Repository navigation
No lowercase on key name in bulkloader #48
Description
Activity
Yeah, so in thinking about this some more,
Recordshould probably have a couple of new properties likesource_id_column,source_id_value, and maybe evensource_filename.Don't think any of that is necessary. Need a compelling use case. But, definitely need an unaltered source_id somewhere.
Don't think any of that is necessary. Need a compelling use case. But, definitely need an unaltered source_id somewhere.
Right, so a publisher downloads record annotations and now needs to link them back to their source data. If they've published records with different
source_idcolumns, then they'll need to know whichsource_id_columnbelongs to which record so that they can make the join on that column with thesource_id_value.Not a realistic scenario. source_id is the collection-level unique identifier, by definition - the closest thing to stable information in the source data. If they are changing source_ids - all bets are off, and knowing what file and field it came from originally doesn't help to re-integrate annotations with the original data.
So @tucotuco and I face-to-faced this issue a bit. The basic use case to support here is publishing CSV files that are generated from different data sources having different
source_idcolumns. This comes into play when publishers want to link these records in VertNet back to their data sources. I think we agree that it's unclear how useful this would be. So, we're tabling the issue for a bit while we think on it and get some feedback from others on the team.+1
Right now the source_id is lowercased before using it to create the key name. Since this is the only record we have of the source_id, it should not be altered.