Subject Re: [IB-Architect] UDF and null
Author Helen Borrie
At 07:05 PM 01-12-00 -0500, you wrote:
>In the long run, the right thing to do with UDF's is to
>replace them (85% confidence).

Oh? why? replace with what?

What does this do for upward compatibility of applications? I was helping
out a few months ago on a project that delegates about 60% of server-side
processing to UDFs; and also constructs objects based on the results of
some complex algorithms within the UDFs. Losing that feature would kill
about two years (so far) of that guy's work, which is a product for
end-users to build their own applications.

>In the mean time, does
>anyone have a suggestion for an alternate interface that
>would allow passing nulls?
>The original design (for those who haven't been here since
>day 1) included three mechanisms for passing values between
>InterBase and a UDF: value, reference, and descriptor. The
>last actually passed a pointer to a structure that included
>the datatype, length, scale, null status, and a pointer to
>the actual data.

What has become of the descriptor option? How I've longed for this !

>That mechanism allowed the very clever (ambitious? overly
>eager? utterly insane?) programmer to write UDF's that were
>data type independent. There has been no great clamor
>for datatype independence from the user base, but there
>is frequent complaint about the absence of null handling.

Reinstate the descriptor option. Is datatype independence such a huge
problem if the UDF language has declarative syntax and good casting support
and the descriptor structure includes vars for input and output datatypes?


All for Open and Open for All
InterBase Developer Initiative ยท