dates in subfiles... Webfacing

I have a date field in a subfile that the user will be updating. The subfile is
initialized rather than cleared (too much logic to let users add records to the
subfile if I clear it). The subfile is loaded on program entry from a disk file,
and the disk file is updated at program close. If there are no records in the
file, the program blows up with bad date when the program is exited; if there
are records in the file, the program blows up with bad date when the program is
entered. My guess is that the date field in the initialized subfile does not get
*loval initialization, but that's only a guess. A dump at time of blowup shows
all fields in question initialized correctly, but that would probably be an
internal field rather than the actual field in the subfile. I've tried both
dates *USA, both dates *ISO, and one of each, to no avail. I'm ready to give up
and define the field as character and do my own date editing, but if there is a
fix, that would save me alot of time.

klokner.amy@partnersmutual.com wrote:
> I have a date field in a subfile that the user will be updating. The
subfile is initialized rather than cleared (too much logic to let users add
records to the subfile if I clear it). The subfile is loaded on program entry
from a disk file, and the disk file is updated at program close. If there are no
records in the file, the program blows up with bad date when the program is
exited; if there are records in the file, the program blows up with bad date
when the program is entered. My guess is that the date field in the initialized
subfile does not get *loval initialization, but that's only a guess. A dump at
time of blowup shows all fields in question initialized correctly, but that
would probably be an internal field rather than the actual field in the subfile.
I've tried both dates *USA, both dates *ISO, and one of each, to no avail. I'm
ready to give up and define the field as character and do my own date editing,
but if there is a fix, that would save me alot of time.
> 
> Thank you.

Does it work ok in green-screen?  If so, what does "blows up" mean? 
What operation was being performed when the failure occurred. Do you 
have a stack trace or something ?

I sent this to you as a Reply, and am repeating it here in case it doesn't work
that way:

I realized after I answered your message on Friday that I hadn't actually
answered your question!  Sorry.

As I enter the program for a policy, I fill the sub-file with any records that
exist. When records do not exist, the program opens fine. I update all files as
the program is exited, and whether or not records exist, the program gets a date
halt. I have to read one record beyond what have been entered into the sub-file
to determine that I am at end, and the program halts when an empty record is
read (whether this be the first or the one after existing data). If records
exist when the program opens, it halts with a date error on the first record
read, when the empty sub-file record is read to be set up for the existing data.
I have dumped the program when it blows up at the beginning or upon exit, and I
can tell from index settings what it is doing. The statement referenced is
always the same, the input line for the date field in the sub-file. 

Green screen has no problems, and when the sub-file is empty, it is pre-filled
with *loval dates. On the web, if the sub-file is empty, the date fields are
blank. 

If there is a fix, I would be grateful. This is apparently the first sub-file
I've had in our web programs that contains a date; rather, this is the first one
with a date that I have initialized rather than cleared. 

Thank you.

Amy Klokner
Partners Mutual Insurance Co

klokner.amy@partnersmutual.com wrote:
> I sent this to you as a Reply, and am repeating it here in case it doesn't
work that way:
> 
> I realized after I answered your message on Friday that I hadn't actually
answered your question!  Sorry.
> 
> As I enter the program for a policy, I fill the sub-file with any records
that exist. When records do not exist, the program opens fine. I update all
files as the program is exited, and whether or not records exist, the program
gets a date halt. I have to read one record beyond what have been entered into
the sub-file to determine that I am at end, and the program halts when an empty
record is read (whether this be the first or the one after existing data). If
records exist when the program opens, it halts with a date error on the first
record read, when the empty sub-file record is read to be set up for the
existing data. I have dumped the program when it blows up at the beginning or
upon exit, and I can tell from index settings what it is doing. The statement
referenced is always the same, the input line for the date field in the
sub-file. 
> 
> Green screen has no problems, and when the sub-file is empty, it is
pre-filled with *loval dates. On the web, if the sub-file is empty, the date
fields are blank. 
> 
> If there is a fix, I would be grateful. This is apparently the first
sub-file I've had in our web programs that contains a date; rather, this is the
first one with a date that I have initialized rather than cleared. 
> 
> Thank you.
> 
> Amy Klokner
> Partners Mutual Insurance Co
> 

Hi Amy,

I guess I don't 100% how a subfile record can be empty but be 
initialized somehow?  Do you mean that you have created the subfile 
record but it does not yet contain any data and is thus considered empty?

In any case I had a chat with the i5/OS developer that works on the 
Workstation component (the part that handles the 5250 and WF data in the 
i5) to find out more.  Apparently even the i5 does not init date fields 
with other than blanks so I must presume that something else is doing it 
- could this be being done by the 5250 display itself?

Probably the best thing to do would be to either init all the records 
that are being written to the display or else handle the exception that 
you receive and consider that as an indication of an empty record?

If you would like us to look into this further you would need to open a 
PMR so that we could get some docs to reproduce and debug the problem in 
the lab.