Core Spec Test Coverage

Consumer / validator tests

• Explicitly not tested

  • Source map being an invalid JSON document (depending on the implementation, this may only test the test harness itself, rather than any interesting properties about the tool)

• Encoding

  • Base64 VLQ
    • VLQs should decode correctly
      • A VLQ with a non-base64 character will fail to decode. (test)
      • A VLQ with one digit and no continuation digits should decode.
      • A negative VLQ with the sign bit set to 1 should decode.
      • A VLQ with non-zero continuation bits (and more than one digit) should decode.
      • A VLQ with a non-zero continuation bit with no further digits should fail to decode.
      • A VLQ should decode with the correct order of digits (least to most significant).
      • A long VLQ with many trailing zero digits will decode.
    • A VLQ exceeding the 32-bit size limit is invalid (note: the spec is unclear on the details of this limit)
    • A VLQ at exactly the 32-bit size limit should be decoded (positive and negative).

• Basic format

  • version field
    • Must be present
    • Must be a number
    • Must be 3
  • file field
    • Optional, allow missing
    • Must be a string
  • sourceRoot field
    • Optional, allow missing
    • Must be a string
  • sources field
    • Must be present
    • Must be an array
    • Array elements must be null or a string
  • sourcesContent field
    • Must be present
    • Must be an array
    • Array elements must be null or a string
  • names field
    • Must be present (note: the spec implies this but implementations may not agree)
    • Must be an array
    • Array elements must be strings
  • mappings field
    • Must be present
    • Must be a string
    • Empty string is valid
  • ignoreList field: see below under "Ignore List"
  • Extra unrecognized fields are allowed

• Index maps

  • ? Must be mutually exclusive with non-index map?
  • file field
    • Optional, allow missing
    • Must be a string? (spec is not clear)
  • sections field
    • Must be present
    • Must be an array
    • An empty sections array is valid
    • Array elements are valid section objects
      • offset field
        • Must be present
        • line field
          • Must be present
          • Must be a number
        • column field
          • Must be present
          • Must be a number
      • map field
        • Must be present
        • Must be an object
        • Must be a valid source map
    • Sections are in order (the spec is not 100% clear, but assumption is increasing numeric order, may need subtests)
    • Sections are non-overlapping (the definition of overlap is not clear, may need subtests)

• Mappings format

  • Each line is separated by ";"
  • A line may consist of zero segments (e.g., ";;")
  • Each line consists only of segments separated by ","
  • Must have greater than zero fields (note: many implementations don't check)
  • Must have 1, 4, or 5 fields
  • The source index must not be out of bounds of the sources array
  • The name index must not be out of bounds of the names array
  • Absolute VLQ values must be non-negative
    • The column must be non-negative
    • The source index must be non-negative
    • The original line must be non-negative
    • The original column must be non-negative
    • The name index must be non-negative
  • Relative VLQ values must be non-negative after adding to previous value
    • The column must be non-negative
    • The source index must be non-negative
    • The original line must be non-negative
    • The original column must be non-negative
    • The name index must be non-negative

• Ignore list

  • An ignore list is optional, may be missing
  • An ignore list can't be a non-array value
    • An ignore list can be empty
    • An ignore list entry must be a non-negative integer
    • An ignore list entry cannot be out-of-bounds of the sources array
  • Ignore list entries are detected and are present
  • Items not specified in the ignore list don't show up as ignored

• Mappings semantics

  • A source map with no mappings does not map any position.
  • A single field segment gets mapped to the correct line and column.
  • A four field segment gets mapped to the correct line and column.
  • A five field segment gets mapped to the correct line and column.
  • When a name is present in a segment, it is correctly mapped.
  • When a source is present in a segment, it is correctly mapped.
  • The second occurrence of a field segment in a line is mapped relative to the previous one.
  • When a new line starts, the generated column field resets to zero rather than being relative to the previous line.
  • For fields other than the generated column, a segment field that has occurred once in a previous line is mapped relatively when it occurs in the next line.
  • Ensure that a transitive source map mapping works as expected
  • Index maps are correctly used in mappings
    • An index map with one sub-map will map correctly.
    • An index map with multiple sub-maps will map correctly, with appropriate offsets for the second and later sub-maps.

• Resolution of sources

  • When sourceRoot is provided, it is prepended to any sources entries and will be mapped with the full URL.
  • If the source URL is an absolute URL, it is resolved as an absolute URL.
  • If the source URL is a relative URL, it is resolved relative to the source map path.

• Wasm support

  • Create versions of the tests that use a Wasm source.

Proposals

Scopes proposal

• Basic format validation
  • originalScopes may be omitted (Q: must originalScopes and generatedRanges both be absent or present simultaneously?)
  • generatedRanges may be omitted
  • originalScopes must be an array (of strings)
    • The length of originalScopes should be less than or equal to the length of sources
    • Can entries be null?
  • generatedRanges must be a string