Check for blank dates

[Adein]

I am using

A            S2CMDT          L  B 21 23DSPATR(CS)  
A                                DATFMT(*MDY)

in my spec for dates. I call them true dates because we commonly use decimals for dates as well.

I created a program to maintain a file, which recently I went back and added some more date fields, again using 'true' dates.

The problem is that when I open a pre-existing item from the table, using a chain, I get errors becase the year portion on the date isn't in the correct year.

 The year portion of a Date or Timestamp value is not in the correct range

So, I tried using MAPVAL, with 01/01/40, 01/01/1940, with a combo of *BLANKS and *ZEROS.

I then tried to check to see if the date was blank or zeros, or

 if mydate = d'01/01/40'

and I can't get the program to compile.

My question is, how do I not read a blank date from the file if it exists? How can I either check for it during the chain, or override the blank with something in the DDS.

Thanks!

Adein

[jamief]

Re: Check for blank dates

try if date = *LOVAL

[Denny]

Re: Check for blank dates

When you compile the program, the RPG compiler create a hidden routine to convert the date data type S2CMDT field in you program to the S2CMDT *MDY field on the screen. If the program field contains a year less than 1940 then you are going to crash.

So, assuming *ISO dates internally, in your program you need to convert the '0001-01-01' in S2CMDT to '1940-01-01'. In the DDS MAPVAL(('1940-01-01' *BLANK)) DATFMT(*MDY) or I prefer DATFMT(*JOB).

Since your date is also an input field, after the EXFMT you will have to test for S2CMDT = d '1940-01-01' and convet it back to '0001-01-01'.

[DeadManWalks]

Re: Check for blank dates

What is the value in the date field?

Also what are your H specs?

[Adein]

Re: Check for blank dates

The thing that is getting me right now, is that no matter what order I put the MAPVAL(('1940-01-01' *BLANK)) DATFMT(*MDY) in (speaking of the date) I can't get the screen file to compile.

I changed the *Blank to *Blanks just in case that was the issue, but no.

It tells me.

 * CPD5117      30        1      Message . . . . :   Syntax of date, time or timestamp value is not valid.
 * CPD7494      20        1      Message . . . . :   Keyword value not allowed.
 * CPD7520      20        1      Message . . . . :   Keyword ignored because of error in list of values. 

With the if date = *loval I get right and left operands are not equal error.


Any more thoughts?

Thanks again!

-Adein

[GLS400]

Re: Check for blank dates

Hi Adein:

I got this:

00030A            DATFLD1         L         DATFMT(*MDY) DATSEP('/')
00040A                                      MAPVAL(('01/01/40' *BLANK))

Here:

http://publib.boulder.ibm.com/infocenter/iseries/v5r3/index.jsp?topic=/rzakc/rzakcmstdfmapval.htm

If the explicit value is a date (L) value, you must use the format that the DATFMT keyword specifies. If the DATFMT keyword is not specified or if DATFMT specifies *JOB, then you must use the *ISO format. Also, you must use the separator that the DATSEP keyword specifies. If the DATSEP keyword is not specified for the separator specified on DATSEP is *JOB, a slash (/) must be used for the separator.

Best of Luck
GLS

[Adein]

Re: Check for blank dates

What is the value in the date field?

Also what are your H specs?

The date is blank. I added a date field to the program after someone started entering in information. Now, I could fix all this by writing a quick script to fill the dates, but I am trying to get the real answer.

H DATFMT(*MDY)

[GLS400]

Re: Check for blank dates

Hi Adein:

 MAPVAL(('1940-01-01' *BLANK)) DATFMT(*MDY)

If the DATSEP keyword is not specified for the separator specified on DATSEP is *JOB, a slash (/) must be used for the separator.

1. You can't use - without using datsep('-')
2. 1940-01-01 is NOT *MDY format
see my post above for IBM reference material

Best of Luck
GLS

[Adein]

Re: Check for blank dates

Hi Adein:

I got this:

00030A            DATFLD1         L         DATFMT(*MDY) DATSEP('/')
00040A                                      MAPVAL(('01/01/40' *BLANK))

Here:

http://publib.boulder.ibm.com/infocenter/iseries/v5r3/index.jsp?topic=/rzakc/rzakcmstdfmapval.htm

Best of Luck
GLS

That seemed to work. That is a great reference!

[Adein]

Re: Check for blank dates

I am still getting a crash when loading the program.

Right after the *INZR I do the following for each of the dates.

IF ISTARDT = d'01/01/40';
   S2ASGDT = d'01/01/40';
ELSE;                    
   S2ASGDT = ISASGDT;    
ENDIF;

 A            S2ASGDT         L  B 19 23DATFMT(*MDY)               
 A                                      MAPVAL(('01/01/40' *BLANK))
 A                                      DSPATR(CS)

With S2 being the screen date and IS being the physical file variable name from a chain. The program compiles, but when you open a record that already existed before I added the dates, it fails.

I really appreciate the help everyone!

-Adein

[tomholden]

Re: Check for blank dates

what is the value of ISASGDT when it blows up? you should verify that that value is a valid date prior to assigning the value to the S2 field.

[Adein]

Re: Check for blank dates

what is the value of ISASGDT when it blows up? you should verify that that value is a valid date prior to assigning the value to the S2 field.

Ok, now I am all kinds of backwards. I don't know my dates from the next one...I guess. Ok. If I wrkqry the file I get 01/01/0001 for the dates. I don't understand how that is possible....

If I run a debug statement, I get to the chain, it fails on the read.

[jamief]

Re: Check for blank dates

01/01/0001 = *USA format

[jamief]

Re: Check for blank dates

Im lost as to what you are doing.

On my system this

 S2CMDT          L  B 21 23DSPATR(CS)  
A                                DATFMT(*MDY) 

would be this - putting format of *MDY would only confuse it! cause its a length of 10

S2CMDT D(10*ISO-)

[JonBoy]

Re: Check for blank dates

What is the format of the date on the database?

If it is the default of *ISO then a "blank" date field will probably contain 0001-01-01 since that is not acceptable for a two digit year date it can never work.

Maybe you have an *MDY on the H-spec? so all dates being read are forced to *MDY and away she blows. That will happen within the CHAIN/READ itself.

Only cure is to ensure that the date field in question is read into the program with a four digit year. Do that by explicitly defining the date field in the program. Then you can test it for *LOVAL and adjust to the 2 digit year *lowval (1949-01-01) if needed.