Skip to content

RAID candidate Conformist patterns #3

Description

@TJackhammer

This issue is the rolling backlog of candidate Conformist patterns for RAID review (.raid/patterns.md).

RAID's /raid:review appends here whenever a live review surfaces a recurring convention that isn't yet a shipped pattern. Periodically these graduate into .raid/patterns.md (after blind-validation), or get pruned if they don't recur.

How it was seeded

First onboard mined merged-PR review comments from the 400 most-recent reviewed PRs (verilator#6592verilator#7824), split time-interspersed into a 200-PR mine-half and 200-PR holdout-half. 15 patterns shipped — each independently re-fired on the holdout. See .raid/patterns.ledger.json for exactly which PRs were examined.

Watch list (re-fired weakly or overlap a shipped pattern — promote if they recur)

  • store-node-ptr-not-string — store an AstNode*, not a precomputed name string/fileline (overlaps the O(n²) pattern)
  • build-ast-not-text — build real AST nodes rather than literal C++/Verilog text or passed name strings
  • vl-restorer — save/restore visitor member state with VL_RESTORER
  • sort-arbitrary-lists — keep enum/option entries sorted
  • p-suffix / m_ naming conventions

Deliberately excluded (generic — a non-Verilator reviewer already catches these)

const-everywhere, avoid auto, no single-statement braces, generic dedup-into-common-function, and "error vs warning by other-simulator behavior" (real policy, but not detectable from a diff).


Seeded by /raid:onboard.

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