Re: File locking / field locking

["Charles Wilt" <charles.wilt@xxxxxxxxx>]

On Fri, Oct 17, 2008 at 9:56 AM, Doug Palme <DPalme@xxxxxxxxxxx> wrote:
>   In a subroutine the program sets a HIVAL on the logical and then does a
>   readp in order to get the highest shipping number.  At this point I
>   believe this field is locked and no other process can access read that
>   record.

The record is locked, but that doesn't mean no other process could read it.

Usually anything can simply read a locked record, what can't be done
is reading the record for update (thus the other process is trying to
lock the record itself.)

If you need to prevent other processes from even just reading the
record.  Then you'd want to use commitment control with commitment
lock level of *ALL (?)

>  it then increrments the count and writes it to a variable.  The
>   other variables are setup and then the program finally writes a new record
>   to the Physical file, which at that point the lock is released correct?

Maybe.  It depends on if you are using commitment control or not and
if so, what lock level you are using.

With *CHG or *CS, the yes the lock is released.  With *ALL, the lock
is not released until you commit.

Note: SQL has a couple of extra lock levels (aka isolation levels).
Also, the interaction between two processes is dependent on the
isolation levels of both processes.

HTH,
Charles