Part of #5.
_apply_row_selection (src/daflip/services.py:410) is only called on the non-chunked path (services.py:587). With --input-chunk-size set, --rows 0:100 is accepted and the full file is converted.
Note the non-chunked path already degrades quietly here too: an unparseable --rows prints a warning and continues with the full dataset rather than erroring.
Fix: apply the row window across the chunk stream, or reject the combination explicitly.
Part of #5.
_apply_row_selection(src/daflip/services.py:410) is only called on the non-chunked path (services.py:587). With--input-chunk-sizeset,--rows 0:100is accepted and the full file is converted.Note the non-chunked path already degrades quietly here too: an unparseable
--rowsprints a warning and continues with the full dataset rather than erroring.Fix: apply the row window across the chunk stream, or reject the combination explicitly.