Managing wwv_flow_files table

Question

Lets say I have a file upload feature on one of my pages.
When the page is submitted, the file is uploaded to the wwv_flow_files table.
I have a after submit process to parse the file, load it into a collection and process the data. When the process ends successfully, it does a DELETE from wwv_flow_files.
I put a validation to upload only certain types of files based on the extension. If the validation fails, the file stays around in the wwv_flow_files table.
If I have a after submit parse routine to parse/validate the file and that fails, the file stays around in the wwv_flow_files table.
If I have any othe error in my after submit process that causes it to exit via my raise_application_error(), the file stays around in the wwv_flow_files table.
In this situation, what is the recommeneded way to properly purge the wwv_flow_files table?
I havent yet put COMMITs anywhere in any of my HTML DB apps because the engine implicitly does a commit at the end of every page view, but this file upload thing seems to be its own little "autonomous transaction".
Thanks

Answer 1

Vikas - If you can do a delete after your process
runs successfully, why can't you do a delete when you
encounter one of the other situations (before raising
an unhandled exception)? Let me explain all my "use cases".
1. On Submit validation that checks the "extension" of the file. This is a simple PL/SQL expression that checks lower(:P1_FILENAME) like '%.csv'.
If this validation fails, the file hangs around in the wwv_flow_file_objects$ table, how can I prevent this?
2. If the validation succeds, I proceed to my one and only after submit process that does something like
parse_file(:P1_FILENAME);
if something then
raise_application_error(-20000,'something failed');
end if;
if otherthing then
raise_application_error(-20000,'otherthing failed');
end if;The parse_file procedure parses the file and stores it in a collection and if everything succeeds it does a delete from wwv_flow_files where filename=:P1_FILENAME;
How can I make sure that the file doesnt stick around in wwv_flow_files no matter what?
For a catch-all technique that recognizes fringe cases, consider a dbms job
that runs from your workspace.What would this job look like?
Even a rollback in your own page process would be too
late to prevent it. A rollback in your process would
prevent the file from appearing in the wwv_flow_files
view but would not prevent it from remaining in the
underlying wwv_flow_file_objects$ table.Not sure I understand this. The wwv_flow_files is a simple wrapper view    FROM wwv_flow_file_objects$
   WHERE security_group_id = wwv_flow.get_sgid;So why would something be in the table but not in the view?