I’m assuming PERS_NID is too big to comfortably load into an array, so we’ll do it with individual selects.

1) Create a temp table mirroring PERS_NID. Run through the file once looking at PERS_NID for exact matches. For each exact match, knock that row out of the temp table. This reduces the chance of a bad guess in the subsequent steps.

2) Create a local procedure – let’s call it Test-All – which takes two input parms: the SSN being tested and the suspect character position. This procedure tries the temp table for each digit 0 through 9 in that position until it finds a match (presumably by calling another procedure, not by using 10 distinct queries).

3) If the SSN being tested is short one character, do a loop to insert ‘x’ in each position where a character could have been missed and call Test-All specifying that position. (e.g. if you have 12345678, call it for x12345678 with suspect position as 1, 1×2345678 with suspect position as 2, and so on through 12345678x.)

4) Parse the SSN being tested for specials or alphas. If one is found, that’s your suspect character position for calling the Test-All procedure.

5) If no specials or alphas were found in step 2, run Test-All for each position 1 through 9 in sequence, looking for a single wrong digit.

6) If still no match, do a loop to parse the SSN under test and swap pairs of adjacent digits. For instance, if the test is 123456789, then the first loop iteration should be 213456789, the second should be 132456789, and so on. Test each against the temp table.

7) If still no match, do a similar loop to swap pairs of nonadjacent digits.

If still no match, the SSN has more than one problem. Print it on a report and send an email to the idiots who sent you the bad import file.

Move ‘Y’ To $RowInTbl

FROM ps_fciap_rnctl param
WHERE param.OPRID = $prcs_oprid
AND param.RUN_CNTL_ID = $prcs_run_cntl_id
End-Select
!
If $RowInTbl = ‘N’
Show ‘This program must be run from the Process Monitor.’
Stop Quiet
End-If
(Any other validations go here)
(Continue your job)
Enjoy!

If it’s really a problem that deserves 100 hours of your time to fix, perhaps you should reconsider your algorithm. It sounds like you’re comparing the data dictionaries of two databases to keep one in sync with the adds and deletes of the other. Are you performing complete outer joins? That’s very thorough and a lot of extra work if you’re expecting a 99% match. Another approach would be for each database instance to run an SQR to sort the database objects and write them to disk files. Then another SQR program could read both files into arrays and compare the arrays row by row.

This way we can test a program under development using SQRW and then develop the run control tables and page after the SQR has been unit tested.

Came upon your website and perhaps you can offer some assitance. When we upgraded a while back from PeopleSoft v8.8 to v9, we also upgraded SQL Server to 2008. Part of this was to update selected SQR programs that utilize outer joins. Most of them I could deal with however it seems that SQR programs fail when they have bind variable(s) in the outer join portion. The identical statement runs just fine in our Query Analyzer tool when I replace the bind variables:

begin-SELECT
CT.WRKDAY_ID
CT.DUR
P.TRC
P.TL_QUANTITY
FROM PS_SCH_CLND_VW CT LEFT OUTER JOIN PS_TL_RPTD_TIME AS P ON P.DUR = CT.DUR AND P.EMPLID = &_J.EMPLID AND P.EMPL_RCD = &_J.EMPL_RCD AND P.TRC = $Sck_Cd
WHERE CT.SCHEDULE_ID = &WT.SCHEDULE_ID
AND CT.DUR BETWEEN $Q_Begin AND $Q_End
ORDER BY CT.DUR DESC
end-SELECT

This is what I get in the log file:
(SQR 1303) Error in SQL (perhaps missing &name after expression):
SELECT CT.WRKDAY_ID, CT.DUR, P.TRC, P.TL_QUANTITY FROM PS_SCH_CLND_VW CT LEFT OUTER JOIN PS_TL_RPTD_TIME AS P ON P.DUR = CT.DUR AND P.EMPLID = ? AND P.EMPL_RCD = ? AND P.TRC = ? WHERE CT.SCHEDULE_ID = ? AND CT.DUR BETWEEN ? AND ? ORDER BY CT.DUR DESC
(SQR 1304) Check SELECT columns, expressions and ‘where’ clause for syntax.

SQR for PeopleSoft: Program Aborting.

This is executed within a local procedure. That said, the old way of doing the outer join used the variables and *= . This worked fine.

In Query Analyzer with the bind variables replaced:
SELECT
CT.WRKDAY_ID,
CT.DUR,
P.TRC,
P.TL_QUANTITY
FROM PS_SCH_CLND_VW CT LEFT OUTER JOIN PS_TL_RPTD_TIME AS P ON P.DUR = CT.DUR AND P.EMPLID = ‘0717′ AND P.EMPL_RCD = 0 AND P.TRC = ‘PS1′
WHERE CT.SCHEDULE_ID = ‘STNDFIREB’
AND CT.DUR BETWEEN ‘2012-02-06′ AND ‘2012-02-20′
ORDER BY CT.DUR DESC

Yields:

OFF 2012-02-20 00:00:00.000 NULL NULL
OFF 2012-02-19 00:00:00.000 NULL NULL
OFF 2012-02-18 00:00:00.000 NULL NULL
OFF 2012-02-17 00:00:00.000 NULL NULL
WRKDAY14HR 2012-02-16 00:00:00.000 PS1 14.000000
WRKDAY14HR 2012-02-15 00:00:00.000 PS1 14.000000
WRKDAY10HR 2012-02-14 00:00:00.000 PS1 10.000000
WRKDAY10HR 2012-02-13 00:00:00.000 PS1 10.000000
OFF 2012-02-12 00:00:00.000 NULL NULL
OFF 2012-02-11 00:00:00.000 NULL NULL
OFF 2012-02-10 00:00:00.000 NULL NULL
OFF 2012-02-09 00:00:00.000 NULL NULL
WRKDAY14HR 2012-02-08 00:00:00.000 PS1 14.000000
WRKDAY14HR 2012-02-07 00:00:00.000 PS1 14.000000
WRKDAY10HR 2012-02-06 00:00:00.000 NULL NULL

Thx,
Geoffrey

let $x = datenow()

You should make sure that $x is a date variable, not a string variable. You do that in the setup block:

begin-setup
declare variable
date $x
end-declare
end-setup

Alternately, you could write:

let $x = datetostr(datenow(), ‘mm/dd/yyyy hh:mi:ss’)

Your note indicates that this is working properly. I assume you insert a row with the $x variable placed in a date, datetime, or string field. The insert should work with all three field types, but you would handle the fields differently when reading them back into an SQR program.

What type of field is &AUDIT_STAMP? Since you used to_char successfully, I assume it is a date or datetime field (which is a distinction in PeopleTools, not in the Oracle database). In that case, you can use to_char, edit, or datetostr to convert it to a string. You would not use strtodate() – that is a function that starts with a string, and you are starting with a date.

If &AUDIT_STAMP is a string, and you are storing it as a date, you may need datetostr() to format a string to output that date.