Invalid CSV according to spec
Stiivi opened this issue · comments
The included CSV has floating number of fields per row, sometimes exceeding number of fields defined in the metadata.
Moreover it contains invalid integer values for NumericCode – there is one Nil value (I guess representing an empty field) and some -'s. This can be fixed with changing the field type to string though.
Indeed, and it seems that the last two columns (WithdrawalDate and Remark) are missing/empty for all records.
@peterdesmet not all records - those start being used for the last set of records (the obsolete codes where these are relevant). Solution here is either to add the commas or, perhaps more usefully, to split this data file into 2 parts: currency codes in use and those which are obsolete.
@rgrp Sorry, should have checked further. Personally, I wouldn't split the file in two, as you would have to move records from one file to the other when a currency becomes obsolete. But I don't know the main intended use of this dataset: splitting might have an advantage I don't know.
@peterdesmet the more I think about it the more I incline to a split (you can always just move the code when it becomes obsolete). The advantage is that most people just want the active codes and this will save them having to filter them out. wdyt?
I have no strong opinion one way or the other. Would do what is most userfriendly and not too difficult to maintain.