fix(dive-computer): resolve CCR transmitter cylinders from every transmitter reading

[alpheios-one]

Related Issue

Closes #2318

Summary

On a Shearwater CCR dive with a diluent and an oxygen transmitter, both transmitters report in the same sample, but ProfileSample.pressureBar / tankIndex only hold the last reading, which on the Petrel 3 is always the oxygen transmitter. The resolver relied on that pair, so:

• the O2 cylinder was labeled with the breathed diluent (the "breathed gas" rule ran before the oxygen default from fix(dive-log): auto-assign 100% O2 when a cylinder's role is set to oxygen supply #1972),
• the diluent cylinder only got the right gas if the O2 transmitter happened to miss a sample, and otherwise fell back to the first gas mix (OC1 99/0),
• the bailout gas without a transmitter was imported as Back Gas,
• the diluent end pressure never got the surfacing trim from Read cylinder end pressure at surfacing, not at the end of the recording #1290 and kept the post-surfacing bleed-down (e.g. 13.4 bar instead of 77.6 bar).

All four were confirmed against the raw downloads of eight real dives and Shearwater Cloud (details in #2318).

Changes

All in parsed_tank_resolver.dart:

• Oxygen cylinder (usage oxygen): the computer's own link, else an oxygen-tagged gas mix, else the 100/0 default. The breathed-gas rule no longer applies to it.
• Diluent cylinder (usage diluent): the computer's own link, else the most-breathed diluent-tagged gas, else the first diluent-tagged gas. Without any diluent-tagged gas the previous rules still apply (keeps the fix(dive-log): auto-assign 100% O2 when a cylinder's role is set to oxygen supply #1972 test behavior).
• Surfacing trim: surfacing points read every transmitter from tankPressuresBar (same fallback rule as groupPressuresByTank), so each cylinder is trimmed at surfacing, not only the last one reported.
• Gases without a transmitter get the sensorless roles (usage tag, then the CCR bailout ranking from feat(dive-computer): rank sensorless CCR bailout gases against each other #1752). The bailout ranking compares against every untagged gas of the dive, including one on a transmitter, so a bailout cylinder with its own transmitter does not promote a leaner deco gas to Bailout. The breathed diluent is listed first among them, so the dive-level diluent (fix(dive-computer): derive the CCR diluent gas mix from the resolved Diluent cylinder #2034) is the one used when the diluent has no transmitter.
• _firstFreeIndex also accounts for tankPressuresBar, so synthesized cylinders never collide with a transmitter index.

Deliberately unchanged: the breathed-gas rule for untagged tanks still keys on the sample's own tankIndex. Crediting every reporting transmitter would hand each tank the dive's main gas and override a computer's own tank-to-gas link on open circuit dives. Attributing gas per transmitter properly (e.g. by pressure drop) is left as a possible follow-up.

How to Test

Needs a Shearwater CCR dive downloaded with two AI transmitters named D1/O2 (or an HP CCR setup).

1. Open Dive Computers, select the Shearwater, and use Re-parse all dives.
2. Open a CCR dive: the O2 cylinder shows 100/0, the diluent cylinder the breathed diluent, the open circuit bottom gas is Bailout and 99/0 is Deco.
3. The diluent cylinder's end pressure matches the value at surfacing (compare with Shearwater Cloud or the pressure graph), not the last reading after the dive.

Test Plan

• New parsed_tank_resolver_ccr_transmitters_test.dart models the real download (gas list OC1 99/0, OC5 15/55, DIL 15/55; both transmitters in every sample, O2 reported last) and failed before the fix with exactly the values seen in the field (O2 15/55, diluent 99/0, diluent end 13.4 bar)
• Guard tests: an open circuit tank keeps the computer's own gas link while another transmitter reports last; a CCR dive with only an O2 transmitter picks the breathed diluent over an unused programmed one; a bailout on its own transmitter keeps a 50% gas as Deco
• flutter test test/features/dive_computer passes (606 tests)
• flutter analyze clean on the changed areas, dart format clean
• Pre-push hook (format, analyze, affected tests) passed
• Manual testing: not done in this PR

Note: three tests under test/features/dive_log (dive_header_map_bottom_edge_test, deco_filter_providers_test, equipment_attr_filter_providers_test) fail locally on Windows. They do not import anything from dive_computer and are unrelated to this change.

[github-actions]

📦 Build artifacts for this PR · commit 9738c41

Platform	Download
Android (APK)	android-apk
macOS	macos-build
Windows	windows-build
Linux	linux-build

Artifacts expire in 7 days. Downloading requires being signed in to GitHub. macOS needs two extractions: unzip the downloaded artifact, then unzip the submersion-macos.zip inside it to get a runnable submersion.app. The build is ad-hoc signed — right-click → Open on first launch.

_{Updated automatically on each push.}

[codecov]

Codecov Report

❌ Patch coverage is 98.33333% with 1 line in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...e_computer/data/services/parsed_tank_resolver.dart	98.33%	1 Missing ⚠️

📢 Thoughts on this report? Let us know!

[Copilot]

Copilot review overview

🟡 Changes recommended

Preserve legacy single-transmitter readings when applying surfacing trim.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• Preserve legacy single pressure reading for tank 0 · New

What changed in this PR

Fixes CCR transmitter cylinder resolution, gas assignment, sensorless roles, and surfacing pressure trimming.

Changes:

• Resolves oxygen and diluent gases across transmitter readings.
• Corrects bailout/deco role inference and index allocation.
• Adds comprehensive CCR regression tests.

File	Summary
test/​features/​dive_computer/​data/​services/​parsed_tank_resolver_ccr_transmitters_test.dart	Adds CCR multi-transmitter regression coverage.
lib/​features/​dive_computer/​data/​services/​parsed_tank_resolver.dart	Updates cylinder, gas, role, pressure, and index resolution.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[alpheios-one]

@ericgriffin This is complete and ready for testing and review. All CI checks are green, the Copilot finding is fixed with a regression test and resolved, and the branch merges cleanly. The quickest check is "Re-parse all dives" on a Shearwater with CCR dives that use two transmitters (see How to Test in the description).

[JaspalX]

out of interest, how does the petrel know which transmitter is oxygen? it just gets a pressure, not a gas mix..so i don't think that assumption is safe, but I may be wrong ofc. I happen to dive o2 as my 3rd sensor (first two dil/bailout), but I could easily put the o2 pressure sensor as the first one.

[alpheios-one]

Good question. It doesn't rely on the transmitter's position. On a CCR dive, libdivecomputer's Shearwater parser tags each transmitter from the name the diver gives it on the Petrel: a name starting with "O" becomes oxygen, one starting with "D" becomes diluent (shearwater_predator_parser.c, DC_FIELD_TANK). In HP CCR mode the diluent and O2 channels are fixed by Shearwater. So an O2 sensor named e.g. "O2" is recognized on T1 just as well as on T3.

If the transmitters keep their default names (T1..T4), no usage is reported and this PR changes nothing for them: the gas falls back to the previous rules and the role comes from the transmitter registry.

libdivecomputer already reported this correctly. The bug was on the Submersion side:

1. The resolver checked which gas was "breathed while this transmitter reported" before looking at the transmitter's usage. On the loop the active gas is always the diluent, so the O2 cylinder was labeled with the diluent mix.
2. Our plugin wrapper keeps a single pressureBar/tankIndex pair per sample, which only holds the last transmitter reported, on a two-transmitter Petrel always the O2 one. The resolver and the surfacing trim read only that pair, although tankPressuresBar carries every transmitter. So the diluent cylinder fell back to the first gas mix, and its end pressure kept the post-surfacing reading.
3. Gases without a transmitter only got the open circuit role heuristic, ignoring the diluent tag and the CCR bailout ranking.

This PR fixes all three: oxygen and diluent cylinders resolve by their usage before the breathed-gas rule, the surfacing trim reads every transmitter, and gases without a transmitter get the same roles as on a sensorless dive.

[JaspalX]

ah, cool. i'd forgotten about the prefixes. thanks for explaining

[Copilot]

Copilot review overview

🟢 Approval recommended

No unresolved blocking issues were identified, and all approval assessments are positive.

Review effort: Lite
Findings: None

Resolved since last review (1)

• Preserve legacy single pressure reading for tank 0