Re: user record updates, locking, and logical vs physical record access

On 05-Aug-2016 14:58 -0500, Stone, Joel wrote:

<<SNIP>> what if the file being updated is a logical view with only
10% of the total fields in the record. In that case, one would have
to work with the PF just in case one of the non-visible fields was
changed by another user? <<SNIP>>

The application can only effect change to the fields visible through that LF record format. Thus by comparison of the previous record-image to the current record-image, necessarily inclusive only of the fields included in the format, the application knows whether the data that is capable of being changed has not been changed, since, by someone else, so the update is safe; i.e. same as for the physical record, but the posited logical record has less data available both to compare and as able to be changed.

That said, I can imagine that a triggered _before event_ logic for a secondary changer, could potentially introduce a conflict for the physical data of a change made by another updater to that same row, but the business rules implemented by a trigger probably would not generally be something for which such conflicts should arise; i.e. probably such a problematic trigger had implemented something that would not properly qualify as a business-rule, for which the operation must always occur irrespective the means to effect that I\O request, including that of logical vs physical.