sqlmc_list_coverage: Result
One shape for both halves of the call. A tool has one output schema, and two tools would let a callerreach the second without ever seeing the first.
One shape for both halves of the call. A tool has one output schema, and two tools would let a caller reach the second without ever seeing the first.
confirmationRequired says which half to read: true means nothing was connected to and the
counts describe what a scan would do; false means they describe what it did. The plan fields
are filled in either way, because a caller reading a finished scan still wants to know how many
databases it covered and whether any was production.
| Field | |
|---|---|
ok |
A refused plan is not ok. A scan that was stopped is |
errors[] |
{ code, name, path } |
listName |
The target group |
confirmationRequired |
True when this is a plan awaiting agreement |
confirmationToken |
Pass back to run it. Null on a finished scan |
targetCount |
How many databases the scan covers |
alreadyCached |
How many have a schema on disk. Not a shortcut — a scan re-reads all of them, because an answer assembled from a three-week-old copy is not an observation. It says how much of the work is genuinely new |
productionTargets[] |
{ connection, database, isProduction } — named rather than counted |
scanned |
How many were read successfully. Zero on a plan |
failed |
How many could not be read |
cancelled |
Whether it stopped early. What it gathered is still reported |
findings[] |
{ schema, name, column, drift: { missingFrom, scanned, typeVariants[] } } |