Re: 5250 and un-do

I use a field NOT NULL GENERATED ALWAYS FOR EACH ROW   ON UPDATE AS ROW
CHANGE TIMESTAMP,    

By comparing that field I can be sure the record has not changed.  It
also is very useful to follow a journal...


On 12/19/2017 03:39 PM, Nathan Andelin wrote:

Just to follow up on some of the comments so far. Good points all around
regarding database normalization.

I've never found any use for locking a record while it is displayed, given
a multi-user environment. I shake my head in awe.

I like Henrik's idea of including an update counter in a record, and
checking to see if it has changed since the current user last read the
record. That's more efficient and easier to implement than comparing
multiple record buffers in memory, each containing a copy of a record in a
different state.

For applications that are only required to update say one column in a row,
you end up needing a record buffer on the update operation that is
distinguished from the record buffer used in the read-from-screen
operation. Use an assignment statement to copy a column value from one
buffer to the other. An SQL Update statement does that, under the covers.

Though it would be possible, I wouldn't expect a super-basic utility like
DFU to have an undo key. Use cases for undo fall in the realm of
application development.

-- Este e-mail fue enviado desde el Mail Server del diario ABC Color --
-- Verificado por Anti-Virus Corporativo Symantec --