Skip to content

[General]: The manual's for-loop rules give downto loops an empty variable type and zero iterations #102

Open

Description

@shuklaayush

Problem

The language manual (ISA_LRM, "For statements") says the loop variable "is
incremented each time round the loop if the to form is used or decremented each time
round the loop if the downto form is used". Two later paragraphs then state, for both
forms:

The variable can be thought of as if it was declared let i : {e1..e2} := … at the start
of the statement list.

If e1 > e2, the loop executes 0 times. If e1 == e2, the loop executes 1 time. The loop
executes Max(1 + e2 - e1, 0) times.

For for i := 31 downto 0, these give i : {31..0}, an empty set, and
Max(1 + 0 - 31, 0) = 0 iterations. CRC32_Mod_2 (for i := 31 downto 0) and
GF2_P8_Multiply (for i := 14 downto 8) need e1 - e2 + 1 iterations.

Proposed fix

Preserve the sentences about evaluating the bounds once and the loop variable's
immutability and scope (ISA_LRM). Let e1 and e2 denote the evaluated start
and end bounds. Replace only the explanation of the variable's type and the iteration
count with:

For `for i := e1 to e2`, the loop variable has type {e1..e2}
and the body executes Max(1 + e2 - e1, 0) times.

For `for i := e1 downto e2`, the loop variable has type {e2..e1}
and the body executes Max(1 + e1 - e2, 0) times.

AI disclosure

Assisted-by: Claude

Claude assisted with source analysis, the proposed correction and drafting this report.
Reviewed with assistance from Codex.

Pinned source: definitions.xml.gz at revision 35eface1b7ee003a1cda95005ec5fd83de658a64.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions