Dynamic physical table name vs. Cache

Hello, Experts!
I'm facing quite an interesting problem. I have two physical tables with the same structure but with a different data. Requirement is to show same reports with one or another table. Idea is to have dynamically changed physical table name with session variable usage. Session variable can be change in UI so it was working until cache was turned on. When cache is turned on logical statements sent to OBI backend are the same even for different values of session variable that stores physical table name. Once cache is populated every users will get values from cache. This is possible source of discrepancy because some users might run reports with tableA values and some with tableB values.
Are there any options to set OBI to use data related to proper physical table name (i.e. accordingly to session variable value)? Model clone is not an option because it will be way to hard and complex to maintain both, beside same reports need to work sometimes with one table name and sometimes with other...
PS. Cache is set to be common for all users.
Lucas

thank you, I've found another way to make it running. In fact there are two ways of doing it: filter LTS and have all data filtered from single table with session variable or use fragmentation content also with session variable.
Now tricky part is to set variable from UI, currently I'm using issue raw sql: call NQSSetSessionValue( 'String SV_SIGNOFF=aaa;' ) but I have to figure out how to change session non system variable value without need of administrator user rights.
There is GoURL method, but it's not working...
2. Add In ORACLE_HOME/bifoundation/web/display/authenticationschemas.xml
<RequestVariable source="url" type="informational" nameInSource="lang"
biVariableName="NQ_SESSION.LOCALE" />
inside the top <AuthenticationSchemaGroup> </AuthenticationSchemaGroup> tag

Similar Messages

Dynamic Physical File names on AL11

Hi,
I have the following requirement.
I have almost 25 files to be exported onto AL11 using OpenHub from various infoproviders and I used the "Logical File name" option to do it.
I would like to have the physical file names as " ZINFOPROV_<DATE>_<TIME> " and Infoprovider tech. name should get populated dynamically somehow.
I was looking at the following thread for the same req : Re: Open Hub File name Change
I defined my logical file name definition as
Logical file    Z_SALES
Name            Used for SALES
Physical file   ZINFOPROV_<DATE>_<TIME>
Data format     ASC
Applicat.area   BW
Logical path    Z_SALES
I would like this ZINFOPROV populated dynamically.
Any have any inputs/suggestions.
Regards,
Kumar

Hi Kumar,
You have a very good question and also shown the viewer of your post the way forward. Well i got your answer from your question only. Considering <DATE> as one kind of literals, you need a literal to identify the infoprovider as one part of your filename. Now my answer is not direct resolution but it may lead you to your answer. Here are the list of other literals that might be of use to you.....
Extract from the help.......
Physical path name
Platform-specific physical path under which files are stored. It most contain the reserved word <FILENAME> as a placeholder for the file name. IT can also contain further reserved words (see below).
Use
The physical path name is used by the function module FILE_GET_NAME at run time to create a complete platform-specific file name.
Procedure
Specify a path. You can include the following reserved words in pointed brackets. These are replaced with current values at runtime.
Reserved Word    Replacement Text
<OPSYS> Operating system in call
<INSTANCE> R/3 application instance
<SYSID> R/3 application name in SY-SYSID
<DBSYS> Database system in SY-DBSYS
<SAPRL> R/3 release in SY-SAPRL
<HOST> Host name in SY-HOST
<CLIENT> Client in SY-MANDT
<LANGUAGE> Log on language in SY-LANGU
<DATE> Date in SY-DATUM
<YEAR> Year in SY-DATUM, 4-character
<SYEAR> Year in SY-DATUM, 2-character
<MONTH> Month in SY-DATUM
<DAY> Day in SY-DATUM
<WEEKDAY> Day of the week in SY-FDAYW
<TIME> Time in SY-UZEIT
<STIME> Hour and minute in SY-UZEIT
<HOUR> Hour in SY-UZEIT
<MINUTE> Minute in SY-UZEIT
<SECOND> Seconds in SY-UZEIT
<PARAM_1> External parameter 1
<PARAM_2> External parameter 2
<PARAM_3> External parameter 3
<P=name> Name of a profile parameter (see Report RSPARAM for valid values)
<V=name> Name of a variable (stored in variable table)
<F=name> Return value of a function module Naming convention for this function module: FILENAME_EXIT_name
<Y=name> Return value of a function module Naming convention for this function module: Y_FILENAME_EXIT_name
<Z=name> Return value of a function module Naming convention for this function module: Z_FILENAME_EXIT_name
Examples
Local directory for temporary files under Windows NT: D:\GLOBAL\ARCHIVE\<PARAM_1>\<FILENAME>
Local directory for temporary files under DOS:C:\TMP\<FILENAME>
Local directory for temporary files under UNIX: /usr/<SYSID>/local/tmp/<CLIENT>/<FILENAME>
Dependencies
The function module for the run-time variable <F=name> must meet the following requirements:
Name
The name must begin with "FILENAME_EXIT_".
Parameter
An export parameter with the name "OUTPUT" must exist. No structure may exist for this parameters.
Import parameters are only supported if they have default values.
Table parameters are not supported.
Valid example: FILENAME_EXIT_EXAMPLE.
I hope this leads you to your resolution.
Regards
Raj
Edited by: Rajesh J Salecha on Oct 9, 2009 1:48 PM - Not able to make it properly formatted...
Edited by: Rajesh J Salecha on Oct 9, 2009 1:54 PM

Dynamic DB table name in insert statement

In the following command i was able to build <dyn_table> dynamically but how do we do that for <dyn_wa>. table name should be populated dynamically here.
for ex: pa0001, pa0002, .... and I have these names in an IT table
INSERT <dyn_wa> FROM TABLE <dyn_table>
ACCEPTING DUPLICATE KEYS.
Pls help...Thanks

Take it as charater type.
This code works for me.
TABLES : ztest123.
DATA : tab_name(10) TYPE c.
tab_name = 'ZTEST123'.
DATA : wa LIKE ztest123.
wa-name = 'Mahesh'.
INSERT (tab_name) FROM wa.
Thanks
Mahesh

Dynamically assigning table name in select statement

how can i assign the table name dynamically in Select statement?
i tried following code
create or replace procedure proc1
as
x varchar2(100);
y varchar2(10);
begin
x='UNIT_MASTER';
execute immediate 'select unit_code into y from x where
rownum=1';
dbms_output.put_line(y);
end;
the procedure is created but when i execute the procedure the
error is shown in the execute immediate statement

Do the following :
Create or replace procedure pro1 as
x varchar2(100);
y varchar2(10);
begin
x := 'UNIT_MASTER';
EXECUTE IMMEDIATE 'select unit_code from '||x||' where rownum
= 1' INTO y;
dbms_output.put_line(y);

Dynamic physical table source schema

Hi All,
Is it possible to dynamically change the physical schema name ?
I am in process of creating a report, but the report should shows the data based on logged in user.
NOTE: There is a one to one mapping between BI user and oracle database schema.
Any advice is greatly appreciated.

Hi Venky ,
I think proxy authentication is what I want.
In my case BI Server's has fixed connection to 'BI_Account' username.
I have created oracle database account's ( say account1 and account2 . I already have BI Server accounts with same name) which has 'connect through' privileges.
I tested in sqlplus that it when user connect to BI_Account[account1]/BI_Account@my_db it is being connected to 'account1' schema, which is exactly what I want.
How do I replicate this in OBIEE ?
what I have to do?
How do I tell BI Server that 'BI_Account' is a proxy account ?
How do I tell BI Server that when user 'account1' get the data from account1 schema ?
What else I have to do ?

SQL or PL/SQL : dynamically insert table name in a SQL Statement

Hi,
We have a strange requirement - we need to dynamically use the table names in a SQL Query. E.g:
select * from :table_name
The table_name will be chosen from a list. I have tried this in SQL as well as PL SQL - but, I have been unsuccessul so far.
Can you guys please help me solve this puzzle ?
I hope I have explained my quesion clearly - if not, please do let me know if some more details are necessary.
Regards,
Ramky

The following is the anonymous block that im using in a report in HTMLDB. My problem is Line Number 9. The bind variable contains the chosen table name at
the run time.
Variable "qry_stmt" contains the query to be returned, so that result set for that query will be displayed in the report.
If I hard code the table name(rather that passing it through bind variable) in the
qry_stmt string, Im getting the result sets for that query. But if I pass through
bind variable at run time, its still generating the string correctly( im printing
using a print statement at line number 14). But its returing the following report
error
report error:
ORA-01403: no data found
Please advice/help me in this.....
declare
qry_stmt varchar2(1000);
p_table varchar2(30) := 'EMP';
P_ENAME varchar2(1000);
begin
IF :p2_TABLE_NAMES IS NOT NULL THEN
qry_stmt := 'select * from '||TRIM(:P2_TABLE_NAMES); -- Line Num 9
execute immediate qry_stmt; --into P_ENAME;
ELSE
qry_stmt := 'SELECT 1 FROM dual ';
END IF;
htp.p(qry_stmt);--Line Num 14
return qry_stmt;
EXCEPTION WHEN NO_DATA_FOUND THEN
NULL;
end;
Thanks and Regards,
Ramky

How to dynamic set table names

In our application, the table name can be set at runtime by the customer. So is there a way in Toplink JPA to set table name at runtime (thru annotation or orm.xml) or programatically? Thanks.

Our application will host several domains at once. Each domain will get its own set of database tables for each module it uses. We keep them seperate by using a unique prefix. An example would be domainA.com's blog post table would be called domainA_blogPost, while domainB.com's will be domainB_blogPost. Both tables are the same, except for the name. The name needs to be specified at runtime since the central service controller tells the rest of the application which domain it is servicing and thus which set of tables to use. We will be adding domains dynamically all the time and we can't afford to create new entity sets and recompile the back end for every domain we add. I have been looking at the API and wonder if this is as simple as implementing a solution in our application, or am I going to have to tweek the implementation of toplink?

Physical table name - abstract table name mapping in CMP Bean

Hello,
I am using a EJB 2.x bean deployed on a SJSAS 9 server. EJB has problem associating the EJB to the actual table
Details are as follows:
EJB name: AccountCMP
Actual tablename : account
Error
ejb store called
IM: postInvokeentity3.AccountBean145996470_ConcreteImpl@d6be89
EJB5071: Some remote or transactional roll back exception occurred
com.sun.jdo.api.persistence.support.JDODataStoreException: JDO76400: Got a JDBC SQLException while executing the SQL statement:
insert into "ACCOUNTCMP"("DESCRIPTION", "ACCOUNTNAME", "BALANCE", "ACCID") values ( ?, ?, ?, ?).
Please examine the SQLException for more information.
NestedException: java.sql.SQLException: ORA-00942: table or view does not exist
FailedObjectArray: [entity3.AccountBean145996470_JDOState@5332ca]
Please note that the insert statement has the ejb-name (I think so, as my CMP EJB is called ACCOUNTCMP) instead of the actual tablename (account).
I do have the account.dbschema file and sun-cmp-mappings.xml in the NB 5.5 project, but the mappings.xml file is not included in the jar by NB. Is that expected, or an problem.
sun-cmp-mappings.xml
<sun-cmp-mappings>
<sun-cmp-mapping>
    <schema>account</schema>
    <entity-mapping>
      <ejb-name>AccountCMP</ejb-name>
      <table-name>account</table-name>
      <cmp-field-mapping>
ejb-jar.xml
<enterprise-beans>
        <entity>
            <ejb-name>AccountCMP</ejb-name>
            <local-home>entity3.AccountLocalHome</local-home>
            <local>entity3.AccountLocal</local>
            <ejb-class>entity3.AccountBean</ejb-class>
            <persistence-type>Container</persistence-type>
            <prim-key-class>entity3.AccountPK</prim-key-class>
            <reentrant>false</reentrant>
            <cmp-version>2.x</cmp-version>
            <abstract-schema-name>account</abstract-schema-name> Thank you for your help.

Hello,
I found out the issue. sun-cmp-mappings.xml file is mandatory, and dont know why, netbeans excludes this file from the jar. I had to edit project.properties file to negate this exclusion.
Another problem was that my table did not have a primary key (and hence by schema file did not have it also). CMP bean code generator did not like this.
Once I fixed these two problems, my CMP started working fine.
Thank you all

Dynamically pass Table name, Column Name and rownum to function

Hi Guys
I wanted to pass the table name, column name and rownum to function and get the relevant value for it. Please guide me to achieve this task
Thanking You
Regards
Lakmal

Thanks,
Here is my test function
CREATE or replace FUNCTION GET_COLUMN_VALUE (tab_name VARCHAR2,column_name VARCHAR2) RETURN varchar2 AS
strsql varchar2 (500);
ColVal varchar2;
BEGIN
strsql:='select '||column_name||' from '||tab_name|| ' Where rownum = 1' ;
EXECUTE IMMEDIATE strsql into ColVal;
RETURN ColVal ;
END;
Message was edited by:
Lakmal Marasinghe

Dynamic BD table name with inner join.

Hi experts,
I'm trying to do something like this:
if syst-datum(4)   > s_budat-high(4) or
   syst-datum4(2) > s_budat-high4(2).
tabname = 'MBEWH'.
else.
tabname = 'MBEW'.
endif.
select m1matnr m1mtart m3~bwkey
from ( mara as m1
      inner join marc as m2
        on m1matnr = m2matnr
      inner join (tabname) as m3
        on m1matnr = m3matnr and
           m2werks = m3bwkey )
where m1~matnr in s_matnr and
        m1~lvorm in s_lvorm and
        m1~mstae in s_mstae and
        m2~werks in s_werks.
endselect.
I get the following error: Incorrect expression "1000 %_#T953" in WHERE condition.
Should I forget the inner join with one dynamic table or is it possible to do?
Thanks in advance, best regards,
André Costa

Hi experts,
I'm trying to do something like this:
if syst-datum(4)   > s_budat-high(4) or
   syst-datum4(2) > s_budat-high4(2).
tabname = 'MBEWH'.
else.
tabname = 'MBEW'.
endif.
select m1matnr m1mtart m3~bwkey
from ( mara as m1
      inner join marc as m2
        on m1matnr = m2matnr
      inner join (tabname) as m3
        on m1matnr = m3matnr and
           m2werks = m3bwkey )
where m1~matnr in s_matnr and
        m1~lvorm in s_lvorm and
        m1~mstae in s_mstae and
        m2~werks in s_werks.
endselect.
I get the following error: Incorrect expression "1000 %_#T953" in WHERE condition.
Should I forget the inner join with one dynamic table or is it possible to do?
Thanks in advance, best regards,
André Costa

Dynamic name for the physical table

Hi Guys,
How to setup dynamic names for the physical table? Where it is useful?*
Pls help me out on this.
thanks

Check this similar post which might be of help dynamic physical table source schema
Cheers,
KK

Physical Table (Dynamic Name) x Physical Table x Alias

Hello,
We are customizing the standard OBI Applications .rpd. but this question is in reference to OBIEE, hence the post here.
I have no problems with Physical Tables and Alias in the Physical layer.
It just happens that the standard OBI Applications .rdp has some other 'physical tables' - the one for: Dim_W_PARTY_ORG_D_Customer that in its properties uses a Dynamic Name pointing to a variable that contains W_PARTY_ORG_D.
There are other similar cases in the vanilla .rpd
I fail to understand why sometimes (most of the time) the vanilla .rpd uses Alias and in these few cases use this strange Physical table. Pls. anyone would care to comment ?
As we expand W_PARTY_ORG_D with new columns... these new columns are NOT automatically part of the Dim_W_PARTY_ORG_D_Customer as it happens in an Alias definition.
Txs. for any comments.
Antonio

Hi Antonio,
From the 11g documentation (http://docs.oracle.com/cd/E14571_01/bi.1111/e10540/physicallayer.htm#i1005872) ;
Select this option to use a session variable to specify the physical table name, similar to catalog and schema objects. This option is available for non-multidimensional data source tables when you select a table type of Physical Table.
You might want to choose this option if you have a multi-tenancy implementation and you want to define a separate physical table name for each customer. Another example would be to select between primary and shadow tables that are valid at different times in your ETL cycle. In both cases, you can assign session variables to dynamically select the appropriate table.
Hope this helps.
Daan Bakboord
http://obibb.wordpress.com

'use dynamic name' and 'caching properites' options for alias table...

Hello everybody,
can anybody please explain 'use dynamic name' and 'caching properites' options for alias tables...
Thanks...
eagerly waiting for a response..
Vijay

You want to create dynamic target table name right?
You can refresh a variable like #GET_SESSION
#GET_SESSION= SELECT <%=odiRef.getSession("SESS_NAME")%> FROM DUAL
My tmp table name like TMP_#GET_SESSION
then in your package refresh #GET_SESSION variable and you can use it.
I hope this can be helpful
Thanks

Regarding physical tables

Hi,
please give me the physical table names which can hold the following:
1. Transfer structure routines
2.CMOD exits
3. Code for Virtual KF and Characteristics.
Thanks,
Vijaya

Hi..
For CMOD Exits tables is MODSAP.
the Remaining two are used in BW .. so better find with any BW consultant.
<b>REWARD IF HELPFUL.</b>

Physical tables

Hi,
please give me the physical table names which can hold the following:
1. Transfer structure routines
2. Code for Virtual KF and Characteristics.
Thanks,
Vijaya

Thanks!

Dynamic physical table name vs. Cache

Similar Messages

Maybe you are looking for