Build a useful investigation brief
Build contrasting source layouts
Create small files with a header, no header, an introductory line, and an all-text first record. Specify expected labels and record counts independently. Include a first data value that resembles a common column name so the comparison does not depend on an unrealistically obvious heading.
Trace the selected row offset
Record the inferred header choice, skipped-line count, and first parsed data row. Compare automatic behavior with the explicit settings supported by the product. A disagreement can isolate inference; matching wrong results may point to how the parser combines header handling with introductory-line skipping.
Follow the confirmed layout
Inspect how the preview’s header and offset choices are passed to the import job. Compare the same file revision and settings at execution. Check whether one stage counts physical lines while another counts parsed records, especially when earlier fields contain quoted line breaks or optional introductory content.
What to carry forward
Produce a minimal layout fixture and the exact row-selection decision that diverges. Keep label inference, line skipping, and setting persistence distinct in the brief. A correction should preserve the first true record and avoid accepting headings as data, rather than merely making the preview labels look familiar.
Technical background: PostgreSQL documentation.
Keep the decision with the work.
Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.
Create an account See the product workflow