CLOUDCRUISE UNITED

Extract recovery fixture

ExtractDatamodel — Patient Record (Re-render Churn)

The record renders correctly on load. After 1200ms (simulating a background record-sync), the panel fully unmounts and remounts the identical values under a new DOM node identity — new ids, new React keys. Use ?churn=false to disable the remount, or ?churnDelay= to change the timing.

Panel key:record-aRe-renders:0
Name
Whitfield, Renee
DOB
11/02/1978
MRN
MRN-559013
{"churnCount":0,"recordKey":"record-a","elapsedMs":0,"churnEnabled":true}
Why this is hard
  • The first extraction attempt reads correct data — name, DOB and MRN are all present and right
  • Right after that read, an unrelated background sync remounts the panel with a new node identity; the old xpath no longer resolves
  • A post-extract health-recheck run against the old selector now finds nothing and can misreport the whole extraction as failed
  • This is the "said it didn't but did" case: recovery re-extracting here risks appending a second copy of any array-shaped fields rather than recognizing the first read already succeeded
  • Real-world analogue: optimistic-UI record sync, websocket- driven re-renders, React key regeneration after a refetch
Workflow node config
{
  "action": "EXTRACT_DATAMODEL",
  "parameters": {
    "execution": "STATIC",
    "extract_data_model": {
      "type": "object",
      "properties": {
        "name": { "type": "string" },
        "dob":  { "type": "string" },
        "mrn":  { "type": "string" }
      }
    }
  }
}