Modifying the cube model & others doubts

Hello Experts,
I have a few doubts about certain BW30B concepts which I would like to seek help from experts in this very impressive forum.
A. From which ODS table does Init/Delta process use to load into connected data target?
B. When data model is changed to add a new attribute, must the Transfer Structure/Data Source be changed as well to include it, just as comm structure, transfer & update rules are modified?
C. To add a new characteristic into the model, I deleted existing cube data & added a new characteristic to this cube to avoid copying existing queries. A new cube is created with existing cube as template. Existing cube data is loaded into this new cube. Then, I adapted txfr & update rules and infosource. One way to complete the change is to load historical data for only the new characteristic from source system and then reload existing data from the new copy cube. The other way is simply to load all history from source system into emptied modified existing cube. I would like to clarify on the steps for the earlier method I mentioned. I cannot understand how data for only the new char can be loaded and then reload from copy cube back into modified, existing cube. Please help to clarify on this method.
D. When adding a new keyfigure, datasource is changed accordingly but its not stated that for char change, data source also need to be changed accordingly. I would think it needs to be changed too for any char / kf added to the model. Please could you confirm this with me.
E. For aggregates, how many maximum characteristics can be used as line item dimensions, not inclusive of data package,time/unit dimensions? Is it that only unit is optional dimension? Please confirm this with me as well.
F. Under what condition must the inventory marker not be updated?
Would be greatful for any useful help given.
warmest regards,
Alfonso Spinelli

Hi Alfonso,
Some answers for you:
A. Data is taken from the Change log table.
B. It depends...if you need to get the data fro the new attribute from the sourec system, then yes the data source and transfer structure will change. If the new data is being populated by reading some values that are already in BW tables, then you would not need to change the datasource. The you would change the comm structure and update rules.
C. The method you are talking about, i.e. create a copy of the cube, move data from original to copy and then reload back to original...is used to preserve the data. If you add a new char to an existing dimension, you need to delete the data of the cube. So this method is the workaround, because you have parked the data and then would not need to reload from the source. However, if you are adding a new char to a new dimension, you do not need to delete the data. Of course for the ne char, you will have data only from this point onwards. If you need all history data for the new char, you would need to reload from the source.
D. The answer to this is similar as point B. If you have the data for that char and can reuse it in your cube load (by coding in your update rules), then you would not need to change the datasource.
E. For line item dimensionswe will always have only one char, be it cube or aggregate. See dimensions are groupings of chars. P, T and U dimensions will always be there, like Time dimension will contain the chars like 0CALMONTH, 0FISCPER...
Hope this helps...

Similar Messages

Help in modifying the Sequence Model

Hi,
I would like to do one of the following:
1) Modify the Single Pass execution entry point, so that, it will call the 'PreUUT' sequence in the same way as the 'testUUTs' entry point.
or
2) Add a new Entry point that does what is described in (1) above.
3) In addition I would like to make an entry point 'PARTIAL TEST' that will, both, call the preUUT and creats a report? (this is in case I select to test only few steps of my sequence). It is required to call the 'preUUT', to generate a report and to collect data with the database in this case.
If it makes things easier, I'm willing to modify the 'Single Pass' to 'Partial Test' instead of adding a new entry point.
Any suggestions/guidlines/examples before I begin to di
g in?
Thanks

Hi Rafi,
My first suggestion would be to make a copy of the sequential process model in your user directory. In my case: C:\Program Files\National Instruments\TestStand 3.1\Components\User\Models. So copy all files from C:\Program Files\National Instruments\TestStand 3.1\Components\NI\Models\TestStandModels to your user directory. These paths will vary depending on where you installed TestStand and which version you're using. This way you will always have the original code to refer to. By default TestStand looks in the user Components directory first to find files so it will use your modified process model. You can change this by going to Configure >> Search Directories.
I don't think it would be any easier to modify Single Pass versus creating a new entry poi
nt. When you create a new entry point you can always copy the code from the Single Pass entry point. I would make the decision based on what options you want the user to have. It may be that you always want the user to enter the serial number. In this case you never want the user to use the Single Pass entry point in its orginal form. So,I would delete this entry point or rename it to Partial Test or leave it Single Pass and modify the code. It really depends on what you want your user to see.
As you dig in if you have more questions please feel free to post back.
Regards,
Sarah Miracle
National Instruments

Discard the Modified version of the data model

Dear all,
In BI 7.0, I have created a cube and activate it. Then, we modify the cube and save . The system now have 2 version(A & M) of the cube. If I decide that I don't want the modified version of the cube, is there any way we can delete the modifed vesion and do the enhance from activate version of the cube?
Best Regards,
Jeff

In BW 3.5 there is a menu entry InfoCube -> Cancel Changes. This removes your changes and sets it back to the activated version. You should have the same entry in BI 7.0.
Best regards
Dirk
Edited by: Dirk Herzog on Mar 11, 2008 1:51 PM

Best way to modify Sequential Process Model for report generation.

I am using the Sequential Process Model in my application and the TestStand Reference Manual, (Figure A-1), clearly shows the following processing sequence:
...<part removed>
Call the Test Sequence
Display the UUT results
Generate a Report
Log Result to a Database
...<more removed>
I want to generate the report BEFORE displaying the results to the operator, or at a minimum, I want to generate the report in parallel with displaying the results to the operator. Currently, the problem I have is that when the test is done I have some automated scripts that take the data file and do some statistical processing on it, but the way the Sequential Process Model is set up, the test might finish but until the operator acknowledges the PASS/FAIL results display, the resulting file is never created. It could be overnight, over the weekend, or several days before an operator comes back and says, "Oh that last test finished, I guess I can press the OK button!", but until they do, I get no data. So I want the report generated no matter what, and right after the test finishes.
Any ideas as to how that might be best accomplished?
Thanks a billion - Ski (noob)

Ray,
Is that new in 4.2 that the engine won't call a callback with nothing in it? I just did it and it seemed to work fine. I'm using 4.1.1 though.
Ski,
Maybe there is a better solution for what you want. Are you using the SequentialModel? What version of TS do you use? Why does the report have to be written before the pass/fail banner displays? The pass/fail banner gets displayed in the PostUUT callback. Like Ray said if you just put that in your client sequence you won't see the banners. However, I'm assuming there is more to this than just that. I'm assuming you want to see the report because of your external analyzer that is gathering the statistical data. And then based on that data you want to allow the user other options. Is this correct?
If so then I would override the PostUUT callback and then use a different callback (possible the ProcessCleanup callback) to displaly the banners. You could even do this without modifying the process model (which I always try to avoid). Just override both the PostUUT and ProcessCleanup callbacks. And then put code in the ProcessCleanup to behave like you need.
Or if you want you can modify the process model and create a new callback lower in the process model. Then have that new one do the post report analysis.
Just some thoughts.
jigg
CTA, CLA
teststandhelp.com
~Will work for kudos and/or BBQ~

Batch testing using the Sequencial Model

One of our Test Engineers is uncomfortable with the Batch Model provided with TestStand 2. Is there an easy way of using the Sequential Model as a Batch Model (testing multiple UUTs). The way I understand it, there is much more flexibility and capability built into the Batch Model, but I'm having a hard time convincing him and our boss of that. It seems as though by the time you get done modifying the Sequencial Model to do Batch testing, you've basically created a custom Batch Model anyway. And if that's the case, why not use something that was already designed specifically for that purpose. Am I looking at this correctly?

jtc,
you are correct in your understanding regarding the basics of the two different models. The sequential model is designed to test a single UUT at a time. The batch model expands this to test as many UUTs in parallel as you like, but the execution is synchronized on some points for the UUTs. This is done because we asume that the UUTs are connected together somehow e.g. placed on the same carrier. So each execution first handles things for all UUTs once (like retrieving/positioning of the carrier) and then starts to test each "socket" (meaning each UUT). After each socket has finished, the carrier is removed and a new one has to be placed in the fixture.
Since i dont know TS 2.0 (i started with 3.0) i am not sure where the differences are regarding processmodels. In TS 3.x, you have a third processmodel called "ParallelModel" which does not contain logical dependencies between the sockets.
Another thing about the BatchModel is that you can insert "synchronized sections" where you can "change the way of execution in parts of your sequence". One possible setting is e.g. "One thread only" which makes sure that the sockets synchronize at the beginning of the section and then only ONE socket executes the content of the section. After this, all sockets continue testing.
Another "problem" which is encountered by many beginners is that both parallel and batch model copy the initial sequence for all sockets. So each socket does the same tests which is propably not wanted.
Nevertheless, you can change the execution by including all tests for all sockets. Then create appropriate preconditions where you check for the socket you are currently working on. The needed Lookupstring would be "RunState.TestSockets.MyIndex"....
hope this helps,
Norbert B.
CEO: What exactly is stopping us from doing this?
Expert: Geometry
Marketing Manager: Just ignore it.

How to invoke the Client Sequence Clearup process in the SequencialModel?

Hi,
I want to know how to directly call the Client Sequence Clearup in the SequencialModel?
I don't wait to modify the main squence files, because many of the clearups in sequence files are different. I want realize it by modifying SequencialModel file.
there are the problem I suffered:
I have added the some process in clearup like close port, close com, but when I terminate the program, clearup not be executed, I need to modify the SequentialMode to realize that it will jump to the Client main sequence to execute the clearup first when I terminate the process, then execute the other clear process in the sequencialmodel process.
please help,
thanks,
Ivan

Hi,
I also faced the same problem in my project. I am not sure whether there are any direct methods, but i achieved it through a round about method.
1. Create a Boolean station global
2. Set it to True when you exit the client sequence – as a Last Step in the client sequence clean up tab
3. Modify the process model to continue the clean up only after the Boolean value is True - may be this should be the first step in the process model's clean up tab
4. Don’t forget to initialize the Boolean (Set to false) when starting the test
Hope it helps
Thanks,
Sasi

End Routine - Modify a record in the cube

Hello Guys,
In the end routine I have to update a field . The transformation is the same cube to the cube. I want the record to be modified like add a value to a field in the end routine. Since it is a cube it creates a new record instead of overwriting the existing record. Is there anyway I can modify the record. I know I can make the existing record 0.00 and then create a new record with the new value.Is there any other solution.
For example:This is the existing record in the cube
sales order           Item No                      Backlog Amount                      Indicator
1000                     10                              1000.00
After applying the end routine it has 2 records ( I modify the record with the indicator value)
sales order           Item No                      Backlog Amount                      Indicator
1000                     10                              1000.00
1000                     10                              1000.00                                    REV
After applying the end routine I need the record to be like overwrite(similar to DSO)
sales order           Item No                      Backlog Amount                      Indicator
1000                     10                              1000.00                                   REV
How to achieve the above result in end routine.
Thanks
Senthil

Hi there,
Since you create new records in end routine in the InfoCube, why not delete the old ones?
You can use the
delete RESULT_PACKAGE where ...
Therefore deleting the old records after inserting the new ones.
Diogo.

Modelling the fields in the cube from different source system.......

Hi,
I need to create one query on the cube.
User want some of the fields in the query which are present in diffrent source system , rest of the fields are present in the cube.
These are master data fields ( 7 fields in total) from other source system.
Cube gets data from one ODS.
some steps I need to do are :
1) create new source system
2) Create corresponding objects in BW.
3) Model the things in ODS and cube
Can somebody let me know what is the best way/steps to model such scenario.
Thanks, Jeetu

Hi,
Replicate both the data sources to BW.
Create 2 infosources/transformations, map them to respective fields and infoobjects.
Now assign them to cube and load the data
hope this helps.
Rgs,
I.R.K

Can I modify a table model outside of the event thread if it's not seen?

I have a JTable that I use to display several table models (I'm extending DefaultTableModel); I switch between the models using JTable.setModel.
If model A is active and currently being viewed, I of course make sure any updates are done using SwingUtilities.invokeLater().
My question is: Can I modify an "inactive" table, say table B, without using invokeLater(), for example B.setValueAt() etc.

Isn't a model connected to an instance by JTable.setModel()? (Or by passing the model as a parameter in new JTable()) Yea. I meant connected to a JTable's UI. Or more specifically, listeners are installed on the model.
if data is changing and the user is not currently viewing that data, I want to change the table model in another threadI would argue that this is thread safe:
1) When a table becomes 'inactive' (not showing in any way) then you set the model to some static empty one.
table.setModel(EMPTY_MODEL);This should happen on the EventDispatchThread. Setting the model to the empty one will uninstall all the listeners on the actual model.
2) Freely modify the table's actual model in another Thread.
3) When the table is about to become 'active', then you set the model back to the original one.
table.setModel(originalModel);This should happen on the EventDispatchThread. In particular, the other thread should execute it with an invoke later. That way, proper synchronization occurs and the EDT sees all the changes made to the original model.

How do I keep from highlighting terms in the search bar when I modify a string in the search bar--other than by clicking three to five times?

I'm not talking about terms in the search results; I'm talking about terms in the search string in the search bar in the upper right corner. (Mine is set to use only Google.) When I've searched on a string and decide to modify the string for a second search (say, changing "Spanish architecture" to "Spanish architectural style"), clicking at the end of the search string first highlights the entire string and then, as I keep clicking, highlights various parts of the string. Sometimes it takes five clicks to get the highlighting to go away so I can start back-deleting normally instead of by entire word or phrase. The Google homepage doesn't do this, whereas all search engines in Firefox do it, so it's clearly a Firefox annoyance.

Left click ONCE to get it's attention. After a moment, then do your work.

I modified the first business card in a template. Does anyone know how to get the other cards on the template to look like the first one without doing each individually? please and thanks!

I modified the first business card in a template. Does anyone know how to get the other cards on the template to look like the first one without doing each individually? please and thanks!

http://www.freeforum101.com/iworktipsntrick/viewtopic.php?t=182&mforum=iworktips ntrick
Peter

How to modify the where clause in this to include other things

How do I modify the WHERE clause in this code:
$expected = array(first_name, last_name, level, language, 'city', 'state', 'zip', );
$query_search = 'SELECT * FROM french;
$where = false;
foreach ($expected as $var) {
if (isset($_GET[$var]) && !empty($_GET[$var])) {
if ($where) {
$query_search .= ' AND ';
} else {
$query_search .= ' WHERE ';
$where = true;
$query_search .= $var . ' = '. GetSQLValueString($_GET[$var], 'text');
to include things like searching multiple columns and IF NULL:
SELECT * from MyTable where (MyCol1 = %s or MyCol1 Is Null) AND (MyCol2 = %s or MyCol2 Is Null)
Where on earth does it go?
Thanks
Brian

My goal is to have a search form and results combination that:
1) allows the user to ignore any of the page’s existing search fields that
they don’t need, (to search with empty form fields);
2) to use whichever search form fields they choose to use even if the
Registrant for whom they are searching did not insert that information when
they registered, (they are filled in the search form but empty in the
database table);
3) and to permit some of the seach fields to search multiple columns, for
example the search form field ‘sonsName’ would look for that information in
three different database table colums, (sonsName1, sonsName2 etc.).
You’ve given me code that permits 1) above to work beautifully. You’ve also
sent along code that should let me accomplish the other two, and though I’ve
been able to recreate below what I want the multiple search column to do, I
can get neither the multiple column nor the IS NULL sections to work in the
code you sent most recently.
How do I take something like this:
$colname_rsSearch = "-1";
if (isset($_POST['first_name_now'])) {
$colname_rsSearch = $_POST['first_name_now'];
$colname2_rsSearch = "-1";
if (isset($_POST['last_name_now'])) {
$colname2_rsSearch = $_POST['last_name_now'];
$colname3_rsSearch = "-1";
if (isset($_POST['sonsName'])) {
$colname3_rsSearch = $_POST['sonsName'];
mysql_select_db($database_connStr_Milo_db, $connStr_Milo_db);
$query_rsSearch = sprintf("SELECT milo.first_name_now, milo.last_name_now,
milo.ws_number, milo.ws_number_3, milo.ws_number_4 FROM milo WHERE
milo.first_name_now = %s AND milo.last_name_now = %s AND (milo.sonsName1 =
%s) || (milo.sonsName2 = %s) || (milo.sonsName3 = %s)",
GetSQLValueString($colname_rsSearch,
"text"),GetSQLValueString($colname2_rsSearch,
"text"),GetSQLValueString($colname3_rsSearch,
"text"),GetSQLValueString($colname3_rsSearch,
"text"),GetSQLValueString($colname3_rsSearch, "text"));
$rsSearch = mysql_query($query_rsSearch, $connStr_Milo_db) or
die(mysql_error());
$row_rsSearch = mysql_fetch_assoc($rsSearch);
$totalRows_rsSearch = mysql_num_rows($rsSearch);
?> (I’ve excluded the IS NULL part from my example, but need it in the final
version)
…and make it work using your code below (which permits someone to search
with empty form fields, but does not yet enlighten me on what to group (with
parenthesis) to make the code execute the multiple column search and the IS
NULL parts of the search.
mysql_select_db($database_connStr_Milo_db, $connStr_Milo_db);
$expected = array('first_name_now', 'last_name_now', sonsName'
$query_search = 'SELECT milo.first_name_now, milo.last_name_now,
milo.ws_number, milo.ws_number_3, milo.ws_number_4 FROM milo';
$where = false;
foreach ($expected as $var) {
if (isset($_POST[$var]) && !empty($_POST[$var])) {
    if ($where) {
      $query_search .= ' AND ';
    } else {
      $query_search .= ' WHERE ';
      $where = true;
$query_search .= '(' . $var . ' = '. GetSQLValueString($_POST[$var],
'text') .
' OR ' . $var . ' IS NULL)';
$search = mysql_query($query_search, $connStr_Milo_db) or
die(mysql_error());
$row_search = mysql_fetch_assoc($search);
$totalRows_search = mysql_num_rows($search);
As always, thank you.

How to transport a VC model other than the simple import/export?

Hello,
What are the other options I have to transport a VC moel from one portal to another other than the export/import option that goes through my local filesystem?
If I will use the portal transport package to transfer the iView generated it won't work since the iView doesn't contain the complied application in it that needs to be deployed to the server. So what are the other option I have? And as a followup question, does VC integrated with NWDI or NWDS in some way?

Hi,
> If I will use the portal transport package to
> transfer the iView generated it won't work since the
> iView doesn't contain the complied application in it
> that needs to be deployed to the server.
Are you Compiling the VC model with "Deploy Model with PAR" checkbox in compiler tab checked? If you create a transport package and add all objects in the model folder, it will work.
What are the issues you are facing in portal transport import?
Regards,
Sooraj

Modifying the index from one filter to other

All,
We accidentely created the "BINARY_SEARCH_INDEX" index in CTXSYS user with INSO_FILTER, but is there any way that i could modify the filter from INSO_FILTER(current) to AUTO_FILTER without rebuilding the indexes.
Any help would be great.
Regards,
~ora

Im working on the same(google) but ..Any help would be great.
Regards,
ora

Write-back dimensions and how do they affect other users of the cube?

If write-back is enabled on a cell or dimension and a value is changed, will it affect other users of that data as well?
When the cube is rebuilt will those new values/updates be overwritten?

Hi Shifrbit,
Writeback is implemented as ROLAP partition. Analysis Services creates a table in relational database where it stores writeback data. User do not need to worry about the mechanism Analysis Server is using to store writeback data. You can re-process your
cube without concern of loosing data. Writeback will not disappear after process the cube. For the detail information about it, please refer to the link below to see
Affect of Cube Changes section.
http://msdn.microsoft.com/en-us/library/ms145568.aspx
Hope this helps.
Regards,
Charlie Liao
TechNet Community Support

Modifying the cube model & others doubts

Similar Messages

Maybe you are looking for