dev meeting 2026 04 01

Agenda

• Diff regression fix (Ali and Shon)

  • Should we indeed revert (2) in https://github.com/ocaml/dune/pull/13696#issuecomment-4144269511

• Post-action dependency refinement (Ali)

  • We (Tarides) want to make changes to the engine to allow for rules to optionally refine their dependencies
  • The refined dependencies will then be included in the target hash
  • Doing so allows rules to refine their dependencies and choose how their targets become invalidated
  • This will be an engine change which will be used by guarded experimental rules
    • One based on ocamlobjinfo processing
    • Maybe others

• Relocatable compiler work needs review (Ali)

  • https://github.com/ocaml/dune/pull/13321
  • We now detect relocatable or relocatable-compiler in the switch as our source of truth
  • This is a recent decision by upstream opam
  • We need to introduce overlays for the older compilers however due to the change in structure.
  • There are still compiler rebuilds occasionally, but the reason is not clear. Likely env issues, but a more general way of detecting them would be nice.

• When encountering (sandbox always) why do we choose copy and symlink over hardlink? (Ali)

  • Should we be preferring hardlink all the time?
  • Question came up adding hardlinking sandboxes to Windows https://github.com/ocaml/dune/pull/13987

• Formatting multi-line strings (Ali)

  • https://github.com/ocaml/dune/pull/13758
  • is this something we want, should we have a look at it?