Schema file

medakiran · Post by **medakiran** » Fri Sep 23, 2005 7:18 pm

Hi,

Can anyone, please tell how can i generate a Schema file?
I have a requirement where, there are like 19 sequential stages in a job, using the same meta data, so i know tht instead of specifying the column definitions for each and every stage i can use a Schema file instead, but, i just dont know how to generate that.
So, someone do kindly help me.

Thanks,
kiran

roy · Post by **roy** » Sun Sep 25, 2005 1:16 am

Hi,
You can search here and the docs to find relevant info regarding schema files.

As for your 19 stages there are several options.
Then most simple and trivial would be building your first stage and output link.
After that simply use copy & paste to get the remaining 18 and you only need to change the file name to move on.

IHTH,

trobinson · Post by **trobinson** » Mon Sep 26, 2005 7:08 am

The problem with manually copy and pasting is maintenance when a new field is added. It is MUCH easier to generate the schema file once and maintain that for all 19. I'll assume Ascential's intent was to use table definitions to generate schema files. It can be done this way but the generated schema file is not very good. It still needs to be manually tweaked in order to work correctly and once generated it doesn't update very well. It would be nice to maintain a table definition that represents the schema file and only maintain that, generating a new schema file from the table def as needed but the functionality is too squirrelly.

roy · Post by **roy** » Mon Sep 26, 2005 8:00 am

And what do you intend to do with the added column/s? auto propegate it?
What if only one or several of them are changed and allothers remain the same?
I'm not certain that is a better solution, but I guess all solutions have upps and downs pros and cons.

If the shema is the same what about the entire process of data?
If it is also the same how about making a multipe instance job? (assuming changes apply to all files might work)

Brethart · Post by **Brethart** » Mon Sep 26, 2005 1:47 pm

I remember seeing schema in the generated osh script in the job design. You can try include one seq stage with proper meta data and then view generated osh (all in DS Designer - if you dont hv such option, then it has to be enabled from DS admin), copy the schema defined and paste into a new file. I hvn't done this yet, but this should work.

roy wrote:And what do you intend to do with the added column/s? auto propegate it?
What if only one or several of them are changed and allothers remain the same?
I'm not certain that is a better solution, but I guess all solutions have upps and downs pros and cons.

If the shema is the same what about the entire process of data?
If it is also the same how about making a multipe instance job? (assuming changes apply to all files might work)

roy · Post by **roy** » Tue Sep 27, 2005 12:51 am

Or you can build one and a dummy dataset and use the dataset management tool to view the file's schema.
But this is info you can find in a simple search here

ray.wurlod · Post by **ray.wurlod** » Tue Sep 27, 2005 1:09 am

Open the table definition in Manager.
Select the Layout tab.
Choose the Parallel option.
Copy the schema, paste into any text editor.
Save schema file.

medakiran · Post by **medakiran** » Tue Sep 27, 2005 9:48 am

ray.wurlod wrote:Open the table definition in Manager.
Select the Layout tab.
Choose the Parallel option.
Copy the schema, paste into any text editor.
Save schema file.

Thanks a lot ray, i got that.
kiran