Re: (E) Extender and %Error on WRITE

Tony Carolla wrote:
> 
> When a WRITE(E) operation results in the %ERROR BIF being on, can I be
> certain that the record _wasn't_ written to disk?

If you have a *AFTER trigger for *INSERT, and the trigger signals an
error, you'd get %ERROR turned on, but the record would still have been
written.  That would get status 1024, though.  There may be other
scenarios that would get an error after the record was written, but I
can't think of any.

> ...
> I have a module that attempts to create a record in a uniquely-keyed file,
> and it uses a DOU %ERROR loop (it has been discussed here before). It uses
> SETGT, READPE to detect an existing record, and if it finds one, it
> increments the key component by one, and WRITE(E)'s. For some reason, it
> went haywire and created five digits of records in the file. I am wondering
> if, at the end of the loop, %ERROR was on, yet a record was written anyways?

Could you explain more (some pseudo code?) about how your loop is
supposed to work, and what exactly happened when it went haywire?