Download Abinitio Interview Ques PDF

TitleAbinitio Interview Ques
File Size488.1 KB
Total Pages30
Document Text Contents
Page 1


Page 2


Page 15


Page 29

Ans: The SQL parser for dependency analysis currently has some limitations. In
particular, it is unable to handle database–specific extensions to the ANSI SQL
standard, nested SELECTs, and unions. Improvements to the SQL parser are
planned for future releases of the product.

Details: To work around these limitations, use a graph/sandbox parameter for the
problematic SQL stmts; & add a _REPO parameter to replace the runtime value with
a value that has the same dependencies and that the SQL parser can parse. The
_REPO mechanism enables you to remap a standard parameter to an EME
datastore–specific value.

Why am I getting SQL analysis errors when I try to check my project into the EME?

Short answer

There might be differences between the table format and the DML on some component. Try
regenerating the DML before checking the project into the EME.


There are many kinds of SQL analysis errors; some common ones are that a field cannot be found
in a given record format or table, or that there is no record format for the table. The EME relies on
the GDE to provide information about the structure of the database tables referenced in the graph.
For the EME to create a dataset object for an Input Table component and use it in dependency
analysis, you must associate the component in the GDE with the correct table and schema, and
have the GDE generate the appropriate record format.

When you have a database component automatically generate its record format, the GDE queries
the database for the structure of all tables referenced. This information is not used by the graph in
execution but is stored in the .mp file for dependency analysis. The EME is not only concerned
with the record format of the port (for the data coming out of the Input Table component), but also
with the record formats of all relevant tables (for the data conceptually coming into the
component). This is necessary to do dependency analysis through the database, as an earlier graph
might write these tables.

If you change which table or schema the component references (for example, if you use a
configuration variable for the table or schema and you change its value), you must regenerate the
record format before checking your project into the EME datastore. Otherwise, the graph's
information about the table structure will still correspond to the original table — even if the record
format might have been changed to reflect the current table. For this reason it is important not to
edit the generated record format by adding or removing fields.

If these errors occur at checkin, that means the errors are not a direct result of the checkin but
rather of dependency analysis, which is often performed along with checkin. You can check
projects into the EME without doing dependency analysis as part of the checkin process (that is,
using translation only), which will improve your check-in performance.

NOTE: In Version 2.14, dependency analysis is not part of the checkin process.

Similer Documents