Umbrella issue. When --input-chunk-size is passed, convert_data delegates to _handle_chunked_conversion and returns at src/daflip/services.py:575-579 before any of the other option handling runs. Several documented flags are accepted on the command line, produce no warning, and have no effect.
This is the failure mode that matters: the command exits 0 and prints "Conversion successful!", so the option looks honoured. On a large file the user only discovers otherwise much later, if at all.
Sub-issues:
A shared consideration across all of these: where an option genuinely cannot be supported in chunked mode, the CLI should fail loudly rather than proceed silently.
Umbrella issue. When
--input-chunk-sizeis passed,convert_datadelegates to_handle_chunked_conversionand returns atsrc/daflip/services.py:575-579before any of the other option handling runs. Several documented flags are accepted on the command line, produce no warning, and have no effect.This is the failure mode that matters: the command exits 0 and prints "Conversion successful!", so the option looks honoured. On a large file the user only discovers otherwise much later, if at all.
Sub-issues:
--compression/--compression-levelignored when chunking--schema-fileignored when chunking--rowsignored when chunking--output-chunk-sizeis accepted and validated but never usedschemacommand reads the entire file for non-CSV formatsA shared consideration across all of these: where an option genuinely cannot be supported in chunked mode, the CLI should fail loudly rather than proceed silently.