ODI File Datastore

Question

Hi All,
I have a confusion regarding physical and logical length in ODI file datastore.
I have a fixed width file where a col c2 had datatype as string(30).
I defined that column in datastore as string>Physical length 30 >Logical length 30
My interface failed with error as"
ORA-12899: value too large for column "S0_IDM"."C$_0S0_EAGLE_DWHCDC_CHRG_MST"."C2_CHARGE_DESC" (actual: 31, maximum: 30)"
When I increased the logical length to 255,the interface worked fine.
Physical length still being the same 30.
How different is this?
Any help on this will be appreciated.
Thanks and Regards
Reshma

Answer 1

This is not from any official documentation, but here is my take after a few moments thought
Everything you do in the ODI designer is based on the logical architecture. Only at runtime is this manifested into a physical implementation i.e. connection strings are materialized etc. When you perform an integration ODI generally creates a few temporary tables prefixed with C$, I$ etc to be able to perform the necessary data movement and transformations required to, for example, populate a target datastore (table). In your example, your flat file will be materialized into such a temporary table before its contents are manipulated (or not) and loaded to the target datastore. When ODI generates this code it is using the logical length issued in the DDL that generates the temporary table column lengths, the physical column is ignored.
Now in your scenario this is not a problem as constraints such as these do not matter to the physical version of the file i.e. if you were to write back to the file it would not matter if you wrote back 255 characters or 31. This could be a problem if you were using database tables and varying the logical vs. physical lengths but usually you reverse engineer database tables using ODI rather than doing it manually so this mitigates that.
Anyway, in short, I think the logical lengths should be taken as representing what will be manifested in the materialization of the temporary objects used to manged / transform data from the source models (C$tables) and target models (I$tables)  whereas the physical lengths indicate what the underlying physical representation of those models actually are.
EDIT: After reading a bit of documentation logical actually represents the length whereas physical is related to the number of bytes required to store the data. Therefore you could have a situation with multi-byte characters where the physical length could be greater than the logical length but not really the other way around.