Rename node name attribute to identifier

[lorisleiva]

Phase D, mechanical opener. Renames the v2 name node attribute to identifier across all hand-written package source — a 1:1 rename with no behavioural changes.

Stacked on #1158 (the v2 regeneration base).

What changed

• .name member accesses on renamed node kinds → .identifier (programNode, accountNode, instructionNode, errorNode, eventNode, pdaNode, constantNode, definedTypeNode, instructionAccountNode, instructionRemainingAccountsNode, providedNode, structFieldTypeNode, enumVariantTypeNode, variablePdaSeedNode, the value nodes, and all link nodes).
• name: object-literal keys constructing those nodes → identifier:.
• update*Visitor config reads (updates.name, argUpdate.name) → identifier, since those config types are Partial<…NodeInput> which use identifier in v2.
• The 'name' in node narrowing sites — which type-check against the v2 Node union but are runtime no-ops (NodeSelector.ts, NodePath.ts, getDebugStringVisitor.ts default branch) — switched to 'identifier' in node / node.identifier, plus the selector example in the README.

Deliberately left untouched (not the name attribute)

• Raw Anchor IDL reads (idl.*.name, 'name' in field), commander .name(), error-context name: keys, projected plain-object keys, and anything inside removed-node guards (argumentValueNode, resolverValueNode, instructionArgumentLinkNode) — handled in the per-package adaptation PRs.
• NodeSelector's camelCase(name) !== node.identifier comparison: v2 identifiers are no longer camelCase-normalised, so the comparison semantics need a case-fold rethink. Cast to string here to keep the rename compiling; the semantic fix is a selector follow-up.

Note for per-package PRs

instructionArgumentNode is a removed node in v2, but its construction sites still had name: → identifier: renamed here (harmless — they become struct fields, which use identifier). Don't treat those as fully migrated; the argument→data rework happens per-package. Similarly fieldDiscriminatorNode.name → path and accountFieldValueNode → accountDataValueNode are out-of-scope shape changes left for the per-package PRs.

Status

Mechanical rename only; packages are not yet green (numeric/transform/removed-node work follows per-package). Every remaining 'name' error in src/ sits inside a removed-node branch. test/ and generated files are untouched.

[changeset-bot]

⚠️ No Changeset found

Latest commit: 2804f7e

Merging this PR will not cause a version bump for any packages. If these changes should not result in a new version, you're good to go. If these changes should result in a version bump, you need to add a changeset.

This PR includes no changesets

When changesets are added to this PR, you'll see the packages that this PR includes changesets for and the associated semver types

Click here to learn what changesets are, and how to add one.

Click here if you're a maintainer who wants to add a changeset to this PR

[lorisleiva]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• Adapt errors to Codama v2 #1167
• Rename node name attribute to identifier #1162 👈 (View in Graphite)
• Regenerate the node types from the v2 spec #1158
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[lorisleiva]

@trevor-cortex

[trevor-cortex]

Summary

Mechanical name → identifier rename across hand-written src/ in the dynamic-*, nodes-from-anchor, validators, visitors-core and visitors packages. I went through all 87 files: every changed line is a faithful 1:1 swap, the update*Visitor config reads (updates.identifier, argUpdate.identifier) line up with the Partial<…NodeInput> types, and the value-node/link-node accesses I spot-checked (accountValueNode, accountBumpValueNode, pdaSeedValueNode, link nodes) all have identifier: IdentifierString in the regenerated @codama/node-types. Removed-node branches (argumentValueNode, resolverValueNode, instructionArgumentLinkNode) are consistently left on .name as described. Formatter re-wraps look right.

Two in-scope misses the TS check can't catch

The completeness claim in the description rests on 'name' does not exist compiler errors. There are two places that use the 'name' in node narrowing pattern, which compiles cleanly against the v2 Node union but silently returns false at runtime for every node:

1. packages/visitors-core/src/NodeSelector.ts ~L36 — if (name && (!('name' in node) || camelCase(name) !== node.name)). This means every name-bearing selector ([programNode]foo, [errorNode]${name}, [definedTypeNode]bar, the selector argument of updateAccountsVisitor/updateDefinedTypesVisitor/updateProgramsVisitor, deleteNodesVisitor, …) will match nothing. Several files in this very PR now build selectors from .identifier (deduplicateIdenticalDefinedTypesVisitor, addPdasVisitor, updateAccountsVisitor) and would be no-ops. Needs 'identifier' in node / node.identifier. (Separately, and for a follow-up rather than this PR: camelCase(name) !== node.identifier will also need a decision now that v2 identifiers aren't camelCase-normalised — case-fold/underscore-strip compare per the brand docstring, or normalise both sides.)

2. packages/visitors-core/src/getDebugStringVisitor.ts ~L115 — the default: branch return 'name' in node ? [node.name] : [];. This is the generic fallback that prints the name for accountNode, instructionNode, definedTypeNode, pdaNode, structFieldTypeNode, enum variants, variablePdaSeedNode, accountValueNode, pdaSeedValueNode, etc. — i.e. most of the debug output. Same fix. (Anchored a nit on the link-node branch a few lines above.)

Worth a quick rg "'name' in" across packages/*/src to confirm those are the only two.

Notes for the description / follow-ups

• instructionArgumentNode is also a removed node in v2 — InstructionNode now carries data?: TypeNode and there's no InstructionArgumentNode in @codama/node-types (the remaining packages/nodes/src/InstructionArgumentNode.ts helper imports a type that no longer exists). This PR still renames name: → identifier: on instructionArgumentNode({...}) constructions and ix.arguments[].name reads (nodes-from-anchor InstructionNode.ts/InstructionArgumentNode.ts, createSubInstructionsFromEnumArgsVisitor, setInstructionDiscriminatorsVisitor, updateInstructionsVisitor, resolve-standalone-pda, encode-instruction-arguments, …). That's harmless — those sites will become struct fields, which do use identifier — but it's not in the description's renamed-kinds list and doesn't fit the "removed-node branches are left untouched" rule either. Might be worth a line in the description so the per-package PRs don't treat them as done.
• Other non-rename shape changes I noticed while cross-checking, all correctly out of scope here but easy to lose track of: fieldDiscriminatorNode.name → path (dynamic-parsers/discriminators.ts, getDebugStringVisitor, nodes-from-anchor fieldDiscriminatorNode('discriminator')); accountFieldValueNode → accountDataValueNode (resolve-consumed-members.ts, resolve-injected-value.ts).

For subsequent reviewers

Nothing else needs deep attention — this is the rename it says it is. Once the two 'name' in node sites are fixed I'm happy to approve.