Hi Carlos, On 30/11/2013 4:27 AM, Carlos Ferreira wrote:
My tables have all the same number of columns ( 1 column only.. of BLOBs.) and Simon Slavin suggested I could use only one big table where I add an extra column to identify the table name.
I would consider searching by an integer ID rather than a string if that is possible.
This seems quite a nice approach, because it does seems to be more memory efficient in terms of disk usage. However my question is the following: Let's assume that TABLE now designates my real tables that can be either SQLITE tables or sub groups of records inside one big real SQLite table I have to load to memory and save to DB groups of these TABLE at the same time ( by saving I refer to update or save the blobs inside each table ).
From the sound of it, you don't (and can't) delete a single row from the table. Is that intentional?
What if faster? Accessing a table in SQLite and updating deleting or adding new records Or Querying the records of one table in such a way that the select records have a field = Table Name..and then adding and updating these records.
My gut feeling is that accessing a smaller table is likely to be faster but a lot depends on the number of records. I don't think you would see much difference for a few thousand or few tens of thousands records.
That said, since you know the exact query that you want to perform, you may want to look into partial indexes as a way to speed up these queries by avoiding a full table scan.
http://www.sqlite.org/partialindex.html Best Regards, Mohit. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
