Issues

Wrong code (dccpeep): assigned 32-bit parameter read through a pointer after an empty loop returns the old value (pass_elim_dead_ix_stores forgets the escape at an LS label)

davidly/dcc#273 · opened

on Nov 13, 2026

Wrong code: in a loop, `a[1] = (0 ? K : x) - 1.0f` with a 32-bit parameter `x` assigned after the loop subtracts 1 from `x` on every iteration

davidly/dcc#272 · opened

on Nov 13, 2026

1

Preprocessor: a macro parameter named like a numeric suffix (`f`, `u`, `L`) is substituted inside the constant (`9.0f` becomes `9.0a1`)

davidly/dcc#271 · opened

on Nov 13, 2026

Wrong code: `(int32_t)(uint32_t)x` converted to float converts the unsigned value when `x` is a signed 8- or 16-bit value

davidly/dcc#270 · opened

on Nov 13, 2026

2

Wrong code: `!f` is 0 when the float `f` is negative zero

davidly/dcc#269 · opened

on Nov 13, 2026

Wrong code: `(uint32_t)x` converted to float converts the signed value when `x` is a signed 32-bit expression

davidly/dcc#268 · opened

on Nov 13, 2026

Wrong code: `f + (x * y)` drops the product when `x` is an integer, or `y` is a local/parameter and `x` is a global, element, member or sub-expression

davidly/dcc#267 · opened

on Nov 13, 2026

1

Valid C rejected: `return` does not convert between `float` and a 32-bit integer (`float f(int32_t x) { return x; }`)

davidly/dcc#266 · opened

on Nov 13, 2026

Valid C rejected: comparison against a negative float constant combined with another operand (`x < -1.0f - n`) — "MIR emission is required"

davidly/dcc#265 · opened

on Nov 13, 2026

1

Wrong code: `++`/`--` on a float is a no-op on a scalar and a 16-bit integer increment on an array element

davidly/dcc#264 · opened

on Nov 13, 2026

Compile error: compound assignment to an integer lvalue with a float right operand is rejected (`l *= f`, `n += 1.5f`)

davidly/dcc#263 · opened

on Nov 13, 2026

Compile error: a float constant that must be folded or tested for truth is rejected (`-(-2.5f)`, `!2.5f`, `2.5f && n`, `if (2.3f)`)

davidly/dcc#262 · opened

on Nov 13, 2026

1

Compile error: the value of a float `++`/`--` is rejected (`g = f++`, `if (f-- > x)`, `while (f-- > 0)`)

davidly/dcc#261 · opened

on Nov 13, 2026

Compile error: a static float initializer must be a bare constant; `(2.5f)`, `-(2.5f)`, `1.0f + 2.0f` are rejected

davidly/dcc#260 · opened

on Nov 13, 2026

Wrong result: float division (`__fdiv`) is 1 ulp off in about a fifth of cases, in both directions

davidly/dcc#259 · opened

on Nov 13, 2026

A 16-bit local assigned in one arm of an `if … else`, then updated, loses the assignment when an early-return `if` follows

davidly/dcc#258 · opened

on Nov 13, 2026

In a `for` or `while` loop with a 32-bit counter, `l = i;` copies the counter after its increment

davidly/dcc#257 · opened

on Nov 13, 2026

A 32-bit local assigned in a loop and passed directly to a call keeps its old value after the loop

davidly/dcc#256 · opened

on Nov 13, 2026

2

dccpeep: `preserve_ix_word_signed_range` makes `a > x || x < k` compare a 16-bit local with itself when `a` is a struct member already on the stack

davidly/dcc#255 · opened

on Nov 13, 2026

dccpeep: `cp8_to_bit3` turns `8 == v` into a test of bit 3, so 9, 12, 47… compare equal to 8

davidly/dcc#254 · opened

on Nov 13, 2026

1

Valid C rejected (DCC-E1002): a constant divided by constant zero in code that never runs; the message does not mention division

davidly/dcc#253 · opened

on Nov 13, 2026

dccpeep: `small_const_incr_carry_skip` drops `x + K` when the sum is then copied (HL read as separate `l`/`h` is not seen)

davidly/dcc#252 · opened

on Nov 13, 2026

1

Wrong code: nested-block 32-bit local initialised from another local, stored to a 16-bit element/member/`*p`, writes a garbage high byte

davidly/dcc#251 · opened

on Nov 13, 2026

2

Wrong code: a `for`-init declaration shares a stack slot with its enclosing block's locals when an earlier sibling scope declares a local

davidly/dcc#250 · opened

on Nov 13, 2026

1

Wrong code: 8-bit local and another variable assigned in opposite arms of an `if … else`, then `(int)c + x` loses `x`

davidly/dcc#249 · opened

on Nov 13, 2026

Navigation Menu

Repository navigation

Issues

Wrong code (dccpeep): assigned 32-bit parameter read through a pointer after an empty loop returns the old value (pass_elim_dead_ix_stores forgets the escape at an LS label)

Wrong code: in a loop, `a[1] = (0 ? K : x) - 1.0f` with a 32-bit parameter `x` assigned after the loop subtracts 1 from `x` on every iteration

Preprocessor: a macro parameter named like a numeric suffix (`f`, `u`, `L`) is substituted inside the constant (`9.0f` becomes `9.0a1`)

Wrong code: `(int32_t)(uint32_t)x` converted to float converts the unsigned value when `x` is a signed 8- or 16-bit value

Wrong code: `!f` is 0 when the float `f` is negative zero

Wrong code: `(uint32_t)x` converted to float converts the signed value when `x` is a signed 32-bit expression

Wrong code: `f + (x * y)` drops the product when `x` is an integer, or `y` is a local/parameter and `x` is a global, element, member or sub-expression

Valid C rejected: `return` does not convert between `float` and a 32-bit integer (`float f(int32_t x) { return x; }`)

Valid C rejected: comparison against a negative float constant combined with another operand (`x < -1.0f - n`) — "MIR emission is required"

Wrong code: `++`/`--` on a float is a no-op on a scalar and a 16-bit integer increment on an array element

Compile error: compound assignment to an integer lvalue with a float right operand is rejected (`l *= f`, `n += 1.5f`)

Compile error: a float constant that must be folded or tested for truth is rejected (`-(-2.5f)`, `!2.5f`, `2.5f && n`, `if (2.3f)`)

Compile error: the value of a float `++`/`--` is rejected (`g = f++`, `if (f-- > x)`, `while (f-- > 0)`)

Compile error: a static float initializer must be a bare constant; `(2.5f)`, `-(2.5f)`, `1.0f + 2.0f` are rejected

Wrong result: float division (`__fdiv`) is 1 ulp off in about a fifth of cases, in both directions

A 16-bit local assigned in one arm of an `if … else`, then updated, loses the assignment when an early-return `if` follows

In a `for` or `while` loop with a 32-bit counter, `l = i;` copies the counter after its increment

A 32-bit local assigned in a loop and passed directly to a call keeps its old value after the loop

dccpeep: `preserve_ix_word_signed_range` makes `a > x || x < k` compare a 16-bit local with itself when `a` is a struct member already on the stack

dccpeep: `cp8_to_bit3` turns `8 == v` into a test of bit 3, so 9, 12, 47… compare equal to 8

Valid C rejected (DCC-E1002): a constant divided by constant zero in code that never runs; the message does not mention division

dccpeep: `small_const_incr_carry_skip` drops `x + K` when the sum is then copied (HL read as separate `l`/`h` is not seen)

Wrong code: nested-block 32-bit local initialised from another local, stored to a 16-bit element/member/`*p`, writes a garbage high byte

Wrong code: a `for`-init declaration shares a stack slot with its enclosing block's locals when an earlier sibling scope declares a local

Wrong code: 8-bit local and another variable assigned in opposite arms of an `if … else`, then `(int)c + x` loses `x`