Skip to content

docs(prior-art): dated citation check of every rules.yaml value #217

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
obvious-autobuild wants to merge 1 commit into
base: main
Choose a base branch
from
+58 −0
Draft

docs(prior-art): dated citation check of every rules.yaml value #217

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
58 changes: 58 additions & 0 deletions docs/04-prior-art-and-codes.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,6 +38,64 @@ The demo's `server/rules.yaml` (PR #11) uses these values, except the 1 ft wall
| **NEC 230.85** | An outdoor emergency disconnect. The 2026 NEC moves it into 230.70. |
| ComEd opening rules | UNVERIFIED. Cite Austin Energy or Texas Gas Service instead. |

## Citation check

Every value in `server/rules.yaml` was rechecked against the source it cites on 2026-10-09. The file is identical on `main` and `t3/server` as of that date (compared with `git diff`), so one table covers both branches. This is a check on whether citations are accurate. It is not a determination that any placement, value or installation is safe, code-compliant or approved; the server still reports UNSURE wherever a scan does not settle a check.

Statuses: **verified** means the source says what the citation claims; **partly supported** means it supports the number with different wording, scope or role; **nothing to check** means the row claims no public source. ICC's Digital Codes viewer renders code text only in a browser and answers automated checkers with 403, so NEC and IRC wording was confirmed from public pages that quote the code; each such page is marked *secondary*.

### Values citing public sources

| rules.yaml value | Cited source | Source reachable | What the source says | Edition or date | Status |
| --- | --- | --- | --- | --- | --- |
| `clearances.gas_ft` = 3.0 ft | Base help page; Austin Energy §1.9; Texas Gas Service | All three. Base renders in a browser; the Austin Energy and Texas Gas PDFs download directly and their text extracts cleanly. | Base: "at least 3 feet apart from gas meters". Austin Energy §1.9.2(C)(4): meters shall not sit "within a circle radius of 3 feet of gas meters, regulators, relief valves, and electrical apparatuses". Texas Gas: electric meter or outlet "3' H x 3' V" from the regulator's relief-valve vent. | Base page edited 2026-09-30; Austin Energy manual effective 2023-12-12; Texas Gas sheet revised 01/22 | Base verified. Austin Energy and Texas Gas partly supported: both are meter-siting rules that put electric equipment 3 ft from gas equipment; neither regulates battery placement. Applying them to the battery is a demo choice. |
| `clearances.ac_ft` = 3.0 ft | Base help page | Yes | "Each battery is roughly the size of an AC unit and needs 3 feet of clearance from gas meters, AC units, fences, other batteries, or other obstructions." The sentence sits in the photo-checklist article (10280641), not the placement article (10280705) the gas rule quotes. | Edited 2026-07-29 | Verified on the value. The citation should name which help page. |
| `clearances.battery_ft` = 3.0 ft | Base help page; IRC 2021 R328 | Both. Base yes; ICC 403s to automated checks. | Base: same sentence as above. IRC R328.3.1: "Individual units shall be separated from each other by not less than 3 feet, except where smaller separation distances are documented to be adequate based on large-scale fire testing" (*secondary*, a city ESS handout quoting the code; UL 9540A is the large-scale fire test method that exception refers to). | Base edited 2026-07-29; IRC 2021; SEAC bulletin 2021-11-30 ties the listing rule to UL 9540 | Partly supported. The 3 ft is verbatim; "unless UL 9540A testing allows closer" paraphrases the fire-testing exception rather than quoting it. |
| `clearances.opening_ft` = 3.0 ft | IRC 2021 R328.4 | Loads in a browser; 403 to automated checks | R328.4, item 3: "Outdoors or on the exterior side of exterior walls. Energy storage systems shall be located not less than 3 feet from doors and windows directly entering the dwelling unit." (*Secondary*; ICC's own viewer titles R328.4 "Locations".) | 2021 IRC | Verified. The wording matches the citation, including "directly entering the dwelling". Garage doors are not named in the code text, which the citation already says. |
| `facing.min_ft` = 3.0 ft | Base help page | Yes | "3 feet of clearance from gas meters, AC units, fences, other batteries, or other obstructions" (photo-checklist article 10280641). | Edited 2026-07-29 | Verified on the value. The citation should name which help page. |
| `headroom.min_ft` = 6.5 ft (placeholder) | NEC 110.26(A)(3), applied to the battery as a demo choice | Loads in a browser; 403 to automated checks | "Section 110.26(A)(3) 'Height of Working Space' requires the height of the working space to be at least 6 feet 6 inches" (*secondary*). | NEC; value unchanged across 2017 to 2023 editions | Verified as a citation. The battery application is a demo choice, which the placeholder flag already says. |
| `meter_working_space.width_ft` = 2.5 ft | NEC 110.26(A)(2) | Loads in a browser; 403 to automated checks | The working space is "at least 30 inches in width or the width of the equipment whichever is greater" (*secondary*). | As above | Verified. 2.5 ft = 30 in. Base's own help page says "30 in high x 36 inches wide", which this file already calls a garbled version. |
| `meter_working_space.depth_ft` = 3.0 ft | NEC 110.26(A)(1) | Loads in a browser; 403 to automated checks | Table 110.26(A)(1): for 0 to 150 volts to ground, depth is 900 mm (3 ft) in every condition (*secondary*). | As above | Verified. 3.0 ft = 36 in. |
| `route.max_ft` = 20.0 ft | Base help page | Yes | "Should be installed within 20 feet of the electrical meter." | Edited 2026-09-30 | Verified. |

### Values citing the repo's own estimates

| rules.yaml value | Cited source | Source reachable | What the source says | Edition or date | Status |
| --- | --- | --- | --- | --- | --- |
| `battery.width_ft` = 2.583333 ft | Base Core spec via this file's Hardware bullet, which cites pv magazine | Yes | The article gives "dimensions of 39.5 by 30.68 by 22 inches" (height by width by depth). *Secondary*: press coverage of Base's spec. | pv magazine, 2026-08-04 | Verified. The source string rounds 30.68 in up to 31 in; 31/12 = 2.583333 ft. |
| `battery.depth_ft` = 1.833333 ft | Same | Yes | 22 in, same sentence. | Same | Verified. 22/12 = 1.833333 ft. |
| `battery.height_ft` = 3.291667 ft | Same | Yes | 39.5 in, same sentence. | Same | Verified. 39.5/12 = 3.291667 ft. |
| `errors.tap_ft` = 0.3 ft | docs/00, Conventions | In repo | "0.3 ft for an AR tap", described as "day-1 estimates, untested". | docs/00 on main | Verified. |
| `errors.vlm_ft` = 1.5 ft | docs/00, Conventions | In repo | "1.5 ft for a position from photo detection". | Same | Verified. |
| `errors.mesh_ft` = 0.5 ft | docs/00, Conventions | In repo | "0.5 ft for the LiDAR mesh". | Same | Verified. |
| `errors.plane_ft` = 0.75 ft (placeholder) | None; untested estimate | n/a | server/README repeats "plane 0.75, an untested estimate". | server/README on main | Consistent; no public source claimed. |
| `errors.tape_ft` = 0.05 ft (placeholder) | "Tape measure read to the nearest half inch" | n/a | 0.05 ft is 0.6 in; half an inch is 0.0417 ft. | n/a | Partly supported by its own wording: the value rounds the half-inch resolution up. |
| `errors.wall_ft` = 0.3 ft | docs/00, Conventions | In repo | Walls come from AR taps, estimated at 0.3 ft. | Same | Verified. |
| `errors.meter_ft` = 0.3 ft | docs/00, Conventions | In repo | The meter is an AR tap, same estimate. | Same | Verified. |
| `errors.drift_per_ft` = 0.16 ft/ft | S1 real-data evals (PR #12) | In repo; PR merged | The ADVIO table reports ARKit-versus-ARCore distance disagreement at median/p90 2.8/8.2 in after 3 ft, 8.6/22.3 after 10 ft and 25.7/60.1 after 30 ft, on a 2018 iPhone 6s running ARKit 1.0. The margins in the rules comment (0.3 ft + 0.16 ft/ft = 0.78, 1.9 and 5.1 ft) cover those p90 values. | PR #12 | Partly supported. The numbers exist, but PR #12 calls them "a disagreement between two trackers, not a bound on ARKit's error". Its position error against ADVIO's truth runs p90 18.6 to 132.6 in, two to three times the 0.16 ft/ft allowance, and its scale error reads 5 to 17% short, not "mostly ~7%". |
| `sweep.step_ft` = 0.166667 ft | docs/00, Decisions | In repo | The sweep "slides the battery's footprint along that line in 2 in steps". | Same | Verified. 2/12 = 0.166667 ft. |
| `sweep.wall_join_ft` = 0.6 ft (placeholder) | "Twice the AR tap error estimate" | n/a | 2 x 0.3 = 0.6. | n/a | Consistent with `tap_ft`. |
| `sweep.meter_to_wall_max_ft` = 3.0 ft (placeholder) | "Input sanity bound" | n/a | No external claim. | n/a | Nothing to check. |
| `clearances.drive_ft` = 5.0 ft (placeholder) | "No public value (docs/04, Public rule values)" | In repo | This file's table lists it as a demo placeholder. | Same | Verified as claimed: no public source. |
| `clearances.pool_ft` = 10.0 ft (placeholder) | "No public value (docs/04, Public rule values)" | In repo | Same. | Same | Verified as claimed. |
| `clearances.wall_equipment_ft` = 0.0 ft (placeholder) | "No public value; 0 means only overlap fails" | n/a | No public claim. | n/a | Nothing to check. |
| `route.confident_reach_ft` = 15.0 ft (placeholder) | "No public value" | n/a | No public claim. | n/a | Verified as claimed. |
| `route.height_ft` = 1.0 ft (placeholder) | "Cable run height above ground" | n/a | No public claim. Base's "within 1 foot of the wall" is a different 1 ft and is not cited here. | n/a | Nothing to check. |
| `route.corner_allowance_ft` = 0.5 ft (placeholder) | "Day-1 estimate: a little per corner; untested" | n/a | No public claim. | n/a | Nothing to check. |
| `ground.allowed` = concrete, gravel, lawn, mulch; `ground.drivable` = drive; `ground.source` (placeholder) | `schemas/scene.schema.json` | In repo | The schema's ground-type enum is drive, concrete, gravel, lawn, mulch, deck. The allowed list picks four of six; the drivable list picks drive. | Schema on main | Verified. "Which surfaces are allowed is a demo choice" is accurate. |

### Flagged

1. The AC, fence and other-battery clearance comes from the photo-checklist help article (10280641); the placement article (10280705) carries only the gas-meter 3 ft, the 20 ft meter distance and the 1 ft wall distance. `ac_ft`, `battery_ft` and `facing.min_ft` say just "Base help page". Name the article.
2. The Hardware bullet credits the battery with "an 11 kW inverter" from the pv magazine article, and calls its listing "a UL listing". The article reports the dimensions and says Base "certified [it] under relevant UL and IEEE safety standards"; it does not give an inverter rating or name a specific standard. Trim or re-cite.
3. Austin Energy §1.9.2(C)(4) and the Texas Gas sheet are meter-siting rules: electric meters and outlets stay 3 ft from gas equipment. Neither governs batteries. The gas_ft row works by analogy, the way `headroom.min_ft` already declares it does.
4. IRC R328.3.1's exception is "large-scale fire testing", which the publicly readable copy (a California-edition handout) ties to CFC 1207.1.5 rather than naming UL 9540A. The substance is the same; the wording in `battery_ft` paraphrases.
5. `drift_per_ft`'s citation quotes p90s of tracker disagreement and calls the scale error "mostly ~7%", where PR #12 measured 5 to 17% and found position error against truth at 2 to 3 times the allowance. Tighten the wording so the number and its context match.
6. `tape_ft` = 0.05 ft rounds the stated half-inch resolution (0.0417 ft) up. Use 0.0417 ft, or say the value is rounded.
7. The two ICC viewer links are the right canonical pointers, but their text is not machine-readable and the checker cannot see it. This file's code quotes were confirmed from public secondary copies; the links stay.

All 16 URLs in this file were link-checked on 2026-10-09 with markdown-link-check and browser headers: 11 answered normally and 5 answered 403 (Esri's blog and the four Roboflow Universe pages); a first pass without browser headers also flagged the two ICC viewer pages. Each of the flagged pages loads in a browser or through a crawler, so the 403s are bot blocking, not dead links. The Roboflow dataset sizes match the counts quoted in this file (402, 1.4k and 1.9k images).

## How others do it

- **Tesla Powerwall** runs a guided self-survey on the phone of about 30 minutes, then designs remotely from the uploaded photos.
Expand Down