bede
commented
2 years ago Thanks @KuzminaAnna
- I'm intrigued by this discrepency as Git blame shows that the function for schema selection has not changed in > 3 months. However this error message is delivering intended behaviour. Because the poor legacy decision to create a different upload CSV schema for paired, unpaired and bam uploads, it is necessary to infer which schema has been provided prior to performing schema-based validation. An empty instrument field prevents inference of the schema and thus fails as a prevalidation check. This error message is accurate and clear if somewhat unwieldy.
- This change was made to increase consistency with other messages
- This is more technically accurate than the old message
- This is more technically accurate than the old message
- This is more technically accurate than the old message
- This is expected behaviour, because today's date is something that changes! If you like I can make it say
today's dateinstead of a changing date literal. - This is because the allowed values are are stored as an unordered set whose display order is stochastic. I will sort these in order to normalise display order
7 EC auto tests were broken as a result of the recent changes in the error messages phrasing. Here's the full list of changes we've discovered so far:
Preconditions: csv with 1 record, where instrument_platform = "" expected error message: "instrument_platform cannot be empty" actual error message: "could not infer upload CSV schema. For Nanopore samples, column 'instrument_platform' must be 'Nanopore', and either column 'fastq' or column 'bam' must be valid paths. For Illumina samples, column 'instrument_platform' must be 'Illumina' and either columns 'bam' or 'fastq1' and 'fastq2' must be valid paths."
Preconditions: csv file without headers expected: "Failed to parse CSV (Index sample_name invalid)" actual: "failed to parse upload CSV (Index sample_name invalid)"
Preconditions: csv with 3 records, where Country 1 = "United States Of America", Country 2 = "FRA", Country 3 = "UK" expected: "United States Of America is not a valid ISO-3166-1 alpha-3 country code" actual: "United States Of America is not a valid ISO 3166-1 alpha-3 country code"
expected: "UK is not a valid ISO-3166-1 alpha-3 country code" actual: "invalid region (ISO 3166-2 subdivision) for specified country"
expected: "One or more regions are not valid ISO-3166-2 subdivisions for the specified country" actual: "invalid region (ISO 3166-2 subdivision) for specified country"
expected: "!oxfords is not a valid ISO-3166-2 subdivision name" actual: "invalid region (ISO 3166-2 subdivision) for specified country"
expected: "One or more regions are not valid ISO-3166-2 subdivisions for the specified country" actual: "invalid region (ISO 3166-2 subdivision) for specified country"
expected: "One or more regions are not valid ISO-3166-2 subdivisions for the specified country" actual: "invalid region (ISO 3166-2 subdivision) for specified country"
Preconditions: csv with 1 record, where collection_date is empty. expected: "collection_date must be in format YYYY-MM-DD between 2019-01-01 and 2022-11-15" actual: "collection_date must be in format YYYY-MM-DD between 2019-01-01 and 2022-11-21"
Preconditions: csv with 3 records, where Control 1 = "", Control 2 = "None", Control 3 = "" expected: "None in the control field is not valid; field must be either empty or contain the one of the keywords negative, positive" actual: "None in the control field is not valid; field must be either empty or contain the one of the keywords positive, negative"