sql: Add unsigned int types (MaterializeInc#14304)

This commit adds 16-bit, 32-bit, and 64-bit unsigned integer types to
Materialize. In addition it also adds various casts to and from the
unsigned int types.

The binary format of these types is big endian, which is the same as
every other numeric type. These types have separate OIDs from the
signed integer types so psql and other drivers don't confuse these for
signed integers.

No arithmetic operators are added for unsigned integers in this commit.
Those can be added in a follow up commit.

Works towards resolving #7629

doc/user/content/sql/types/_index.md

@@ -18,7 +18,7 @@ Materialize's type system consists of two classes of types:
 ## Built-in types
 
 Type | Aliases | Use | Size (bytes) | Catalog name | Syntax
------|---------|-----|--------------|----------------|-----
+-----|-------|-----|--------------|----------------|-----
 [`bigint`](integer) | `int8` | Large signed integer | 8 | Named | `123`
 [`boolean`](boolean) | `bool` | State of `TRUE` or `FALSE` | 1 | Named | `TRUE`, `FALSE`
 [`bytea`](bytea) | `bytea` | Unicode string | Variable | Named | `'\xDEADBEEF'` or `'\\000'`
@@ -33,8 +33,12 @@ Type | Aliases | Use | Size (bytes) | Catalog name | Syntax
 [`oid`](oid) | | PostgreSQL object identifier | 4 | Named | `123`
 [`real`](float) | `float4` | Single precision floating-point number | 4 | Named | `1.23`
 [`record`](record) | | Tuple with arbitrary contents | Variable | Unnameable | `ROW($expr, ...)`
+[`smallint`](integer) | `int2` | Small signed integer | 2 | Named | `123`
 [`text`](text) | `string` | Unicode string | Variable | Named | `'foo'`
 [`time`](time) | | Time without date | 4 | Named | `TIME '01:23:45'`
+[`uint2`](uint) | | Small unsigned integer | 2 | Named | `123`
+[`uint4`](uint) | `uint` | Unsigned integer | 4 | Named | `123`
+[`uint8`](uint) | | Large unsigned integer | 8 | Named | `123`
 [`timestamp`](timestamp) | | Date and time | 8 | Named | `TIMESTAMP '2007-02-01 15:04:05'`
 [`timestamp with time zone`](timestamp) | `timestamp with time zone` | Date and time with timezone | 8 | Named | `TIMESTAMPTZ '2007-02-01 15:04:05+06'`
 [Arrays](array) (`[]`) | | Multidimensional array | Variable | Named | `ARRAY[...]`

doc/user/content/sql/types/integer.md

@@ -72,6 +72,7 @@ To | Required context
 [`oid`](../oid) | Implicit
 [`real`/`double precision`](../float) | Implicit
 [`text`](../text) | Assignment
+[`uint2`/`uint4`/`uint8`](../uint) | Depends on specific cast
 
 #### To `integer` or `bigint`
 
@@ -85,6 +86,7 @@ From | Required context
 [`numeric`](../numeric) | Assignment
 [`real`/`double precision`](../float) | Assignment
 [`text`](../text) | Explicit
+[`uint2`/`uint4`/`uint8`](../uint) | Depends on specific cast
 
 ## Examples
 

doc/user/content/sql/types/uint.md

@@ -0,0 +1,103 @@
+---
+title: "Unsigned Integer types"
+description: "Express unsigned integers"
+menu:
+  main:
+    parent: 'sql-types'
+aliases:
+  - /sql/types/uint2
+  - /sql/types/uint4
+  - /sql/types/uint8
+---
+
+## `uint2` info
+
+Detail | Info
+-------|------
+**Size** | 2 bytes
+**Catalog name** | `mz_catalog.uint2`
+**OID** | 16,460
+**Range** | [0, 65,535]
+
+## `uint4` info
+
+Detail | Info
+-------|------
+**Size** | 4 bytes
+**Catalog name** | `mz_catalog.uint4`
+**OID** | 16,462
+**Range** | [0, 4,294,967,295]
+
+## `uint8` info
+
+Detail | Info
+-------|------
+**Size** | 8 bytes
+**Catalog name** | `mz_catalog.uint8`
+**OID** | 14,464
+**Range** | [0, 18,446,744,073,709,551,615]
+
+## Details
+
+### Valid casts
+
+For details about casting, including contexts, see [Functions:
+Cast](../../functions/cast).
+
+#### Between unsigned integer types
+
+From    | To      | Required context
+--------|---------|--------
+`uint2` | `uint4` | Implicit
+`uint2` | `uint8` | Implicit
+`uint4` | `uint2` | Assignment
+`uint4` | `uint8` | Implicit
+`uint8` | `uint2` | Assignment
+`uint8` | `uint4` | Assignment
+
+#### From unsigned integer types
+
+You can cast unsigned integer types to:
+
+To | Required context
+---|--------
+[`numeric`](../numeric) | Implicit
+[`real`/`double precision`](../float) | Implicit
+[`text`](../text) | Assignment
+[`smallint`/`integer`/`bigint`](../integer) | Depends on specific cast
+
+#### To `uint4` or `uint8`
+
+You can cast the following types to unsigned integer types:
+
+From | Required context
+---|--------
+[`boolean`](../boolean) (`integer` only) | Explicit
+[`jsonb`](../jsonb) | Explicit
+[`oid`](../oid) (`integer` and `bigint` only) | Assignment
+[`numeric`](../numeric) | Assignment
+[`real`/`double precision`](../float) | Assignment
+[`text`](../text) | Explicit
+[`smallint`/`integer`/`bigint`](../integer) | Depends on specific cast
+
+## Examples
+
+```sql
+SELECT 123::uint4 AS int_v;
+```
+```nofmt
+ int_v
+-------
+   123
+```
+
+<hr/>
+
+```sql
+SELECT 1.23::uint4 AS int_v;
+```
+```nofmt
+ int_v
+-------
+     1
+```

src/adapter/src/catalog.rs

@@ -2271,6 +2271,9 @@ impl<S: Append> Catalog<S> {
             CatalogType::Int16 => CatalogType::Int16,
             CatalogType::Int32 => CatalogType::Int32,
             CatalogType::Int64 => CatalogType::Int64,
+            CatalogType::UInt16 => CatalogType::UInt16,
+            CatalogType::UInt32 => CatalogType::UInt32,
+            CatalogType::UInt64 => CatalogType::UInt64,
             CatalogType::Interval => CatalogType::Interval,
             CatalogType::Jsonb => CatalogType::Jsonb,
             CatalogType::Numeric => CatalogType::Numeric,
@@ -4553,6 +4556,9 @@ impl ExprHumanizer for ConnCatalog<'_> {
                     .join(",")
             ),
             PgLegacyChar => "\"char\"".into(),
+            UInt16 => "uint2".into(),
+            UInt32 => "uint4".into(),
+            UInt64 => "uint8".into(),
             ty => {
                 let pgrepr_type = mz_pgrepr::Type::from(ty);
                 let pg_catalog_schema =

src/adapter/src/catalog/builtin.rs

@@ -837,6 +837,72 @@ pub const TYPE_ANYCOMPATIBLEMAP: BuiltinType<NameReference> = BuiltinType {
     },
 };
 
+pub const TYPE_UINT2: BuiltinType<NameReference> = BuiltinType {
+    name: "uint2",
+    schema: MZ_CATALOG_SCHEMA,
+    oid: mz_pgrepr::oid::TYPE_UINT2_OID,
+    details: CatalogTypeDetails {
+        typ: CatalogType::UInt16,
+        array_id: None,
+    },
+};
+
+pub const TYPE_UINT2_ARRAY: BuiltinType<NameReference> = BuiltinType {
+    name: "_uint2",
+    schema: MZ_CATALOG_SCHEMA,
+    oid: mz_pgrepr::oid::TYPE_UINT2_ARRAY_OID,
+    details: CatalogTypeDetails {
+        typ: CatalogType::Array {
+            element_reference: TYPE_UINT2.name,
+        },
+        array_id: None,
+    },
+};
+
+pub const TYPE_UINT4: BuiltinType<NameReference> = BuiltinType {
+    name: "uint4",
+    schema: MZ_CATALOG_SCHEMA,
+    oid: mz_pgrepr::oid::TYPE_UINT4_OID,
+    details: CatalogTypeDetails {
+        typ: CatalogType::UInt32,
+        array_id: None,
+    },
+};
+
+pub const TYPE_UINT4_ARRAY: BuiltinType<NameReference> = BuiltinType {
+    name: "_uint4",
+    schema: MZ_CATALOG_SCHEMA,
+    oid: mz_pgrepr::oid::TYPE_UINT4_ARRAY_OID,
+    details: CatalogTypeDetails {
+        typ: CatalogType::Array {
+            element_reference: TYPE_UINT4.name,
+        },
+        array_id: None,
+    },
+};
+
+pub const TYPE_UINT8: BuiltinType<NameReference> = BuiltinType {
+    name: "uint8",
+    schema: MZ_CATALOG_SCHEMA,
+    oid: mz_pgrepr::oid::TYPE_UINT8_OID,
+    details: CatalogTypeDetails {
+        typ: CatalogType::UInt64,
+        array_id: None,
+    },
+};
+
+pub const TYPE_UINT8_ARRAY: BuiltinType<NameReference> = BuiltinType {
+    name: "_uint8",
+    schema: MZ_CATALOG_SCHEMA,
+    oid: mz_pgrepr::oid::TYPE_UINT8_ARRAY_OID,
+    details: CatalogTypeDetails {
+        typ: CatalogType::Array {
+            element_reference: TYPE_UINT8.name,
+        },
+        array_id: None,
+    },
+};
+
 pub const MZ_DATAFLOW_OPERATORS: BuiltinLog = BuiltinLog {
     name: "mz_dataflow_operators",
     schema: MZ_CATALOG_SCHEMA,
@@ -2146,6 +2212,12 @@ pub static BUILTINS_STATIC: Lazy<Vec<Builtin<NameReference>>> = Lazy::new(|| {
         Builtin::Type(&TYPE_ANYCOMPATIBLENONARRAY),
         Builtin::Type(&TYPE_ANYCOMPATIBLELIST),
         Builtin::Type(&TYPE_ANYCOMPATIBLEMAP),
+        Builtin::Type(&TYPE_UINT2),
+        Builtin::Type(&TYPE_UINT2_ARRAY),
+        Builtin::Type(&TYPE_UINT4),
+        Builtin::Type(&TYPE_UINT4_ARRAY),
+        Builtin::Type(&TYPE_UINT8),
+        Builtin::Type(&TYPE_UINT8_ARRAY),
     ];
     for (schema, funcs) in &[
         (PG_CATALOG_SCHEMA, &*mz_sql::func::PG_CATALOG_BUILTINS),

src/adapter/src/coord/timestamp_selection.rs

@@ -73,6 +73,9 @@ impl<S: Append + 'static> Coordinator<S> {
                 ScalarType::Int16 => evaled.unwrap_int16().try_into()?,
                 ScalarType::Int32 => evaled.unwrap_int32().try_into()?,
                 ScalarType::Int64 => evaled.unwrap_int64().try_into()?,
+                ScalarType::UInt16 => evaled.unwrap_uint16().into(),
+                ScalarType::UInt32 => evaled.unwrap_uint32().into(),
+                ScalarType::UInt64 => evaled.unwrap_uint64(),
                 ScalarType::TimestampTz => {
                     evaled.unwrap_timestamptz().timestamp_millis().try_into()?
                 }

src/expr/src/scalar.proto

@@ -337,6 +337,51 @@ message ProtoUnaryFunc {
         google.protobuf.Empty trunc_float32 = 216;
         google.protobuf.Empty trunc_float64 = 217;
         google.protobuf.Empty trunc_numeric = 218;
+        google.protobuf.Empty cast_int16_to_uint16 = 219;
+        google.protobuf.Empty cast_int16_to_uint32 = 220;
+        google.protobuf.Empty cast_int16_to_uint64 = 221;
+        google.protobuf.Empty cast_int32_to_uint16 = 222;
+        google.protobuf.Empty cast_int32_to_uint32 = 223;
+        google.protobuf.Empty cast_int32_to_uint64 = 224;
+        google.protobuf.Empty cast_int64_to_uint16 = 225;
+        google.protobuf.Empty cast_int64_to_uint32 = 226;
+        google.protobuf.Empty cast_int64_to_uint64 = 227;
+        google.protobuf.Empty cast_numeric_to_uint16 = 228;
+        google.protobuf.Empty cast_numeric_to_uint32 = 229;
+        google.protobuf.Empty cast_numeric_to_uint64 = 230;
+        google.protobuf.Empty cast_float32_to_uint16 = 231;
+        google.protobuf.Empty cast_float32_to_uint32 = 232;
+        google.protobuf.Empty cast_float32_to_uint64 = 233;
+        google.protobuf.Empty cast_float64_to_uint16 = 234;
+        google.protobuf.Empty cast_float64_to_uint32 = 235;
+        google.protobuf.Empty cast_float64_to_uint64 = 236;
+        google.protobuf.Empty cast_string_to_uint16 = 237;
+        google.protobuf.Empty cast_string_to_uint32 = 238;
+        google.protobuf.Empty cast_string_to_uint64 = 239;
+        google.protobuf.Empty cast_uint16_to_float32 = 240;
+        google.protobuf.Empty cast_uint16_to_float64 = 241;
+        google.protobuf.Empty cast_uint16_to_int32 = 242;
+        google.protobuf.Empty cast_uint16_to_uint32 = 243;
+        google.protobuf.Empty cast_uint16_to_int64 = 244;
+        google.protobuf.Empty cast_uint16_to_uint64 = 245;
+        google.protobuf.Empty cast_uint16_to_string = 246;
+        mz_repr.adt.numeric.ProtoOptionalNumericMaxScale cast_uint16_to_numeric = 247;
+        google.protobuf.Empty cast_uint32_to_float32 = 248;
+        google.protobuf.Empty cast_uint32_to_float64 = 249;
+        google.protobuf.Empty cast_uint32_to_uint16 = 250;
+        google.protobuf.Empty cast_uint32_to_int32 = 251;
+        google.protobuf.Empty cast_uint32_to_int64 = 252;
+        google.protobuf.Empty cast_uint32_to_uint64 = 253;
+        google.protobuf.Empty cast_uint32_to_string = 254;
+        mz_repr.adt.numeric.ProtoOptionalNumericMaxScale cast_uint32_to_numeric = 255;
+        google.protobuf.Empty cast_uint64_to_float32 = 256;
+        google.protobuf.Empty cast_uint64_to_float64 = 257;
+        google.protobuf.Empty cast_uint64_to_uint16 = 258;
+        google.protobuf.Empty cast_uint64_to_int32 = 259;
+        google.protobuf.Empty cast_uint64_to_uint32 = 260;
+        google.protobuf.Empty cast_uint64_to_int64 = 261;
+        google.protobuf.Empty cast_uint64_to_string = 262;
+        mz_repr.adt.numeric.ProtoOptionalNumericMaxScale cast_uint64_to_numeric = 263;
     }
 }
 
@@ -621,5 +666,8 @@ message ProtoEvalError {
         ProtoIncompatibleArrayDimensions incompatible_array_dimensions = 52;
         string type_from_oid = 53;
         ProtoIndexOutOfRange index_out_of_range = 54;
+        google.protobuf.Empty uint16_out_of_range = 55;
+        google.protobuf.Empty uint32_out_of_range = 56;
+        google.protobuf.Empty uint64_out_of_range = 57;
     }
 }