bugc: combine thirteen fixes and additions

packages/bugc-react/CHANGELOG.md

@@ -7,6 +7,17 @@ specification itself are tracked in the root
 
 ## Unreleased
 
+### Added
+
+- The CFG view shows a `block.prevrandao` read as `block.prevrandao`
+  ([#363]).
+- The IR view shows bugc's new `copy` instruction ([#359]).
+
+### Changed
+
+- The IR and CFG views show each operand of a `hash` instruction, which
+  now has a list of `values` ([#365]).
+
 ## 0.1.0-preview.1 — 2026-10-03
 
 Updated `@ethdebug/bugc` to `0.1.0-preview.1`.
@@ -34,3 +45,6 @@ First publication.
 
 [#299]: https://github.com/ethdebug/format/pull/299
 [#300]: https://github.com/ethdebug/format/pull/300
+[#359]: https://github.com/ethdebug/format/pull/359
+[#363]: https://github.com/ethdebug/format/pull/363
+[#365]: https://github.com/ethdebug/format/pull/365

packages/bugc-react/src/components/CfgView.tsx

@@ -349,12 +349,14 @@ function CfgViewContent({ ir }: CfgViewProps): JSX.Element {
             return `${envInst.dest} = block.timestamp`;
           case "block_number":
             return `${envInst.dest} = block.number`;
+          case "block_prevrandao":
+            return `${envInst.dest} = block.prevrandao`;
           default:
             return `${envInst.dest} = ${envInst.op}`;
         }
       }
       case "hash":
-        return `${inst.dest} = keccak256(${formatValue(inst.value)})`;
+        return `${inst.dest} = keccak256(${inst.values.map(formatValue).join(", ")})`;
       case "cast":
         return `${inst.dest} = cast ${formatValue(inst.value)} to ${inst.targetType.kind}`;
       case "compute_slot": {

packages/bugc-react/src/components/IrView.tsx

@@ -160,7 +160,10 @@ function InstructionRenderer({
 
     case "hash":
       add(`${formatDest(instruction.dest)} = hash `);
-      addOperand("value", formatValue(instruction.value));
+      instruction.values.forEach((value, index) => {
+        if (index > 0) add(", ");
+        addOperand("value", formatValue(value));
+      });
       break;
 
     case "cast":
@@ -336,6 +339,16 @@ function InstructionRenderer({
       break;
     }
 
+    case "copy":
+      add("memory[offset: ");
+      addOperand("offset", formatValue(instruction.offset));
+      add(", length: ");
+      addOperand("length", formatValue(instruction.length));
+      add(`] = copy ${instruction.location}[`);
+      addOperand("source", formatValue(instruction.source));
+      add("]");
+      break;
+
     default:
       add(`; unknown instruction: ${(instruction as { kind?: string }).kind}`);
   }

packages/bugc-react/src/utils/irDebugUtils.ts

@@ -90,10 +90,6 @@ export function extractInstructionDebug(
       operands.push({ label: "operand", debug: instruction.operandDebug });
       break;
 
-    case "hash":
-      operands.push({ label: "value", debug: instruction.valueDebug });
-      break;
-
     case "cast":
       operands.push({ label: "value", debug: instruction.valueDebug });
       break;

packages/bugc/CHANGELOG.md

@@ -9,6 +9,21 @@ support. Changes to the specification itself are tracked in the root
 
 ### Added
 
+- `block.prevrandao`, the previous block's RANDAO value (EIP-4399), as a
+  `uint256`. It compiles to the EVM's `PREVRANDAO` (`0x44`) ([#363]).
+- `keccak256` over value types: `keccak256(a, b, ...)` hashes the 32-byte
+  words of one or more integers, `address`, `bool` or `bytesN` values, in
+  order. For integers, `address` and `bool` this equals Solidity's
+  `keccak256(abi.encode(a, b, ...))`. A `bytesN` narrower than 32 bytes
+  hashes its word with its bytes at the right end, unlike `abi.encode`;
+  cast it to `bytes32` to match Solidity. A single dynamic `bytes` or
+  `string` argument still hashes its data, and must be the only argument.
+  The optimizer folds a hash of constant words at levels 1 to 3 ([#365]).
+- `push` on a dynamic array in storage, as in Solidity: `a.push(v)`
+  writes `v` at index `a.length` and adds one to the length, and
+  `a.push()` adds a zero element, as for an array of arrays. `a` may be
+  in a mapping or a struct. A push has no value. On an array in memory
+  or a fixed-size array, it is a type error ([#366]).
 - The `%` operator, with the precedence of `*` and `/`. It compiles to
   the EVM's `MOD` (`SMOD` for signed operands), so `x % 0` is `0`, as
   `x / 0` is ([#321]).
@@ -28,6 +43,24 @@ support. Changes to the specification itself are tracked in the root
 
 ### Changed
 
+- The IR `hash` instruction now has `values`, a list, in place of
+  `value` and `valueDebug` ([#365]).
+- An index into a dynamic array in storage now reverts with Solidity's
+  `Panic(0x32)` unless it is less than the array's length, for reads and
+  writes, as an index into an array in memory does. Before, a write past
+  the end wrote there and left the length as it was, so the array's
+  pointer did not show the element ([#366]).
+- Storage arrays now have Solidity's layout. A fixed-size array is
+  inline: its elements start at its own slot, so `[0] a: array<uint256,
+3>` takes slots 0 to 2, and a struct member that is a fixed-size array
+  takes all its slots. A dynamic array still keeps its length in its
+  slot and its elements from `keccak256(slot)`. In either, elements
+  narrower than 16 bytes now share a slot, as many as fit, from its
+  low-order end: an `array<uint8, 40>` takes two slots. Each storage
+  variable's pointer describes the same layout. Before, every array's
+  elements started at `keccak256(slot)`, one element to a slot, while
+  the pointer of a fixed-size array described it inline, and the pointer
+  of an array of narrow elements described them packed ([#364]).
 - An integer literal operand of an arithmetic or comparison operator now
   takes the type of the other operand when its value fits, so with
   `x: int8`, `x < 0`, `x == 1` and `-1 < x` compare as `int8`. A literal
@@ -79,6 +112,8 @@ support. Changes to the specification itself are tracked in the root
 
 ### Fixed
 
+- `a.length` on a dynamic array in storage now reads the length in the
+  array's slot. Before, it did not read that slot ([#366]).
 - At optimization levels 2 and 3, an inlined body's control flow now has
   a `transform: ["inline"]` context: the branch of an `if`, the jumps
   between the body's blocks, each block's `JUMPDEST`, and the jump that
@@ -89,6 +124,61 @@ support. Changes to the specification itself are tracked in the root
   has the branch's context: its source range, its variables (the storage
   variables too) and its transforms. Before, it had no context, so a
   debugger listed no variables while it ran ([#356]).
+- `!` is now a logical not: `!true` is `false`. Before, at every
+  optimization level, it compiled to the EVM's bitwise `NOT`, so `!x` was
+  a non-zero word, and so `true`, for every `x` ([#353]).
+- A source range now ends at its node's last token. Before, a node
+  followed by whitespace or a comment included them, so most `code`
+  ranges had trailing whitespace: `block.number ` in `block.number + 1`,
+  the target `x ` in `x = 1`, a statement up to the next statement, and
+  the `create` and `code` blocks up to the next block ([#362]).
+- A slice of a fixed-size `bytesN` value, as in
+  `keccak256("transfer(address,uint256)")[0:4]`, now gives those bytes as
+  dynamic `bytes`. Before, at every optimization level, the compiler read
+  the value as the memory address of dynamic `bytes`, so the slice
+  reverted with `Panic(0x32)` or held other memory ([#357]).
+- `keccak256` of dynamic `bytes` or a `string` now hashes its data, and
+  `keccak256(msg.data)` hashes the calldata. Before, code generation
+  hashed the word that refers to the data (its memory address, or 0 for
+  `msg.data`). So at optimization level 0, `keccak256("transfer()")` was
+  wrong, while levels 1 to 3 folded it to the right constant; at every
+  level, the hash of a slice or of `msg.data` was wrong ([#360]).
+- The `invoke` context on a caller's `JUMP` into a function now has the
+  call's `arguments`, as the callee's entry `JUMPDEST` does: the `JUMP`
+  leaves them on the stack, and the `JUMPDEST` does not change it.
+  Before, only the `JUMPDEST`'s `invoke` had them, so a debugger that
+  opens the frame on the step after the `JUMP` showed it for one step
+  without its arguments ([#354]).
+- The instructions that run before a function's first statement now have
+  a `variables` context: the memory setup at the start of the code (and
+  `create`) blocks, and a function's entry `JUMPDEST` and prologue. They
+  list the storage variables and the locals in scope, a local without a
+  pointer: a parameter is not yet in its home there, and the `invoke`'s
+  `arguments` point at it on the stack. Before, they listed no variables,
+  so a debugger showed none for those steps ([#358]).
+- Assigning a memory `string` or `bytes` to storage, as in `s = m;`, now
+  stores its bytes as Solidity does: up to 31 bytes in the slot, with
+  twice the length in the low byte; else twice the length plus one in the
+  slot, and the data from `keccak256(slot)`. A slice's bytes past its
+  length are stored as zero. Before, the slot held the value's memory
+  address ([#355]).
+- A slice now copies all its bytes, as in `msg.data[4:]` or `m[1:40]`.
+  The IR has a `copy` instruction, which compiles to `CALLDATACOPY` from
+  calldata and `MCOPY` from memory. Before, a slice copied one word, so a
+  slice longer than 32 bytes held zeros after its first 32 ([#359]).
+- A struct read from storage into a local, as in
+  `let p = players[who];` or `let s = stored;`, is now a copy in memory,
+  as the examples' README describes: reading `p.score` reads the copy,
+  and a write to `p.score` changes the copy, not storage. A field that is
+  a struct is copied too; a struct with an array, mapping, `string` or
+  dynamic `bytes` field is an error. Before, at every optimization level,
+  the local held the struct's first storage word as if it were a memory
+  address, and a write to one of its fields failed code generation
+  ([#361]).
+- A value whose only later uses are as the address, slot or value of a
+  read or write, or as the base of an offset, now gets a home in memory
+  when code generation cannot keep it on the stack. Before, such a value
+  could fail code generation with "Cannot load value" ([#361]).
 - A call to a function with no return type now compiles as a statement,
   as in `bump();`, at every optimization level. The function's `return`
   context has no `data`. Before, IR generation failed with "Cannot convert
@@ -300,4 +390,17 @@ First publication.
 [#349]: https://github.com/ethdebug/format/pull/349
 [#351]: https://github.com/ethdebug/format/pull/351
 [#352]: https://github.com/ethdebug/format/pull/352
+[#353]: https://github.com/ethdebug/format/pull/353
+[#354]: https://github.com/ethdebug/format/pull/354
+[#355]: https://github.com/ethdebug/format/pull/355
 [#356]: https://github.com/ethdebug/format/pull/356
+[#357]: https://github.com/ethdebug/format/pull/357
+[#358]: https://github.com/ethdebug/format/pull/358
+[#359]: https://github.com/ethdebug/format/pull/359
+[#360]: https://github.com/ethdebug/format/pull/360
+[#361]: https://github.com/ethdebug/format/pull/361
+[#362]: https://github.com/ethdebug/format/pull/362
+[#363]: https://github.com/ethdebug/format/pull/363
+[#364]: https://github.com/ethdebug/format/pull/364
+[#365]: https://github.com/ethdebug/format/pull/365
+[#366]: https://github.com/ethdebug/format/pull/366

packages/bugc/examples/README.md

@@ -143,3 +143,28 @@ account.isActive = true;
 accounts[user].balance = accounts[user].balance + 100;
 accounts[user].isActive = true;
 ```
+
+## Hashing
+
+`keccak256` has two forms. The compiler tells them apart by the type of
+the argument:
+
+- **One dynamic `bytes` or `string`** hashes its data:
+  `keccak256("transfer(address,uint256)")`.
+- **One or more value types** (integers, `address`, `bool`, `bytesN`)
+  hashes their 32-byte words, concatenated in order. This equals
+  Solidity's `keccak256(abi.encode(a, b, ...))` for integers, `address`
+  and `bool`:
+
+```bug
+// a toy roll: 2 in 3 is a hit
+let hit = (keccak256(block.number, msg.sender) as uint256) % 3 != 0;
+```
+
+A `bytes` or `string` argument must be the only one.
+
+A `bytesN` narrower than 32 bytes hashes its full word, with its bytes
+at the low-order (right) end, as BUG holds it. Solidity's `abi.encode`
+puts a `bytesN` value's bytes at the high-order (left) end, so for
+`bytes4` and the like the hashes differ from Solidity's. Cast to
+`bytes32` first (which puts the bytes at the left end) to match it.

packages/bugc/examples/advanced/nested-arrays.bug

@@ -3,7 +3,7 @@ name NestedArrays;
 // Example 7: Nested Arrays
 //
 // Expected IR:
-//   // First dimension: matrix[i]
+//   // First dimension: matrix[i], checked against matrix's length
 //   %outer_base = compute_slot kind="array", base=15
 //   %inner_array_slot = add %outer_base, %i
 //
@@ -23,8 +23,15 @@ storage {
 }
 
 code {
-  let i = 2;
-  let j = 5;
+  // An index must be below the length, so push the rows and elements
+  // first: matrix[1] gets two elements
+  matrix.push();
+  matrix.push();
+  matrix[1].push(0);
+  matrix[1].push(0);
+
+  let i = 1;
+  let j = 1;
   let value = 999;
   matrix[i][j] = value;
   let elem = matrix[i][j];

packages/bugc/examples/intermediate/arrays.bug

@@ -2,11 +2,11 @@ name ArraysAndLoops;
 
 storage {
   [0] numbers: array<uint256, 10>;
-  [1] sum: uint256;
-  [2] max: uint256;
-  [3] count: uint256;
-  [4] found: bool;
-  [5] searchValue: uint256;
+  [10] sum: uint256;
+  [11] max: uint256;
+  [12] count: uint256;
+  [13] found: bool;
+  [14] searchValue: uint256;
 }
 
 code {
@@ -21,7 +21,7 @@ code {
   after: deploy
   variables:
     sum:
-      pointer: { location: storage, slot: 1 }
+      pointer: { location: storage, slot: 10 }
       value: 0
   */
   for (let i = 0; i < 10; i = i + 1) {
@@ -78,21 +78,21 @@ code {
   fails: stack overflow during call - needs investigation
   variables:
     sum:
-      pointer: { location: storage, slot: 1 }
+      pointer: { location: storage, slot: 10 }
       value: 285
   */
   /*@test final-max
   fails: stack overflow during call - needs investigation
   variables:
     max:
-      pointer: { location: storage, slot: 2 }
+      pointer: { location: storage, slot: 11 }
       value: 81
   */
   /*@test final-count
   fails: stack overflow during call - needs investigation
   variables:
     count:
-      pointer: { location: storage, slot: 3 }
+      pointer: { location: storage, slot: 12 }
       value: 8
   */
 

packages/bugc/src/ast/spec.ts

@@ -1439,7 +1439,8 @@ export namespace Expression {
     | Expression.Special.MsgValue
     | Expression.Special.MsgData
     | Expression.Special.BlockTimestamp
-    | Expression.Special.BlockNumber;
+    | Expression.Special.BlockNumber
+    | Expression.Special.BlockPrevrandao;
 
   export const isSpecial = (
     expression: Expression.Base,
@@ -1451,6 +1452,7 @@ export namespace Expression {
       Expression.Special.isMsgSender,
       Expression.Special.isBlockNumber,
       Expression.Special.isBlockTimestamp,
+      Expression.Special.isBlockPrevrandao,
     ].some((guard) => guard(expression));
 
   export namespace Special {
@@ -1532,6 +1534,26 @@ export namespace Expression {
     export function blockNumber(id: Id, loc?: SourceLocation): BlockNumber {
       return { id, kind: "expression:special:block.number", loc: loc ?? null };
     }
+
+    export interface BlockPrevrandao extends Base {
+      kind: "expression:special:block.prevrandao";
+    }
+
+    export const isBlockPrevrandao = (
+      expression: Expression.Base,
+    ): expression is BlockPrevrandao =>
+      expression.kind === "expression:special:block.prevrandao";
+
+    export function blockPrevrandao(
+      id: Id,
+      loc?: SourceLocation,
+    ): BlockPrevrandao {
+      return {
+        id,
+        kind: "expression:special:block.prevrandao",
+        loc: loc ?? null,
+      };
+    }
   }
 }
 

packages/bugc/src/ast/visitor.ts

@@ -201,6 +201,7 @@ export function visitDetailed<N extends Ast.Node, T, C = never>(
     case "expression:special:msg.data":
     case "expression:special:block.timestamp":
     case "expression:special:block.number":
+    case "expression:special:block.prevrandao":
       return visitor.expressionSpecial(node as Ast.Expression.Special, context);
 
     default: