fix(javascript): reserve writer capacity for write paths - #3994
Conversation
| // while the cursor advanced, so dump() returned uninitialized tail bytes. | ||
| const fory = new Fory({ compatible: false }); | ||
| const { serialize, deserialize } = fory.register(Type.list(Type.any())); | ||
| const arr = new Array(150000).fill(1); |
There was a problem hiding this comment.
Because every element is 1, writeElementsHeader sets isSame = true, so this test exercises only the aggregate reserve. The per-item mixed-type reserves and the writeDeclared reserve can still be removed without this test failing. Please add a list larger than the initial writer buffer with mixed element types; if writeDeclared is part of this fix, cover the unknown-Struct declared-list reserialization path as well.
| const { serialize, deserialize } = fory.register( | ||
| Type.struct( | ||
| { namespace: "example", typeName: "BigMap" }, | ||
| { m: Type.map(Type.int32({ encoding: "fixed" }), Type.int32({ encoding: "fixed" })) }, |
There was a problem hiding this comment.
This declared int32 map goes through writeSpecificType, so it does not execute the new reserve in MapAnySerializer.write. Please add a Type.map(Type.any(), Type.any()) case large enough to exceed the initial writer buffer, so removing that reserve makes the regression test fail.
2 participants
Why?
BinaryWriter starts with a ~100KB buffer and, for speed, its individual writeXxx methods do no bounds checking. The contract is that callers call reserve(n) first, which grows the buffer if needed. The generated collection write path honors this (it reserves elementFixedSize * count before the loop), but three paths write per-entry data with no reserve at all: the generated declared-map write, MapAnySerializer.write, and CollectionAnySerializer.write.
What does this PR do?
three write paths now reserve writer capacity, matching the existing generated-collection convention (root reserves fixedSize once; per-element loops reserve fixedSize × count; strings reserve internally)
Related issues
AI Contribution Checklist
yes/noyes, I included a completed AI Contribution Checklist in this PR description and the requiredAI Usage Disclosure.yes, my PR description includes the requiredai_reviewsummary and screenshot evidence or equivalent persisted links of the final clean AI review results from both fresh reviewers described inAI_POLICY.md, the Fory-guided reviewer and the independent general reviewer, on the current PR diff or current HEAD after the latest code changes.Does this PR introduce any user-facing change?
Benchmark