Skip to content

fix: Padding now added to Arrow file marker and RecordBatches are being written with correct alignment #95

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Oct 14, 2025
Merged

fix: Padding now added to Arrow file marker and RecordBatches are being written with correct alignment #95

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1b718f4
fix: Correct padding now added to arrow filemarker. Recordbatches ar…
willtemperley Oct 13, 2025
a190494
fix: Pad schema message instead of record batch start. Calculate reco…
willtemperley Oct 14, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
22 changes: 18 additions & 4 deletions Arrow/Sources/Arrow/ArrowWriter.swift
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -144,12 +144,25 @@ public class ArrowWriter { // swiftlint:disable:this type_body_length
withUnsafeBytes(of: CONTINUATIONMARKER.littleEndian) {writer.append(Data($0))}
withUnsafeBytes(of: rbResult.1.o.littleEndian) {writer.append(Data($0))}
writer.append(rbResult.0)
addPadForAlignment(&writer)
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is for "Padding bytes to an 8-byte boundary" https://arrow.apache.org/docs/format/Columnar.html#recordbatch-message , right?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes - I believe that every time a FlatBuffers message is written it needs to be padded. I think that FlatBuffers messages align to 8 bytes internally but not necessarily where they terminate.

let metadataLength = writer.count - startIndex
let bodyStart = writer.count
switch writeRecordBatchData(&writer, fields: batch.schema.fields, columns: batch.columns) {
case .success:
let bodyLength = writer.count - bodyStart
let expectedSize = startIndex + metadataLength + bodyLength
guard expectedSize == writer.count else {
return .failure(.invalid(
"Invalid Block. Expected \(expectedSize), got \(writer.count)"
))
}
rbBlocks.append(
org_apache_arrow_flatbuf_Block(offset: Int64(startIndex),
metaDataLength: Int32(0),
bodyLength: Int64(rbResult.1.o)))
org_apache_arrow_flatbuf_Block(
offset: Int64(startIndex),
metaDataLength: Int32(metadataLength),
bodyLength: Int64(bodyLength)
)
)
case .failure(let error):
return .failure(error)
}
Expand Down Expand Up @@ -293,6 +306,7 @@ public class ArrowWriter { // swiftlint:disable:this type_body_length
case .success(let schemaOffset):
fbb.finish(offset: schemaOffset)
writer.append(fbb.data)
addPadForAlignment(&writer)
case .failure(let error):
return .failure(error)
}
Expand Down Expand Up @@ -379,7 +393,7 @@ public class ArrowWriter { // swiftlint:disable:this type_body_length
addPadForAlignment(&markerData)

var writer: any DataWriter = FileDataWriter(fileHandle)
writer.append(FILEMARKER.data(using: .utf8)!)
writer.append(markerData)
switch writeFile(&writer, info: info) {
case .success:
writer.append(FILEMARKER.data(using: .utf8)!)
Expand Down