[2.x] Adapt dynamic-codecs to Codama v2

[lorisleiva]

This PR adapts @codama/dynamic-codecs to Codama v2.

• Renames getNodeCodec and getNodeCodecVisitor to getNodeValueCodec and getNodeValueCodecVisitor.
• Decodes every integer as a bigint and respects endian. Adds floatTypeNode, fixed points, date-times and durations, which encode as their raw integer.
• Decodes enums as { __kind, __discriminator, data } using raw identifiers. Struct keys are raw identifiers too.
• Applies type node transforms from the innermost to the outermost, including those of link nodes on top of the linked type.
• Adds sentinelCountNode support for arrays, sets and maps.
• Encodes an instruction's data, and resolves injected values from the provides of enclosing instructions.
• Adds ProvidedScope.resolveOrThrow to @codama/visitors-core and the INJECTED_VALUE_NOT_PROVIDED error, thrown when a required injected value is not provided.
• Restores the node stack when a visit throws, so reused visitors keep resolving links correctly.
• Bumps @solana/* dependencies to 8.4.0.

[changeset-bot]

⚠️ No Changeset found

Latest commit: 027a158

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]

• [2.x] Adapt the dynamic-instructions builder to Codama v2 #1194
• [2.x] Reject values of the wrong type when encoding dynamic codecs #1193
• [2.x] Adapt dynamic-address-resolution to Codama v2 #1192
• [2.x] Accept bytes, enum shorthands and struct defaults when encoding dynamic codecs #1191
• [2.x] Add resolveTypePath helper to visitors-core #1190
• [2.x] Adapt dynamic-parsers to Codama v2 #1187
• [2.x] Add withPath and visitPath helpers to NodeStack #1186
• [2.x] Adapt dynamic-codecs to Codama v2 #1185 👈 (View in Graphite)
• [2.x] Add relations, nested argument seeds and legacy PDAs to nodes-from-anchor #1181
• [2.x] Adapt nodes-from-anchor to Codama v2 #1180
• [2.x] Adapt instruction visitors to Codama v2 #1179
• [2.x] Adapt update visitors to Codama v2 #1178
• [2.x] Adapt structure visitors to Codama v2 #1175
• [2.x] Move casing helpers to @codama/fragments #1174
• [2.x] Adapt validators to Codama v2 #1173
• [2.x] Adapt visitors-core to Codama v2 #1172
• [2.x] Adapt nodes to Codama v2 #1171
• [2.x] Adapt errors to Codama v2 #1167
• [2.x] Rename node name attribute to identifier #1162
• [2.x] 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

Ports @codama/dynamic-codecs to the v2 node model. The visitor is restructured around three helpers (visitLinkedNode, getCollectionCodec, applyTransform), with the per-type-node transforms applied via an interceptVisitor layer that wraps every visit — so transforms compose innermost-first and link-node transforms naturally stack on top of the linked type's own. Integers now always decode to bigint (with a transformCodec shim for the sub-64-bit sizes) and honour endian; float/fixed-point/date-time/duration are added; enums move to { __kind, __discriminator, data } with raw identifiers; sentinelCountNode is wired through to kit's sentinel array size. On the visitors-core side, ProvidedScope gains resolveOrThrow (sharing a resolveOutcome core with resolve), and resolveWithin now tracks which injection dead-ended so the new CODAMA_ERROR__INJECTED_VALUE_NOT_PROVIDED names the innermost culprit rather than the enclosing struct.

The code is clean and the test coverage is solid — I checked the pre/post-offset hex expectations against kit's offsetCodec semantics (preOffset in the post-offset scope is the original offset, so ff000000aa00 for the preOffset strategy is right), and the injected-value tests cover the standalone-type, linked-from-instruction, and parent/child-instruction cases well.

Things I verified

• interceptVisitor binds this to the outermost visitor, so visit(node.type, this) inside baseVisitor re-enters through the transform/scope/stack layers at every depth. Transforms on nested fields, enum variant data, count prefixes, etc. all get applied.
• getNodeValueCodec opens frames only for ancestor instructions and leaves the last path node to recordProvidedScopeVisitor, so there's no double-push when the path ends at an instruction.
• Frames are ordered root→leaf and ProvidedScope.find searches innermost-first, so shadowing works as documented.
• resolve's docblock ("all-or-nothing … undefined is returned") is still accurate after the Outcome refactor.

Non-blocking notes

Changeset. I don't see a .changeset/*.md in the file list, and CONTRIBUTING.md asks for one per user-facing concern. This PR renames the public getNodeCodec/getNodeCodecVisitor API, adds ProvidedScope.resolveOrThrow to @codama/visitors-core, and adds an error code to @codama/errors. If the 2.x branch is covered by a seeded major changeset and per-PR entries aren't wanted during candidacy, ignore this — just flagging in case it was an oversight.

Enum links inside provided values resolve against the injection site. In values.ts, visitInjectedValue visits the resolved node with this, so any enumValueNode inside a provided value has its enum link resolved via the current stack.getPath(). If an instruction in program A provides an enumValueNode('foo', …) (relative link) and the injection lives inside a defined type in program B that A links to, the link would resolve in B, not A. Probably fine to leave — it's an edge case and the value author can use an explicit program on the link — but worth knowing for whoever wires up dynamic-parsers/dynamic-instructions on top of this.

Map keys. Maps encode via Object.entries, so keys always reach the key codec as strings. Kit's number encoders coerce, so an integer-keyed map should still round-trip as { "1": … }, but there's no test for non-string keys. Not a regression from v1 as far as I can tell.

For subsequent reviewers

• The enum value shape (data nesting, raw identifiers instead of pascalCase) is a wire-format change for anything consuming this package's output. dynamic-parsers, dynamic-instructions and dynamic-client only get @solana/* bumps here, so presumably their own v2 adaptation follows in later PRs of the stack — worth confirming nothing in those packages currently depends on the old __kind: 'PascalCase' / spread-fields shape.
• pnpm-lock.yaml churn is the 8.3→8.4 @solana/* bump; I didn't audit it line by line.