big strategic question

With Addi present for the back half, the session moved to the hard

planning question: what, actually, does this hack-a-thon commit us to?

With Addi present ad-hoc for parts of the back half, the session

moved to the hard planning question: what, actually, does this

hack-a-thon commit us to?

### Technical debt — the concrete backlog (pp.22–23)

### Test debt (p.23)

- Takes a long time to set/forgery the tests.

- Takes a long time to set/forget the tests.

- So many tests rely on local axes — integration tests exercise the

  whole software (e.g. "button Z clicked" in GitLab-CI).

- GitHub Actions cannot do this — must verify behaviour. You have

- **Plan:** hack-a-thon → deep-dive (two-day session with Eric and the

  LR team) → **second Hack-A-Thon** dedicated to the underlying back-

  end refactor.

- The 2026 beam-cycle outage was discussed in the room: starts ~second

  week of June, ends ~second week of August (one team member noted the

  current schedule drops mid-to-end-of-June and picks back up the second

  week of August; SNS is running two megawatts a month ahead of

  schedule). This is the window the back-end plan must be in place for.

- **Beam outage window: 2026-06-25 → 2026-08-04** (per the SNS FY2026

  Q1–Q3 official schedule, updated 2026-03-01; source

  <https://neutrons.ornl.gov/sns/schedule>). SNS is running two

  megawatts a month ahead of schedule, which is why the pre-outage

  lead time is real. This is the window the back-end plan must be in

  place for — i.e. deep-dive and Hack-A-Thon #2 must both land before

  2026-06-25.

---

   this week that forecloses it.

9. **Follow-on cadence committed.** Current hack-a-thon → a two-day

   deep-dive session (Bogdan / Eric / LR team) → **second Hack-A-Thon

   on the underlying back-end** — all before the mid-June-to-mid-

   August beam outage.

   on the underlying back-end** — all before the 2026-06-25 → 2026-08-04

   SNS beam outage.

10. **Day 3 charge (tomorrow):**

    - Marie: deliver the LR-wants bullet document.

    - Kevin / Glass: Google-Doc folder is the working surface; the

- **Is the new-workflow LR branch feature-complete** relative to the

  production `lr_reduction` refined path? (Becky: "may be some stuff

  missing" — the separate-files structure limits merge risk.)

- **Global vs. per-run parameters** — Becky's xlsx is the forthcoming

  source of truth. List it, then wire the UI to read it.

- **Global vs. per-run parameters** — Becky's

  `LR_reduction_workflow_config_parameters.xlsx` is the source of

  truth; it pre-sorts parameters into seven tiers:

  1. *Per-run required* — `method_per_run`, `DBname`, `RBnum`,

     `RB_Ymin`, `RB_Ymax`, `BkgROI`.

  2. *Per-run often-changed* — `data_x_range`, `AutoScale`,

     `useCalcTheta`, `useBS`, `qmin`/`qmax`/`dqbin`,

     `LambdaMin`/`LambdaMax`, `tof_min`/`tof_max`/`tof_bin`,

     `ThetaShift`, `ScaleFactor`.

  3. *Per run-group required* — `Sname`, `experiment_id` (IPTS).

  4. *Per run-group often-changed* — `Normalize`, `Qline_threshold`,

     `Qnorm`.

  5. *Per run-group optional (good defaults)* — `_Spath_override`,

     `_NEXUSpathRB_override`, `_DBpath_override`.

  6. *Per-experiment / reduction-method globals* — `plotON`,

     `plotQ4`, `save8col`.

  7. *Instrument-based defaults* — `mmpix`, `dSampDet`, `nx`/`ny`,

     `dMod`, `xi_ref`, `dS1Samp`, `IncidentTheta`,

     `emission_coefficients`, `use_emission_time`, `dead_time`,

     `dead_time_tof_step`, `DetResFn`, `DetSigma`, `peak_pad`,

     `peak_type`.

  8. *Not fully used* — `_BINpath_override`, `DTCsubname`,

     `BINsubname`, `errBINsubname`, `useGravity`.

  Wire the UI against these tiers and the global-vs-per-run question

  answers itself.

- **Webreport / monitor.sns.gov integration** — adopt Plotly output

  for a diagnostic-plots tab? Blocker or already in flight?

- **NOIP/NOVA resource model for interactive reduction jobs** — John

- **NDIP/NOVA resource model for interactive reduction jobs** — John

  flagged that the *"need X cores / Y GB"* scaffolding exists, but

  neutron-apps haven't exercised it at interactive-GUI latency. Pilot

  in scope?

## Notes on fidelity

- The Google Doc prepared by Glass is the authoritative record for

  the Mag / Shared / Liquids breakdown and the Web-vs-Qt cost/benefit

  framing; I have summarised those tables rather than restating them

  verbatim. If you spot a divergence between the Google-Doc version

  and my summary, trust the Google Doc.

- Glass's Confluence page (*"Notes from hackathon discussion on Liquids

  Reflectometer requirements for QuickNXS 2026-04-22"*, in the Neutron

  Data Project Documentation space — "NDIP = Managed Software") is the

  authoritative record for the Mag / Shared / Liquids breakdown and the

  Web-vs-Qt cost/benefit framing; I have summarised those tables rather

  than restating them verbatim. If you spot a divergence between the

  Confluence version and my summary, trust Confluence.

- The Teams auto-transcripts are noisy (garbled, "LR-reduction" →

  "referend"/"LRRed"; "mr-reduction" sometimes → "demanded Albertson";

  speaker attribution missing), and the final 20 minutes of Session 3

  are where the decisions actually landed — I trusted the handwritten

  notes over the transcript here.

- A handful of tokens in the handwritten notes were ambiguous: "Nelld"

  on page 3 (during the bad webapp demo); "UDP = Managed Software" on

  p.11 (likely a NOIP term); "linkme/lineal — semantic triples" on

  p.12 (context unclear — appears to be a metadata-model aside).

  Please correct in review.

- Maria Patrou and Addi Malviya-Thakur were noted-present on

  handwritten notes but not listed in the invite-block of the agenda;

  I've treated them as guest / ad-hoc attendees for the MVP tutorial

  and Session 3, respectively.

- Maria Patrou (guest speaker for the MVP tutorial) and Addi

  Malviya-Thakur (ad-hoc attendee, present for parts of Session 3) were

  noted-present in the handwritten notes but not listed on the

  invite-block of the agenda.

---

Please read ~/Desktop/Hack-A-Thon-2026/2026-04-23-Files.md. Please use that information to summarise the Day 2 events and write them to tasking/Hack-A-Thon-2026-Day2-Summary.md for me to review. If there is any other information that you would like to have, please ask me.

Claude produced [

### Prompt 1.2

Regarding LR_reduction_workflow_config_parameters.xlsx, I have downloaded it to ~/Desktop/Hack-A-Thon-2026/LR_reduction_workflow_config_parameters.xlsx if you want to inspect it. A few other notes:

1. "Nelld" is "refl1d" and which has a public github repository at: https://github.com/reflectometry/refl1d

2. "UDP = Managed Software" is "NDIP - Managed Software". It was a note to myself to obtain/extract [a page that Glass was working on during Day 2](https://ornl-neutrons.atlassian.net/wiki/spaces/NDPD/pages/785252365/Notes+from+hackathon+discussion+on+Liquids+Reflectometer+requirements+for+QuickNXS+2026-04-22). I have downloaded it to ~/Desktop/Hack-A-Thon-2026/Notes+from+hackathon+discussion+on+Liquids+Reflectometer+requirements+for+QuickNXS.pdf

3. "linkme/lineal semantic triples" is "LinkML/Semantic Triples" and was a note to myself obtain the [page that Glass referred to during his discussion on Day 2](https://linkml.io/linkml/data/rdf.html) where he related language terms such as subject-object-predicate to naming conventions in code that make more sense.

4. Addi is a female, please correct all references to her in the summary as necessary

5. Addi attended ad-hoc when she could

6. GenX was agreed to be dropped, I accept that resolution from the discussion.

7. Beam-outage dates are 2026-06-25 (begin) to 2026-08-04 (end). I have downloaded ~/Desktop/Hack-A-Thon-2026/sns-fy2026-q1-q3-official-updated-2026-03-01.jpg from https://neutrons.ornl.gov/sns/schedule for you to incorporate into the summary as appropriate