Ann,

The 20 source code references, where for the new built-in function
EXTRACT! Sorry I should have mentioned the name ;-(

I would have expected the same to be true of any new function added to
the engine. Extract should be appropriate guidepost, since I viewed it
as a fair representation of most UDF functions - a keyword which accepts
values and returns a result.

I will also have a look at the implementation of AVG, thanks for the
pointer.


Sean

-----Original Message-----
From: Ann Harrison [mailto:harrison@...]
Sent: Wednesday, November 08, 2000 4:07 PM
To: IB-Architect@egroups.com
Subject: RE: [IB-Architect] UDF Functions

>
>From: Dalton Calford [mailto:dcalford@...]
>
>3.) Is there a reason why the common udf's (or a subset of them) can

not

>become part of the new version?

At 03:31 PM 11/8/2000 -0500, Leyne, Sean wrote:

>I have done some research on this subject, the lack of built-in
>functions has been a pet peeve of mine since I started using IB.
>
>Aside from the fact, that it appears that the definition and
>implementation of the functions is spread across some 20 source code
>files, there is no reason why this can't be done.

Ummm maybe that's why the intention was not to use the existing
UDF mechanism to define built-in functions. The Novell thing
was pretty much a cheap way to get some additional capabilities
to that platform.

Look at the implementation of avg, and imagine something like
a blr_internal_function operator that took a list of blr_function_x
arguments that identified the function.

Regards,

Ann