How do I get different record type into 1 dataset??

[mstaton]

I am creating multiple record types in my transformations. Each record is 256 bytes in size with only the 1st 2 fields are the same on each record. How do I go about getting them into 1 sequential file???

[anbu]

Can you provide sample input and expected output?

[mstaton]

Can you provide sample input and expected output?

Header Record - HDR000000112012009xxxxxxbbbbbb
Detail Record - DTL0000002aaaaa123456zzzzzzzzzz
Trailer Record - TRL0000003x1x1x1x1x1x1x1x1x1x1

Expected result on final file

HDR000000112012009xxxxxxbbbbbb
DTL0000002aaaaa123456zzzzzzzzzz
TRL0000003x1x1x1x1x1x1x1x1x1x1

[anbu]

Your input and output looks identical. So what are you doing with transformations?

Are the different records comes in different input files?

[ray.wurlod]

A Data Set may have only one record schema.

However, you can use subrecords to store variant formats in a Data Set. The penalty is that you have to create and later parse the subrecords.

[kwwilliams]

Your topic says that you want multiple layouts in a dataset, yet your post says that you want multiple layouts in a sequential file - that makes this thread a little confusing.

I'm assuming sequential file, from your transformer you should have three links coming out (if you need them in a particular order in the file you should create an extra column with a number associated with the oder for instance dummy column with value of 1 for header, 2 for detail and 3 for trailer). Use a column export to collapse all of the columns except for dummy into a single column. Then funnel (use a sorted funnel with dummy as your key) and write to a data set.

[ray.wurlod]

In either a sequential file or a data set you could use a record schema that contains three fields, field1, field2 and "the rest" (as a VarChar), at the expense of some parsing.