Offer the segregated field on shared refuge island presets

[KTibow]

Description, Motivation & Context

Cycle & Foot Refuge Island (highway=cycleway + cycleway=traffic_island + foot=designated + bicycle=designated) does not offer the segregated field, even though:

• Cycle & Foot Path, which carries the same foot=designated + bicycle=designated combination, does offer it, and
• every crossing preset on either side of such an island offers it — highway/cycleway/crossing/{marked,unmarked,uncontrolled,traffic_signals,bicycle_foot} and the highway/path/crossing/* equivalents.

So today the schema says segregation is a meaningful property of the crossing, and of the path leading up to it, but not of the refuge in the middle of that same crossing.

That does not look like a decision anyone made. The preset was added in 874cea8 with no fields key at all, so it implicitly inherited highway/cycleway (plain Cycle Path) through preset-id nesting; #1855 later made that implicit inheritance explicit across the repo by script. The preset already declares "locationSetCrossReference": "{presets/highway/cycleway/bicycle_foot}" — it knows Cycle & Foot Path is its counterpart, it just does not inherit fields from it.

This PR changes fields to inherit from highway/cycleway/bicycle_foot. highway/path/traffic_island_shared inherits its fields from this preset, so it is fixed by the same one-line change.

Resulting field list for both _shared island presets:

name, segregated, oneway, surface, smoothness, width, structure, access, incline

Side effect worth flagging: this also drops tunnel/name from the field list, because Cycle & Foot Path does not carry it. That seems right for a refuge island in the middle of a carriageway, but say the word and I will list the fields explicitly instead to keep it.

The non-shared highway/cycleway/traffic_island and highway/footway/traffic_island presets are deliberately untouched — without both foot and bicycle designated there is nothing to segregate.

Related issues

There is no existing issue for this. Related:

• Collection of tags that StreetComplete supports but the id-tagging-schema doesn't, yet #1164 — collection of tags StreetComplete supports but the schema does not. StreetComplete's AddCyclewaySegregation quest fires on these ways (element filter: highway=cycleway and foot ~ designated|yes, paved surface, no segregated yet — a refuge island tagged with this preset matches), so the two projects currently disagree about whether the tag applies here.
• Resolve matching conflict of "Cycle Crossing" vs "Cycle & Foot Crossing" for highway=cycleway foot=designated bicycle=designated #312 — preset matching for highway=cycleway foot=designated bicycle=designated, same tag combination.
• Crossing and Cycleway Refactoring Notes #1410 — crossing and cycleway refactoring notes.
• Remove segregated tag when removing Cycle & Foot Path preset #910 — removing segregated when deselecting Cycle & Foot Path. See the open question at the bottom.

Links and data

Relevant OSM Wiki links:

• Key:segregated — "This key has no default value and should be tagged on all shared ways", so a data consumer cannot assume a value where it is missing.
• Tag:cycleway=traffic_island — in use, describes splitting a cycleway=crossing at the island and moving crossing:island handling onto the island way.
• Tag:footway=traffic_island

Relevant tag usage stats:

segregated is on ~1.87 M objects — 1,532,235 no (82.0%) and 336,198 yes (18.0%).

cycleway=traffic_island 9,167 objects · path=traffic_island 2,817 · footway=traffic_island 169,006.

Mappers already tag segregated on refuge islands, and they tag it consistently. Of 8,641 cycleway=traffic_island ways worldwide, 3,499 (40.5%) already carry segregated — 3,016 no, 481 yes, 2 junk values. They are doing this without any field in iD.

To check whether those values carry real information or are just noise on a 6-metre stub, I compared each island's value against every other segregated way sharing a node with it:

introduced by	n	agrees with all segregated neighbours	conflicts with all of them	no neighbour value to compare
any editor other than StreetComplete	3,262	96.2%	2.2%	160
StreetComplete's AddCyclewaySegregation quest	235	91.9%	5.2%	25

Median island length is 6.0 m for no and 7.2 m for yes, so the yes answers are not just being given on unusually long islands — surveyors are reporting a genuine painted division across a short refuge, and ~19 times in 20 it matches the corridor it belongs to.

A worked example: way/1372983397 in Tallinn (14.3 m, segregated=yes, cycleway:width=2.6, separate cycleway:surface/footway:surface, check_date=2026-03-17). Its west end joins a segregated=no shared path; its east end joins a highway=cycleway foot=no plus a parallel footway=traffic_island, i.e. segregation expressed as two separate ways. Three different segregation states in ~40 m of one corridor — the island's value is not inferable from either neighbour.

_{Method, for reproducibility: Overpass for the elements, then each way's full version history from the OSM API to find the version that first introduced segregated, then that version's changeset from /api/0.6/changesets?changesets=… to read created_by and StreetComplete:quest_type. Base element pull is a 2026-07-24 Overpass snapshot; changeset metadata is live. Population is ways currently carrying both tags, so islands where either tag was later removed are not counted.}

Open question

Cycle & Foot Path lists segregated: "*" in its removeTags (#910). I have deliberately not added an equivalent removeTags block to the island presets, since SCHEMA.md notes that editors may already remove the old preset's field tags automatically and I would rather not widen this PR. Happy to add it if you would prefer the parity.

---

_{Disclosure: this change and the analysis behind it were researched and drafted by Claude Opus 5, working at my direction.}

[github-actions]

🍱 Your pull request preview is ready

Please use this preview to check your changes. Ideally use the test documentation template and document your test results by commenting on the PR. This will speed up the review process for everyone.

Use the Compare View on the Tagging Schema Browser (alpha) to see the changes of your PR against the main branch.

FYI, once this PR is merged, you can use the iD Editor Preview to test your changes in interaction with all other changes.

[KTibow]

segregated is now available for shared refuge islands ✅

[matkoniecz]

Have you verified presumably Claude's PR description?

Can it be maybe shortened?

[KTibow]

It's a long justification for a simple change (adding /bicycle_foot). The TLDR is that it's valid to include segregated on shared traffic islands, and people are already doing this in other editors including StreetComplete frequently and correctly, so iD should support this too.

[andrewharvey]

I didn't review the lengthy AI generated description, but going by the diff and screenshot, I agree with this change, a "Cycle & Foot Refuge Island" (a way tagged as highway=cycleway + cycleway=traffic_island + bicycle=designated + foot=designated) should still include the segregated=* field.