Re: Results of SQR Upgrade Testing

[Benjamin Le <ble@PCC.EDU>]

Denise,
Thanks for posting this valuable information about SQR3 vs SQR4. We too are
in process of upgrading from V3 to V4.3 on our Unix/Oracle box. I would
like to add few founds:

1. Default installation directories (Oracle):
        V3: /opt/sqr/ora/workbench/
        V4: /opt/sqr/ora/ (workbench is gone)
If you have scripts hard coded with old directories, you need to change to
the new directories for V4 to work.

2. SQL handling:
WORK ON V3 BUT NOT V4:
"MONTHS_BETWEEN
   (to_date(&remaining_pays_begin_date,'DD-MON-YYYY'),
    to_date(&end_date,'DD-MON-YYYY')) &date_diff"
WORK ON V4: (arguments need to follow the command, not next line)
"MONTHS_BETWEEN(to_date(&remaining_pays_begin_date,'DD-MON-YYYY'),
    to_date(&end_date,'DD-MON-YYYY')) &date_diff"

3. If using edit mask to print date with 2-digit year from database, SQR V4
will produce an error log as a warning for Y2K.

At 03:29 PM 8/2/99 -0400, you wrote:
>Thanks to everyone who responded to my post last month about upward
>compatibility between versions 3 and 4!  It was very helpful.  We have now
>completed our testing of in-house-written SQR programs, and I thought I
>would share with you what I found.  There were some minor printing
>differences between the two versions, but nothing serious.
>
>1)  I had never noticed it, but version 3 was inconsistent in its use of
>edit masks, depending on the type of field being printed.  If I had an edit
>mask of 99.99, and the value was .5, a &col (either printed with implicit
>PRINT from the select, or using the &col name later with explicit PRINT)
>would print as .50, while a #var would print as 0.50.  An exception was a
>value of 0, which printed as 0.00 no matter what.  Version 4 is consistent,
>and prints 0.50 no matter which type of field is being printed.
>
>2)  I was printing an age (&spouse_age retrieved in another select) using a
>mask of 99.  In version 3, it printed as blank if there was no spouse and a
>null value was returned.  In version 4, it printed as 0.  Easily fixed by
>changing the mask to b9.
>
>3)  I had a program that printed several reports in succession to the same
>.lis file.  Several of the reports had no output.  In version 3, I did not
>get any pages at all for these reports.  In version 4, I got pages with
>headers but no detail.
>
>4)  I think most of you are probably aware that if you print multiple .lis
>files concurrently, version 3 will name the second report rptname.l01, while
>version 4 will name it rptname_01.lis.
>
>5)  This one bugs me, and if anyone has an answer, I would appreciate it!  I
>have a report that uses the wrap parameter on a long character field.  There
>are two lines in my header, and I use BEGIN-HEADING 2.  In version 3, if a
>page break occurs in the middle of printing the wrapped long character
>field, the continuation starts on line 3 of the next page, under the header.
>In version 4, it continues on the 2nd line of the heading.  Because of my
>placement of the field, it is not overlapping anything on my header, but it
>does not look good.  I have tried various things, like making the header
>definition 3 lines, but whenever I change anything like that, the shifting
>of the lines results in my not having any wrapped fields over a page break,
>so I can't tell if they work.
>
>6)  I had asked about "alter session set nls_date_format = 'DD-MON-YYYY';"
>not working any more.  The answer turned out to be using the -zif parameter
>to point to an .ini file that has '"SQR_DB_DATE_FORMAT=DD-MON-YYYY" in it.
>
>7)  Sorting seems to be working a little bit differently.  When sorting
>multiple records with the same sort keys, they came out in different orders
>in the two versions.  This did not affect the outcome of the processing, it
>just made it more difficult to compare the results.
>
>I don't know if anyone will find any of this interesting or helpful, but I
>wanted to give something back to the group.  Sometimes I do know the answer
>to someone's question, but because I read them in digest form, someone else
>always has given the answer already by the time I read it!
>
>Denise White
>Sr IT Application Developer
>Textron Systems


Benjamin Le
System Development
Portland Community College
Sylvania Campus CC 224
12000 SW 49th Ave. Portland, Oregon 97219
Voice: (503)-977-4970   Fax: (503)-977-4987
Mailto:ble@pcc.edu      http://www.pcc.edu