Using Key-exeqry to populate block, next_record draws FRM-41009

Question

I am using web forms 6.0.8.18.3 (6i patch9) and 10g database
I am making a new form from another very similar screen
but presenting format as child to parent view of data
(reversing similar screens parent to child format)
The new screen has 2 t tabs, 2 blocks per tab(not master/detail)
I want to preserve the prior form’s same
Query_data_source_name block property selects (if possible)
(This preserves the enter-query/execute query on the block(works fine)
I want to populate a tabular(table based) block(query only)
based on user entering a number (child) returning one or many rows
(the first field is enterable (control number)and the rest are display)
goal: so the user can query by number field and get child/parent data.
I can do a direct ‘select into’ the fields and create one record ok.
But fails if to_many_records, returns(error)
That’s why I need the cursor/loop/fetch/next_record set up
To return many records.
In a proc called by a key_exeqry trigger to populate the block,
I am using next_record in my loop but get FRM-41009
How can I achieve this via a key_exeqry?
if not
is there a way to ‘break away from the key_exeqry’,
do the loop(in a place where I will not draw a error)
based on passed parameters to populate the block?
Here is my block/Key-exeqry logic --(not ideal)
If entered field is null then
Null; --allows f7/f8 full query usin query data source name(ok)
Elsif not null
Do validate select-–validate number with select
If no data found error message(ok)
If to many rows then
Call multi record loop proc
Do the multi record procedure(program unit)cursor select/loop/fetch(fails /no next_record allowed!)
I am rotating through the loop and land on last record
(The problem is the next_record is not allowed.)
If field not null and a flag not on(=N)
Call single_record_populate procedure(select into) (works fine-ok)
End if;
I saw A. Weiden post that ‘key-exeqry fires only once for the query,
so you won’t have a chance To do a per-record operation in it’.
(so there is a better way)
But Is there any way to do this--do a block populate(with loop) based on a key-exeqry trigger?
(75% of srcreen requirements work ok so far)
Thanks.
here is actual KEY-EXEQRY code
declare
v_parent_mode_id number;
v_ep_num varchar2(4);
v_ep_description varchar2(120);
v_cd_Id number;
v_cd_capture_percent number(6,3);
dummy_alt number;
v_error_flag varchar2(1);
action varchar(80);
p_out_parent_mode_id number(10) := v_parent_mode_id;
p_out_cd_Id number(10) := v_cd_Id;
p_out_ep_num varchar2(4) := v_ep_num;
p_out_cd_Id_Num varchar2(4) := :emission_points.cd_id_num;
begin
v_error_flag     := 'N';
if (:emission_points.cd_id_num is null) then
execute_query;
v_error_flag := 'Y';
elsif
(:emission_points.cd_id_num is not null) then --validate check if entry(chld) exists under parent
begin
     select parent_mode_id, cd_id, cd.capture_percent, ep.num
into v_parent_mode_id, v_cd_id, :emission_points.capture_percent, v_ep_num
FROM EMISSION_POINTS EP,
EP_MODES EPM,
CAPTURING_DEVICES CD,
CONTROL_DEVICES CDT
WHERE EP.ID = EPM.EP_ID AND
EPM.ID = CD.PARENT_MODE_ID AND
CD.CD_ID IS NOT NULL AND
CD.PARENT_MODE_ID IS NOT NULL AND
CD.STACK_ID IS NULL And
cd.fac_id = :FACILITIES.ID and
cd.cd_id = cdt.id and
cdt.num = :emission_points.cd_id_num;
exception
when no_data_found then
disp_warn('Control number '||:emission_points.cd_id_num||' does not exist under parent points for id_number '
||:facilities.scr_fac1);
raise form_trigger_failure;
v_parent_mode_id := null;
v_cd_id := null;
:emission_points.capture_percent := null;
v_ep_num := null;
when too_many_rows then
v_error_flag := 'Y';
--process for multiple records returned
populate_emission_points(p_out_parent_mode_id, p_out_cd_Id, p_out_ep_num, p_out_cd_Id_Num);
end;
end if;
if (:emission_points.cd_id_num is not null) and v_error_flag = 'N' then
     DISP_WARN(':emission_points.cd_id_num=>'||:emission_points.cd_id_num||' v_error_flag=>'||v_error_flag);
--process for single record returned
emission_points_record;
end if;
end;
Edited by: Doug Galayda on Apr 16, 2009 12:36 PM

Answer 1

1. This is really why I did it in pieces:
2. Whole query does work
1.The form previously has a existing form-clause-query
When I put a whold valid select in
It attaches ‘the new from clause qeury’ on to the select.
Select item1, item2, item3
From(select item1, item2 <<my new from clause
I see this effect in a :system.last_query display.
I can’t ignore the properties already there.
They add themselves to(or around) what ever I put into
A new from clause query statement.
So since I have a existing from clause query that says
Select id, fac_id, num, parent_mode_id, cd_id, description, capture_percent
From
What ever I put into a SET_BLOCK_PROPERTY becomes this
Select id, fac_id, num, parent_mode_id, cd_id, description, capture_percent
From( my new set_block_property here
In other words if you had a existing select from clause
And you put a whole select in your substitute from clause query
And you put a :system.last_query display in your key-exeqry immediately
After your set_block_property command
You are in fact replacing the from clause with that statement
(what you put in the statement will follow a ‘from’.
I thought I was doing the correct thing:
The pieces added up to looking correct to, but did not work!
(I can’t get the real :bind value passed in to the join.)
2. You are right when I put the whole query in there
It does work, but only when I use a literal ‘’0031’’ in double quotes,
but I cannot get the bind variable to be ‘seen’
In the set_block_property
I can see my user entered field value as 0031 in a display before the
Set block property is fired.
I need a value of 0031 in a bind variable for the query to work.
And that’s what I am working on.
SET_BLOCK_PROPERTY('EMISSION_POINTS',QUERY_DATA_SOURCE_NAME,
'(SELECT C.ID, A.ID EP_ID,A.FAC_ID,A.NUM,A.DESCRIPTION,C.PARENT_MODE_ID,
C.CD_ID,C.CAPTURE_PERCENT,C.CAPTURE_METHOD,C.DATE_TESTED
FROM EMISSION_POINTS A,
EP_MODES B,
CAPTURING_DEVICES C,
CONTROL_DEVICES CD
WHERE A.ID = B.EP_ID AND
B.ID = C.PARENT_MODE_ID AND
C.CD_ID = CD.ID AND
C.CD_ID IS NOT NULL AND
C.PARENT_MODE_ID IS NOT NULL AND
C.STACK_ID IS NULL and
cd.num = ''0031'')');
Steve Cosner replied to my first post
How can I pass a variable into a Query Data Source Name select's join?
Use one of these:
and cd.num = :Blk_name.Pass_cd_Id_num
and cd.num = :Parameter.Pass_cd_Id_num
and cd.num = :Global.Pass_cd_Id_num
Any one of the above would work. I would use either a hidden item in a control block, or a form parameter.
The sql select you build in the query-data-source name is passed
directly to Oracle for execution, and PL/SQL variables cannot be "seen" by the database.
Instead, you pass a bind variable (note the leading colon),
and Forms then knows to pass the value or location of the bind variable so Oracle can use it.
SET_BLOCK_PROPERTY('EMISSION_POINTS',QUERY_DATA_SOURCE_NAME,
'(SELECT C.ID, A.ID EP_ID,A.FAC_ID,A.NUM,A.DESCRIPTION,C.PARENT_MODE_ID,
C.CD_ID,C.CAPTURE_PERCENT,C.CAPTURE_METHOD,C.DATE_TESTED
FROM EMISSION_POINTS A,
EP_MODES B,
CAPTURING_DEVICES C,
CONTROL_DEVICES CD
WHERE A.ID = B.EP_ID AND
B.ID = C.PARENT_MODE_ID AND
C.CD_ID = CD.ID AND
C.CD_ID IS NOT NULL AND
C.PARENT_MODE_ID IS NOT NULL AND
C.STACK_ID IS NULL and
cd.num = :emission_points.cd_id_num)');
The bind variable :emission_points.cd_id_num in the above
set block property statement appears as its :bind variable name :emission_points.cd_id_num
This fails with frm-40505.
The key problem now it to get the :bind value to be ‘seen’ in the statement.
Thank you so much.
I appreciate your time and help.