This e-mail and any files transmitted with it are the property of the
London Borough of Havering, are confidential, may be subject to legal privilege
and are intended only for the person(s) or organisation(s) named above.
Any unauthorised use, retention, distribution, copying or disclosure
is strictly prohibited. If you receive this email in error, please notify
the sender immediately and delete this e-mail from your system.
WARNING: It is your responsibility to take all necessary steps to
ensure this e-mail and any attachments to it are free from viruses.

If you are worried about being passed a NULL parameter then don't as you
won't. The DBMS will short-circuit expressions in general that would
otherwise need to propagate NULLs.
Regards,
Ian
-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com
[mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of
zhenchen17
Sent: 11 March 2009 08:50
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: [Info-Ingres] how to return a null value in OME?

Hi everyone,
I wonder how to return a null value in OME. The struture
II_DATA_VALUE used in OME is without a member variable to indicate it
is representing a null value. So how can i do if i want to return a
null in OME?
Thanks.
Chenzhen
_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

That is correct but don¡¯t forget to make the advertised length greater by one for the NULL byte.
Regards,
Ian

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com [mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of zhenchen17
Sent: 12 March 2009 01:54
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: Re: [Info-Ingres] how to return a null value in OME?

Hi Karl,
In my case, i don't want a null input but actually i want a null
output.
so what i do is:
1. define fi with a negative result value
2. mark the last byte of the return data to nonzero
Is that right?

Thanks
Chenzhen

On 3ÔÂ11ÈÕ, ÏÂÎç8ʱ31·Ö, Karl & Betty Schendel <schen... (AT) kbcomputer (DOT) com>
wrote:
_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

Hi Ian,
What you said makes me confused.

If i want to return a nullabe integer(i4), what actual length would i
define in FI definiton? sizeof(i4) or sizeof(i4)+1?. Wouldn't DBMS
automatically add 1 byte more when calling the function if it
recognize the function could return a nullable value?

Thanks
Chenzhen

On 3月12日, 下�5时24分, "Ian Kirkham" <Ian.Kirk... (AT) ingres (DOT) com> wrote:

Hi Chenzhen,
You need to say sizeof(i4)+1. This is the convention adopted otherwise the DBMS would be forever checking the sign of the datatype to work out how many bytes needed moving. For a more complicated example returning a 3 byte result 'ABC' in a nullable varchar(10) would be like this:

/varlen\ / used \ / unused \ /NVL
+---+---+---+---+---+---+---+---+---+---+---+---+---+

In this case the data type would be -21 and the length would be 13 (2 byte varchar length, 10 bytes for the data plus 1 byte for null value)

Regards,
Ian

-----Original Message-----
From: info-ingres-bounces (AT) kettleriver...ting (DOT) com [mailto:info-ingres-bounces (AT) kettleriverconsulting (DOT) com] On Behalf Of zhenchen17
Sent: 12 March 2009 13:02
To: info-ingres (AT) kettleriverconsulting (DOT) com
Subject: Re: [Info-Ingres] how to return a null value in OME?

Hi Ian,
What you said makes me confused.

If i want to return a nullabe integer(i4), what actual length would i
define in FI definiton? sizeof(i4) or sizeof(i4)+1?. Wouldn't DBMS
automatically add 1 byte more when calling the function if it
recognize the function could return a nullable value?

Thanks
Chenzhen

On 3月12日, 下�5时24分, "Ian Kirkham" <Ian.Kirk... (AT) ingres (DOT) com> wrote:
_______________________________________________
Info-Ingres mailing list
Info-Ingres (AT) kettleriverconsulting (DOT) com
http://www.kettleriverconsulting.com...fo/info-ingres

On Mar 12, 2009, at 9:02 AM, zhenchen17 wrote:


Nope. You have to use sizeof(i4)+1 in your FI definition.
The DBMS doesn't assume the extra byte for the null indicator.

Karl

Hi Ian
I c.
Thanks very much.
On 3月12日, 下�9时22分, "Ian Kirkham" <Ian.Kirk... (AT) ingres (DOT) com> wrote:

Hi Karl and Ian,
What you said seems have some problems. I tried to return a null
float, so i define fi with -II_FLOAT and the length sizeof(f8)+1. But
when the function is called, DBMS will a return a error code:
"E_AD2005 ADF routine found DB_DATA_VALUE with an invalid length."
But if i define the length only sizeof(f8) , it works fine and return
NULL as i want.

I have no idea what happened, could u take a look and give any reason?
Thanks
Chenzhen

On 3月12日, 下�9时29分, Karl & Betty Schendel <schen... (AT) kbcomputer (DOT) com>
wrote:

On Mar 18, 2009, at 9:12 AM, zhenchen17 wrote:

Did you set the NONULLSKIP flag? I had forgotten this, but you need
to pass NONULLSKIP if the function can return null, whether it can
accept nullable parameters or not. I forget why I did it that way.

Without the NONULLSKIP flag, ad0_fiinit (in adgstartup.c) forces the
result type of the user defined FI's to positive, which would explain
why
you get the AD2005.

If you did set NONULLSKIP, then I'm not sure what is going on, and would
need more details.

Karl

Hi Karl,
Of cuz i set the NONULLSKIP. But it still failed.

I doubt the error comes from the len_check function(adclenchk.c)

case DB_FLT_TYPE:
/* Float datatype, which can either be 4 or 8 bytes long */

/* set default length to 4, in case check fails */
if (adc_rdv != NULL)
{
adc_rdv->db_datatype = DB_FLT_TYPE;
adc_rdv->db_length = 4;
adc_rdv->db_prec = 0;
}

if (adc_dv->db_length != 4 && adc_dv->db_length != 8)
{
/* Precision check is done in adi_encode_... for its callers
** (e.g. PSF), or here for others (such as ABF). */
if (adc_is_usr && adc_rdv != NULL)
{
if (adc_dv->db_length > DB_MAX_F8PREC ||
adc_dv->db_length < 0)
db_stat = adu_error(adf_scb, E_AD2006_BAD_DTUSRLEN, 0);
else
{
/* Do the length to F4 and F8 mapping */
if(adc_dv->db_length <= DB_MAX_F4PREC)
adc_rdv->db_length = 4;
else
adc_rdv->db_length = 8;
}
}
else
{
db_stat = adu_error(adf_scb, E_AD2005_BAD_DTLEN, 0);
}
}
else
{
if (adc_rdv != NULL)
adc_rdv->db_length = adc_dv->db_length;
}
break;

From the code above, it seems the length of float should be 4 or 8.
If I set the length 8+1, the statement:
db_stat = adu_error(adf_scb, E_AD2005_BAD_DTLEN, 0)
is executed and error returned.

Any idea?
Thanks.

Chenzhen

On 3ÔÂ18ÈÕ, ÏÂÎç10ʱ03·Ö, Karl & Betty Schendel <schen... (AT) kbcomputer (DOT) com>
wrote: