Because I'm writing glue code between the SQLite API and a higher level library which provides a standard API across several DBMSs.
In other DBMSs even if a field contains null you can still ask the API what type it 'should' have been if something *had* been put in it. Without this ability, I cant make SQLite work with the library. I don't understand the reasoning for designing SQLite weakly typed like this, it just seems bizarre. On 02/09/10 14:28, Pavel Ivanov wrote: >> My reason for doing this is, if a field is null, I still need to know >> what class it 'should' have been if it had been storing a value. >> > Why do you need that? No matter what you declare field can store any > type of data. And in SQLite there's no "declared storage class". You > are talking either about "affinity", or "declared column type", or > "actual data type stored'. So which one of those 3 exactly you want to > know and why? > > _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users