On Fri, 26 Feb 2010, K. John Wu wrote:

> On the issue of handling string values, the main function to be added 
> would be a function to retrieve the string values, likely something 
> named fastbit_get_qualified_strings, with the following prototype

> const char** fastbit_get_qualified_strings(FastBitQueryHandle  query,
> const char *cname);
>
> For row-wise data accesses, there is already a function named
> fastbit_result_set_get_string.  It should work for any string valued
> columns.
>
> Do you think this might be enough?  Would you have a different suggestion?

that column accessor shoud work. i've been unable to get the add_values() 
stuff to work with TEXT and CATEGORY ibis types (by adding them to the 
switch statement, it dies in tablex::appendString(), i'm still 
investigating.

been hacking capi.cpp in my local tree but there's nothing to share just 
yet. i'm about to throw millions of rows into FB (specifically a sinkhole 
log for malcode infected machines).

> currently, the pointers returned by the fastbit_get_qualified_ttts and 
> fastbit_result_set_get_string are NOT to be deleted by the user code.
>  Instead, the user must call fastbit_destroy_query in order to free
> the memory allocated for those pointers.  Does this work OK for
> python?  Any comments or suggestions to improve this?

seems to but i haven't tried it for long periods of time. i have 
destructors for the Query, ResultSet and FastBit objects that call the 
appropriate cleanup methods.

________
jose nazario, ph.d.                 http://monkey.org/~jose/
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to