[Qt-interest] Using true database cursors with Qt

[Robert Hairgrove]

John Moran wrote:
> Hi Robert,
> 
>> I believe that QSqlQuery provides exactly the same functionality as a
>> cursor. There might be some things you could do to improve performance
>> on the database side of things, though:
> 
>> (1) proper indexing;
> 
> Naturally, I already index things; My real concern is so-called
> "impedance mismatch" though.
> 
>> (2) fetch pre-built thumbnails instead of entire full-sized images to
>> populate your list;
> 
> They're already thumbnails.
> 
>> (3) call QApplication:;processEvents() within your loop which should
>> help keep the GUI responsive.
> 
> That's what I intend to do, between cursor fetches; The point of the
> cursor is that it encapsulates a result set. I don't have to transfer
> all the results between the database and my app at once. I can fetch a
> bunch of them at a time, and call QApplication:;processEvents()
> between fetches. I think it's pretty doubtful that QSqlQuery is
> transparently using a *native* cursor (the only useful kind - I think
> some python DB drivers use the term cursor without actually using
> native cursors). Surely the docs would say so, and surely I'd be able
> to specify how many records I want to fetch at a time, between calls
> to  QApplication:;processEvents(). This would typically work my
> instantiating a cursor class, which we iterate through. Each iteration
> would return a subset of the cursor's result set, perhaps in the form
> of a QSqlQuery. We'd be using a nested loop. This is how the regular
> C++ postgreSQL API, libpqxx, does it.
> 
> Thanks,
> John Moran

I'm not familiar with PostgreSQL, but what you describe sounds something 
like DB2 or Oracle's array fetch. I don't think that Qt provides that 
functionality.

If you want to fetch only a subset of the data, and you have everything 
indexed according to some scalar value (perhaps a sequential number) 
then you could just partition the range of key values according to how 
many rows you wanted to fetch. That would have the advantage of being 
able to write DBMS-independent code.

But look at the docs for QSqlResult which has examples for PostgreSQL 
and other DBMS's. I think you might be able to do it that way, although 
I never tried it myself.