Test one exact data-item target and its recovery path Use a test sample with an existing parameter list and two assigned replicates. Establish the installed service/action contract with the LabVantage owner before any writes. The public action references below are a starting point for that review, not proof that a particular action is exposed by your service or behaves identically in every release. FIRST RUN 1. Identify the enabled operation and its read-back route Record the LabVantage environment, permitted service operation, action name/version if the wrapper exposes actions, and the authorized service identity. Choose a supported query or source view that can retrieve the target and its current value/status before and after the test. Check: The team can point to the actual configured operation and verify its inputs. A business-action name in documentation is not itself an HTTP endpoint. Keep result entry, review and release operations individually identified. LabVantage: LIMS web services and business logic: https://www.labvantage.com/a-quick-guide-to-lims-web-services/ LabVantage: sample lifecycle and review: https://www.labvantage.com/services/customer-success/training/end-user-lims-training-course-syllabus/ 2. Resolve the two targets before entering values Use the identity map below to capture both glucose replicates for LV-S910. Check the parameter-list version and variant against the assigned test definition, then record the unit configured for that measurement. Keep an untouched third dataset or replicate available as a control if practical. Check: The initial read shows two distinct intended targets with the agreed unit. The proposed values 1200 and 1190 belong to different replicates. Reject an unresolved identity instead of filling whichever row happens to be available. 3. Apply an explicit policy for existing and released targets Inspect the selected action’s released-data and missing-object options with the source owner, using the distinctions described below. For this fixed-target exercise, choose a policy that prevents changing released data and does not create another sample, dataset or replicate to recover a missing response. Check: Save the effective option values from the service wrapper, not just the caller’s intended values. Test the policy on a released disposable target and an absent target. Neither should silently acquire a new result under this evaluation policy. 4. Compare entered text with the source’s resulting display After one permitted entry, retrieve the resulting value and associated state through the agreed source route. LabVantage’s EnterDataSet reference explains that result entry can also affect transformed, qualified and displayed values. Capture the applicable calculation/formatting context for the target. Check: Do not compare only the number sent with a formatted report. In a separate fictional rounding test, entering 1.234 may display as 1.23 under the configured rule; retain what was entered and explain the display transformation rather than treating them as different measurements. LabVantage: EnterDataSet and transformed/display values: https://vantagecare.labvantage.com/labvantagedoc/Content/actions/sapphire/action/EnterDataSet.html 5. Interrupt the exchange and reconcile each intended row Use a test harness or saved fixtures to represent a response lost after one result is written. Query both complete target identities before deciding what to resend. Record delivery attempt, expected value, observed value, unit and disposition for each row. Check: A target with matching destination evidence is recovered as matched. A target with a different value is a conflict for investigation. A target whose read fails remains unresolved. Only a confirmed missing eligible target proceeds through the agreed recovery operation. Use three recovery outcomes instead of a single retry flag These are proposed reconciliation decisions for the worked sample. A transport response, scientific value and laboratory release state remain separate evidence. Read-back evidence: Exact target and expected value/unit match Decision: Retain the existing data item; do not create another replicate Recorded outcome: Matched, with target identity and source observation. Read-back evidence: Exact target exists with a different value Decision: Investigate source version, mapping and prior entry before any overwrite Recorded outcome: Conflict, preserving both the expected and observed values. Read-back evidence: No eligible target is confirmed Decision: Resolve missing assignment or release eligibility through the source workflow Recorded outcome: Blocked for target resolution, rather than auto-creating a scientific repeat. Read-back evidence: Read-back request fails or lacks identity Decision: Do not infer that the earlier write failed Recorded outcome: Unresolved; retain the failed lookup and original attempt. Read-back evidence: Matching numeric value but wrong replicate Decision: Reject it as recovery evidence for the intended row Recorded outcome: The requested target remains unresolved despite the coincidental numeric match. LabVantage: LIMS web services and business logic: https://www.labvantage.com/a-quick-guide-to-lims-web-services/ Two replicates with an interrupted acknowledgment Fictional reconciliation worksheet using the documented identity concepts. The shared test definition is GLUCOSE, version 3, variant Routine, parameter Glucose, type Standard; the confirmed unit is mg/L. These are fixture observations, not the output of a live LabVantage execution. sample,dataset,replicate,expected_text,observed_text,observation LV-S910,1,1,1200,1200,Matched LV-S910,1,2,1190,,Unresolved The blank observation for replicate 2 means its value has not been established by read-back. It does not mean zero or prove absence. If a later query finds 1190 on that exact target, both rows can be matched without another write. If it finds 1180, retain the conflict rather than overwriting it during recovery. PROPOSED RECEIVING CHECKS — NOT EXECUTED CUSTOMER VALIDATION Recover a lost response without sending the value again Input: Start with the worksheet above, then provide read-back evidence of 1190 mg/L for dataset 1, replicate 2 under the same test definition. Expected: Both intended results become matched. There are still two scientific replicates and no third replicate created by recovery. Find a conflicting value on the correct target Input: Return 1180 mg/L for the intended second replicate instead of 1190. Expected: The recovery item records expected 1190 and observed 1180 with the same complete target identity. It remains a conflict for review and does not silently replace the observed result. Return the expected number on a different dataset Input: Return 1190 mg/L for dataset 2 while the pending target is dataset 1, replicate 2. Expected: That observation does not resolve the pending row. The later dataset remains separate source evidence; the original target still needs a valid lookup. Replay against released or missing targets Input: Test the effective wrapper policy against a released data item and, separately, a target that has not been assigned. Expected: Under this evaluation’s explicit policy, released content is unchanged and absent targets do not trigger additional sample/dataset/replicate creation. Each failure retains its target identity and actionable reason. HANDOFF Provide the deployed service/action contract, the full two-target identity map, source unit and calculation rules, effective option settings and the before/after read-back evidence. This lets Seal represent an interrupted exchange as specific matched, conflicting or unresolved rows that neil can explain from the source evidence.