Data (4 of 4)
9 specifications, 92 requirements. All libraries
0056 sql-drivers
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/drivers.test.ts: each adapter runs the same script and gives the same rows on the real client: node:sqlite, DuckDB, PGlite and pg over a socket, and bun:sqlite and Bun's SQL on Bun; and driverFor picks the adapter from the shape of the client |
FR-002 | check | bun run verify | suite/drivers.test.ts: each adapter runs the same script and gives the same rows on the real client: node:sqlite, DuckDB, PGlite and pg over a socket, and bun:sqlite and Bun's SQL on Bun; and driverFor picks the adapter from the shape of the client |
FR-003 | check | bun run verify | suite/drivers.test.ts: a client of no known shape is refused with a message that says what is looked for, and a row is made plain: a wide integer a number, an object with no prototype an ordinary one |
0057 sql-describe
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/describe.test.ts: a model is described in three layers, one described model for each module: governance (module, concept, attribute, relationship, code list, code), logical (table, column, keys, indexes) and physical (the SQL of each dialect, the typed table), each element with a stable id |
FR-002 | check | bun run verify | suite/describe.test.ts: a model is described in three layers, one described model for each module: governance (module, concept, attribute, relationship, code list, code), logical (table, column, keys, indexes) and physical (the SQL of each dialect, the typed table), each element with a stable id |
FR-003 | check | bun run verify | suite/describe.test.ts: a model is described in three layers, one described model for each module: governance (module, concept, attribute, relationship, code list, code), logical (table, column, keys, indexes) and physical (the SQL of each dialect, the typed table), each element with a stable id |
FR-004 | check | bun run verify | suite/describe.test.ts: every element of one layer is traced to what it is made into, every end of a trace is an element, and the typed tables the physical layer names exist in the generated files |
FR-005 | check | bun run verify | suite/describe.test.ts: the same models give the same bytes, as JSON, and the source of an element is what the caller says of it |
FR-006 | check | bun run verify | suite/describe.test.ts: what has no definition, an identity or a purpose is a finding that names the element; the real models have none |
0058 sql-osquery
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/osquery.test.ts: the published schema of a version is read, never written by hand: the file says where it came from (the repository, the commit, the SHA-256 of the file it was trimmed from, the licence), the tool that makes it keeps only the fields the importer reads, and a file that is not osquery's is refused |
FR-002 | check | bun run verify | suite/osquery.test.ts: the schema is imported through adopt into a logical schema that is stamped with the version, carries the platforms of each table and of the columns that exist on fewer, and makes the same tables on each dialect; a platform leaves out what is not there |
FR-003 | check | bun run verify | suite/osquery.test.ts: the columns that hold personal data are marked with the DPV-PD category and what they are held for, from a table of rules; the other columns are not |
FR-004 | check | bun run verify | suite/osquery.test.ts: only a SELECT of one statement is run: anything that writes, more than one statement, an empty one and what is not a SELECT are refused as error values before a process is started, words in strings and comments are not code, and a table that acts on the world is refused |
FR-005 | check | bun run verify | suite/osquery.test.ts: only a SELECT of one statement is run: anything that writes, more than one statement, an empty one and what is not a SELECT are refused as error values before a process is started, words in strings and comments are not code, and a table that acts on the world is refused |
FR-006 | check | bun run verify | suite/osquery.test.ts: a SELECT runs as `osqueryi --json`, its rows have the types of the schema, a table that is not on the platform is a typed error that says where it is, and what osquery says (a missing table, a failure, output that is not JSON) is an error value; suite/osquery.test.ts: where osqueryi is installed it is run for real: the machine's own users are rows of the types the schema gives |
FR-007 | check | bun run verify | suite/federate.test.ts: machines asked through osquery are targets: a table that is on one platform and not another is a typed error for the machine that lacks it and rows for the one that has it, the merge has both, and personal columns can be hidden |
FR-008 | check | bun run verify | suite/osquery.test.ts: the schema is made into a catalog for a platform: a table of the platform is there with its columns, one that the platform lacks is not |
0059 sql-federate
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/federate.test.ts: one query is asked of every target, each answer is a value in the order of the plan (rows with when they were collected, or a typed error), a target that fails is a line and not the end of the run, and a result is partial by default |
FR-002 | check | bun run verify | suite/federate.test.ts: one query is asked of every target, each answer is a value in the order of the plan (rows with when they were collected, or a typed error), a target that fails is a line and not the end of the run, and a result is partial by default |
FR-003 | check | bun run verify | suite/federate.test.ts: a quorum says how many targets must answer: below it the result says so and still has what answered |
FR-004 | check | bun run verify | suite/federate.test.ts: a target that is too slow is a timeout, no more than the concurrency are asked at once, a target that gives more rows than the plan allows is an error and its rows are not kept, and redaction changes a value before it is kept |
FR-005 | check | bun run verify | suite/federate.test.ts: the answers are loaded into DuckDB tagged with their target, with a column for each column that every answer agrees the type of (text where they do not), and the merge is SQL over fleet_rows and fleet_targets |
FR-006 | check | bun run verify | suite/federate.test.ts: the answers are loaded into DuckDB tagged with their target, with a column for each column that every answer agrees the type of (text where they do not), and the merge is SQL over fleet_rows and fleet_targets |
FR-007 | check | bun run verify | suite/federate.test.ts: the targets that differ are found as a query: from what most targets have, or from one target, as the rows they lack and the rows they have besides |
FR-008 | check | bun run verify | suite/federate.test.ts: the audit of a run is optional and the caller's: the record has who, what and what each target answered and no rows, a sink of the caller's gets it, and auditTable keeps it in a table of a database |
FR-009 | check | bun run verify | suite/federate.test.ts: machines asked through osquery are targets: a table that is on one platform and not another is a typed error for the machine that lacks it and rows for the one that has it, the merge has both, and personal columns can be hidden |
0060 flow-ledger
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/flow.test.ts: the flow model has the eight stages in order, a closed code list for each thing that is said of one, reasons as data, loads, the contract of a stage, a ledger that is only added to, the idempotent intake with its conflicts, and the manifest with its funnel; every concept says what it means |
FR-002 | check | bun run verify | suite/flow.test.ts: the flow model has the eight stages in order, a closed code list for each thing that is said of one, reasons as data, loads, the contract of a stage, a ledger that is only added to, the idempotent intake with its conflicts, and the manifest with its funnel; every concept says what it means |
FR-003 | check | bun run verify | suite/flow.test.ts: the flow model has the eight stages in order, a closed code list for each thing that is said of one, reasons as data, loads, the contract of a stage, a ledger that is only added to, the idempotent intake with its conflicts, and the manifest with its funnel; every concept says what it means |
FR-004 | check | bun run verify | suite/flow.test.ts: the flow model has the eight stages in order, a closed code list for each thing that is said of one, reasons as data, loads, the contract of a stage, a ledger that is only added to, the idempotent intake with its conflicts, and the manifest with its funnel; every concept says what it means |
FR-005 | check | bun run verify | suite/flow.test.ts: the flow model has the eight stages in order, a closed code list for each thing that is said of one, reasons as data, loads, the contract of a stage, a ledger that is only added to, the idempotent intake with its conflicts, and the manifest with its funnel; every concept says what it means |
FR-006 | check | bun run verify | suite/flow.test.ts: the flow model has the eight stages in order, a closed code list for each thing that is said of one, reasons as data, loads, the contract of a stage, a ledger that is only added to, the idempotent intake with its conflicts, and the manifest with its funnel; every concept says what it means |
FR-007 | check | bun run verify | suite/flow.test.ts: chapter 1: twenty messages through eight stages, each confirmation a row of the ledger with its load id; the last stage that confirmed each message, the parse failure, the destination's rejection, the message that vanished (the gap query returns exactly it), the one that was accepted and is not usable, and 1 + 1 + 1 + 1 + 16 = 20 |
FR-008 | check | bun run verify | suite/flow.test.ts: chapter 1: twenty messages through eight stages, each confirmation a row of the ledger with its load id; the last stage that confirmed each message, the parse failure, the destination's rejection, the message that vanished (the gap query returns exactly it), the one that was accepted and is not usable, and 1 + 1 + 1 + 1 + 16 = 20 |
FR-009 | check | bun run verify | suite/flow.test.ts: chapter 6: a thousand unique deliveries, fifty redeliveries and ten with the same id and other content give a thousand effects, delivery counts that total 1,050, a conflict queue of exactly ten with the originals intact, and effects by load id that sum to a thousand; two workers racing give one first |
FR-010 | check | bun run verify | suite/flow.test.ts: chapter 11: a manifest of 1,000 goes to 975 accepted with 6 parse failures, 13 contract failures and 6 destination rejections and nothing is unexplained; a run with 4 messages silently dropped between two stages has 4 unexplained, attributed to the boundary they went missing at; a batch whose duplicates are reason-coded reconciles at both grains |
FR-011 | check | bun run verify | suite/flow.test.ts: chapter 13: 5,000 bundles in 40 loads of 125, and a load undone by its id removes exactly its 125 rows from the table that carries its id and leaves the other 39 loads byte-identical by hash; the load is marked as undone and the ledger keeps what it confirmed |
0061 meaning-registers
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/meaning.test.ts: the meaning model has mappings with how they relate and when they hold, variances that expire, null rules, and closed code lists for absence, null policy, relationship, mapping status and variance status |
FR-002 | check | bun run verify | suite/meaning.test.ts: the meaning model has mappings with how they relate and when they hold, variances that expire, null rules, and closed code lists for absence, null policy, relationship, mapping status and variance status |
FR-003 | check | bun run verify | suite/meaning.test.ts: the meaning model has mappings with how they relate and when they hold, variances that expire, null rules, and closed code lists for absence, null policy, relationship, mapping status and variance status |
FR-004 | check | bun run verify | suite/meaning.test.ts: the meaning model has mappings with how they relate and when they hold, variances that expire, null rules, and closed code lists for absence, null policy, relationship, mapping status and variance status |
FR-005 | check | bun run verify | suite/meaning.test.ts: the meaning model has mappings with how they relate and when they hold, variances that expire, null rules, and closed code lists for absence, null policy, relationship, mapping status and variance status |
FR-006 | check | bun run verify | suite/meaning.test.ts: translate gives the approved mapping that holds at the time, nothing for a value no mapping covers or one that has ended or is only a draft, and says ambiguous where two hold; coverage names the unmapped values most seen first |
FR-007 | check | bun run verify | suite/meaning.test.ts: translate gives the approved mapping that holds at the time, nothing for a value no mapping covers or one that has ended or is only a draft, and says ambiguous where two hold; coverage names the unmapped values most seen first |
FR-008 | check | bun run verify | suite/meaning.test.ts: a null rule keeps a value that is there, rejects, keeps the kind of absence or replaces with a default as the policy says, and an empty string is as missing as a null |
FR-009 | check | bun run verify | suite/meaning.test.ts: the variance register finds what is accepted past its date and what is open, and leaves accepted ones that have not expired and resolved ones |
FR-010 | check | bun run verify | suite/meaning.test.ts: profile gives rows, nulls, distinct values and the lowest and highest of a column with counts, on every engine |
FR-011 | check | bun run verify | suite/meaning.test.ts: the generator makes the same rows for the same seed, other rows for another, the rate of absence and of duplicates asked for, and no value that is not synthetic |
0062 contracts-time-control
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/contract.test.ts: the contract model has contracts with their fields, periods, controls with their runs (only added to), consents and canary tokens; the contracts, periods and consents are annotated effective-dated with members that exist; every concept is defined |
FR-002 | check | bun run verify | suite/contract.test.ts: the contract model has contracts with their fields, periods, controls with their runs (only added to), consents and canary tokens; the contracts, periods and consents are annotated effective-dated with members that exist; every concept is defined |
FR-003 | check | bun run verify | suite/contract.test.ts: the contract model has contracts with their fields, periods, controls with their runs (only added to), consents and canary tokens; the contracts, periods and consents are annotated effective-dated with members that exist; every concept is defined |
FR-004 | check | bun run verify | suite/contract.test.ts: the contract model has contracts with their fields, periods, controls with their runs (only added to), consents and canary tokens; the contracts, periods and consents are annotated effective-dated with members that exist; every concept is defined |
FR-005 | check | bun run verify | suite/contract.test.ts: the contract model has contracts with their fields, periods, controls with their runs (only added to), consents and canary tokens; the contracts, periods and consents are annotated effective-dated with members that exist; every concept is defined |
FR-006 | check | bun run verify | suite/contract.test.ts: the contract model has contracts with their fields, periods, controls with their runs (only added to), consents and canary tokens; the contracts, periods and consents are annotated effective-dated with members that exist; every concept is defined |
FR-007 | check | bun run verify | suite/contract.test.ts: the contract model has contracts with their fields, periods, controls with their runs (only added to), consents and canary tokens; the contracts, periods and consents are annotated effective-dated with members that exist; every concept is defined |
FR-008 | check | bun run verify | suite/contract.test.ts: the compatibility matrix tells a breaking change to readers from one to writers: a removed field, a retyped one, one that became nullable or a label raised break readers; a required field added breaks writers; an optional field added breaks nobody |
FR-009 | check | bun run verify | suite/contract.test.ts: holdsAt reads a record as of an instant on every engine: it began at or before it and has no end or ends after it, and the end is not included |
FR-010 | check | bun run verify | suite/contract.test.ts: a closed period takes no change: a change dated in it is placed as an adjustment of the next open period, one in an open period stays in it, one in no period is placed nowhere, and reopening needs a reason |
FR-011 | check | bun run verify | suite/contract.test.ts: the control matrix computes passing, failing, waived, overdue and never-run from the latest run and the frequency of each control |
FR-012 | check | bun run verify | suite/contract.test.ts: consent is a test: a use with a consent in force for its purpose passes, and one outside its dates, for another purpose, withdrawn or with none is returned |
FR-013 | check | bun run verify | suite/contract.test.ts: a planted canary token is found in text outside the place that holds it and not in that place |
FR-014 | check | bun run verify | suite/contract.test.ts: a label is the highest of what a dataset is made from, however far back, and a derived dataset labelled lower is a violation that names what makes it so |
FR-015 | check | bun run verify | suite/contract.test.ts: the layer report says by table where the physical layer disagrees with the logical, in words, and says nothing where they agree |
0063 operating-model
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/operating-model.test.ts: the operations model has the waiting-on ledger, decision records that can replace one another, limits, the copy inventory, environments with releases and a promotion log that is only added to, workloads with their recorded statements, the signal catalog and the freshness rules, with closed code lists and every concept defined |
FR-002 | check | bun run verify | suite/operating-model.test.ts: the operations model has the waiting-on ledger, decision records that can replace one another, limits, the copy inventory, environments with releases and a promotion log that is only added to, workloads with their recorded statements, the signal catalog and the freshness rules, with closed code lists and every concept defined |
FR-003 | check | bun run verify | suite/operating-model.test.ts: the operations model has the waiting-on ledger, decision records that can replace one another, limits, the copy inventory, environments with releases and a promotion log that is only added to, workloads with their recorded statements, the signal catalog and the freshness rules, with closed code lists and every concept defined |
FR-004 | check | bun run verify | suite/operating-model.test.ts: the operations model has the waiting-on ledger, decision records that can replace one another, limits, the copy inventory, environments with releases and a promotion log that is only added to, workloads with their recorded statements, the signal catalog and the freshness rules, with closed code lists and every concept defined |
FR-005 | check | bun run verify | suite/operating-model.test.ts: the operations model has the waiting-on ledger, decision records that can replace one another, limits, the copy inventory, environments with releases and a promotion log that is only added to, workloads with their recorded statements, the signal catalog and the freshness rules, with closed code lists and every concept defined |
FR-006 | check | bun run verify | suite/operating-model.test.ts: the operations model has the waiting-on ledger, decision records that can replace one another, limits, the copy inventory, environments with releases and a promotion log that is only added to, workloads with their recorded statements, the signal catalog and the freshness rules, with closed code lists and every concept defined |
FR-007 | check | bun run verify | suite/operating-model.test.ts: the operations model has the waiting-on ledger, decision records that can replace one another, limits, the copy inventory, environments with releases and a promotion log that is only added to, workloads with their recorded statements, the signal catalog and the freshness rules, with closed code lists and every concept defined |
FR-008 | check | bun run verify | suite/operating-model.test.ts: the waiting-on ledger lists what is open, the oldest first, as overdue once its date has passed, and drops what has come |
FR-009 | check | bun run verify | suite/operating-model.test.ts: a decision that replaces another is accepted and says what it replaced, the replaced one is marked, and only accepted decisions are in force |
FR-010 | check | bun run verify | suite/operating-model.test.ts: limits are found when the time to act has come: expired ones and ones within the days of notice their owner needs, the soonest first, and not those further off |
FR-011 | check | bun run verify | suite/operating-model.test.ts: copies of data with no owner, never refreshed or not refreshed within the days given are found, and a copy that is owned and fresh is not |
FR-012 | check | bun run verify | suite/operating-model.test.ts: a release is promoted only in the order of the environments and to production only with evidence, a refused promotion writes nothing, parity says which environments are behind, and real data outside production is found |
FR-013 | check | bun run verify | suite/operating-model.test.ts: a release is promoted only in the order of the environments and to production only with evidence, a refused promotion writes nothing, parity says which environments are behind, and real data outside production is found |
FR-014 | check | bun run verify | suite/operating-model.test.ts: a signal is made only as the catalog defines it: an unknown name, other attributes or a counter that goes down are refused, and a sample carries its kind and unit |
FR-015 | check | bun run verify | suite/operating-model.test.ts: freshness is computed from the latest load of a source that was not undone: fresh within the hours, stale past them, never loaded where there is none |
FR-016 | check | bun run verify | suite/operating-model.test.ts: the recorded statements of a workload are replayed and compared with the rows and hash recorded: the same after no change, different after one, and anything but a single select is refused and not run |
0064 recovery-proof-identity
| Requirement | Mechanism | Who | Note |
|---|---|---|---|
FR-001 | check | bun run verify | suite/proof.test.ts: the proof model has drills (kept, only added to), calculation runs with the hash of their inputs and outputs, reference cases, lineage edges and entity links, with closed code lists and every concept defined |
FR-002 | check | bun run verify | suite/proof.test.ts: the proof model has drills (kept, only added to), calculation runs with the hash of their inputs and outputs, reference cases, lineage edges and entity links, with closed code lists and every concept defined |
FR-003 | check | bun run verify | suite/proof.test.ts: the proof model has drills (kept, only added to), calculation runs with the hash of their inputs and outputs, reference cases, lineage edges and entity links, with closed code lists and every concept defined |
FR-004 | check | bun run verify | suite/proof.test.ts: the proof model has drills (kept, only added to), calculation runs with the hash of their inputs and outputs, reference cases, lineage edges and entity links, with closed code lists and every concept defined |
FR-005 | check | bun run verify | suite/proof.test.ts: the proof model has drills (kept, only added to), calculation runs with the hash of their inputs and outputs, reference cases, lineage edges and entity links, with closed code lists and every concept defined |
FR-006 | check | bun run verify | suite/proof.test.ts: the proof model has drills (kept, only added to), calculation runs with the hash of their inputs and outputs, reference cases, lineage edges and entity links, with closed code lists and every concept defined |
FR-007 | check | bun run verify | suite/proof.test.ts: a rollback drill counts what a load wrote, undoes it by its id, shows that what was removed is what was written, that none of its rows remain and that the rows of the four other loads are as they were, and keeps the drill |
FR-008 | check | bun run verify | suite/proof.test.ts: a replay that is idempotent and a restore leave a table as it was and pass, one that changes it fails and says so, and each drill is kept with how long it took |
FR-009 | check | bun run verify | suite/proof.test.ts: a calculation is recorded with the hash of its inputs and outputs and reproduced from them: reproduced with the same inputs and version, and different-outputs after a change of rounding, different-inputs with other inputs, unknown for a run not recorded |
FR-010 | check | bun run verify | suite/proof.test.ts: reference cases hold a calculation to the value expected within the tolerance: a case within passes, one outside fails with the difference |
FR-011 | check | bun run verify | suite/proof.test.ts: lineage is walked upstream and downstream, nearest first, each dataset once and without looping on a cycle, and its edges feed the label check of the contract model |
FR-012 | check | bun run verify | suite/proof.test.ts: records of three sources are linked by rules tried in order (a key as it is, a key written one way, a similarity), each with the rule and score that joined it and a stable entity id, a record nothing links stays alone, and the links are kept |
FR-013 | check | bun run verify | suite/proof.test.ts: records of three sources are linked by rules tried in order (a key as it is, a key written one way, a similarity), each with the rule and score that joined it and a stable entity id, a record nothing links stays alone, and the links are kept |