[GLOBAL-2] [MEDIUM] Submodules pull conflicting openzeppelin-contracts/forge-std versions

[thedavidmeister]

Submodules pull conflicting versions of openzeppelin-contracts and forge-std

Transitive submodules pin different versions of shared dependencies. Solidity merges them under a single remapping target, so whichever version foundry.toml resolves first wins, and the others are silently ignored. The mismatch is invisible at compile time and easy to drift further.

lib/openzeppelin-contracts versions across submodules:

• lib/rain.factory/lib/openzeppelin-contracts — v4.9.5-2 (dc44c9f1...)
• lib/rain.interpreter.interface/lib/openzeppelin-contracts — v4.9.5 (bd325d56...)
• lib/rain.math.fixedpoint/lib/openzeppelin-contracts — v4.8.3 (0a25c194...)
• lib/openzeppelin-contracts-upgradeable (top-level) — v4.9.3 (3d4c0d57...)

lib/forge-std versions across submodules:

• top-level — v1.6.1
• lib/rain.factory/lib/forge-std — v1.9.3
• lib/rain.interpreter.interface/lib/forge-std — v1.7.6
• lib/rain.math.fixedpoint/lib/forge-std — v1.5.2

Fix: bump lib/rain.math.fixedpoint to a release that tracks v4.9.x oz / a current forge-std, or add explicit per-import remappings so each consumer resolves its expected version. Pinning a single oz/forge-std at the top level via remappings.txt would also work but only if every submodule's API surface is compatible.

[thedavidmeister]

Closing as not planned (superseded) — this submodule-version-conflict is being made moot by the soldeer migration (#475), which removes git submodules entirely. NB: #475 is currently OPEN, so the submodule conflict technically still exists on main until it lands; if the migration is abandoned, reopen this.

[thedavidmeister]

Reopening — this was closed as superseded by the soldeer migration #475, but #475 is still an OPEN (stalled) draft, so the submodule-version conflict still exists on main. An issue covered/made-moot by an OPEN PR isn't closeable until that PR actually lands; this stays open until #475 merges (or is formally abandoned).

[thedavidmeister]

🤖 ai:producer
Design question — awaiting human design ruling before any PR: which fix shape for the conflicting submodule dependency versions — (a) bump lib/rain.math.fixedpoint (+ others) to versions tracking oz v4.9.x / current forge-std, (b) explicit per-import remappings so each consumer resolves its pinned version, or (c) neither, because the soldeer migration (#475, currently a stalled open draft) removes the submodule graph wholesale and moots both. First-principles question it fails: interim dependency-pin churn changes compiled bytecode in a repo whose dependency system is mid-migration — landing (a)/(b) now may be work the migration immediately deletes, but waiting leaves a silent version conflict live on main. (automated design-gate note)