You
File a bug for the CSV import failing on large files with a BOM.
Assistant
Created BUG-1531 for the failing CSV import and linked it to the 3 related reports.
- Severity: high, affects imports above 5 MB.
- Repro: import a file with a UTF-8 BOM and quoted newlines.
- Owner: Data Platform, notified in the team channel.
You
Can you find the cause?
Assistant
The parser strips the BOM only after splitting on newlines, so a quoted newline in the first row breaks the header. Fix: strip the BOM before tokenizing.
You
Why only above 5 MB?
Assistant
Files over 5 MB are read in chunks, and the BOM is only handled in the first chunk path. Smaller files take a different code path that happens to skip it.
You
Post an update on BUG-1531 with the cause.
Assistant
Added a comment to BUG-1531 with the root cause and the proposed fix, and set the owner to Data Platform.