> The point that Steve was making is that your application
> SHOULD "recover"
> from this error by recognizing this record for what it is --
> a case where
> a "logical" transaction has been interrupted, so any
> "recovery" should roll-
> back to the beginning of the logical transaction, read &
> delete this error record, and continue processing.

No, it should NOT "delete this error record", because there is no "error
record"; in the absence of other problems ("bad variable block structure"
being a favorite), the record that was retrieved when FREAD returned with
CCL/151 is exactly like every other record in the message file, and should
be treated exactly the same way.

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *