Why?

[George Simon]

Now back to our scheduled program...

I've created 2 new tables.  "bugauto" and "bugchoose" and one new
processing_only format with no processing called "bugform"
I've put this into a new menu line reading:

6 -----------------------------------------------------------------------
  E? Bug Test                        ?
  \fp\rreport gfsales -f bugform -v bugchoose -a -y bugauto

The "bugauto" table is identical to the automatic table except that I'm not
executing any of the lookups to other files.  I did not remark out the lines
but because all of them were being done from subroutines, I simply remarked
out the subroutine lies.

The "bugchoose" table simply reads:

     LABEL             D E F I N E   P R O C E S S I N G
5.0.09RN
----------------------------------------------------------------------------
---
  1  -------   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -
       ? If: @rn="24"
       Then: select
  2  -------   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -
       ? If:
       Then: end

The "bugform" processing-only format jas no processing at all.

When I run the menu option, I still get the error.

I also added one line to the top of the automatic table that reads:

     LABEL             D E F I N E   P R O C E S S I N G
5.0.09RN
----------------------------------------------------------------------------
---
  1  -------   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -
       ? If: @fn="" and 135 co "using"
       Then: call "getlong"; end
  2  -------   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -   -

When I run that using \fp\rreport gfsales -f compare -v choosepe -a -u

I still get the error.

Now what?

George Simon (IT Department)
American River Logistics, LTD
614 Progress St.
Elizabeth, NJ  07205
Phone:(908)354-7746      Fax:(908)354-7491
mailto:george at worldest.com
http://www.americanriverintl.com/



-----Original Message-----
From: Kenneth Brody [mailto:kenbrody at bestweb.net]
Sent: Thursday, July 01, 2004 6:16 PM
To: George Simon
Cc: filePro mailing list
Subject: Re: Why?

(Lack of proper quoting corrected.)

George Simon wrote:
[...]
> There are several records that would make it crash.  I altered the
selection
> table to read:
> If @rn="24"     'one of the records that would make it crash
> Then: select
> If:
> Then: end
> It selects the one record and crashes.

Good.  That means that the sort/select processing can be eliminated,
and replaced by your 2-line version above.

> > You say that it crashes when called from automatic, so can you eliminate
> > the output processing, which will not yet have been executed for this
> > record, and have it still crash?
> Do you mean like create a processing-only format that does nothing and has
> not code?
> I'll try it.
> Yes, it also crashes.  I created a processing-only format called
"donothing"
> and I replaced "compare" with "donothing" on the menu line.  The selection
> table is still selecting via @rn="24".  The record gets selected and
> crashes.

Good.  The output format and the output processing can be eliminated
and replaced by your null versions, and it still crashes.

[...]
> > How many lookups are in the processing tables that are left?
> In automatic processing?  Let me count... 13.
> What if I just remark out those lookups and the lines associated with
them?
> Is that going to affect what you are trying to determine?

If it still crashes, then that's not a problem.  If it doesn't crash,
then you can either experiment to narrow down which ones need to be
kept and which ones can be eliminated and still have it crash, or you
can simply keep all 13.

> My suggestion was to replace all of the lookups here with lookups that
> use R=("1") rather than whatever you're currently using.

You now have at most 13 lookups to modify ro R=("1") and see if it
still crashes.

[...]
> > You say the entire file set takes 600MB.  Does it really need that much
> > data to duplicate the crash?
>
> No, the entire application takes approx. 600 MB of data, some of which is
> under NDA.  That data is not necessary to create the crash.  It is just
> easier for me to send you the entire thing than to have to pick and choose
> which files and data I should send you.

On the other hand, you say that you can't send the entire application
because of the NDA.

However, at this point you have at most 14 files (assuming that you must
keep all 13 lookups, and they're each a different file), and if it still
fails with R=("1") then all would be needed from the lookup files would
be the map, a 1-record key, and the automatic index definitions.  (And
all that you need from the main file are the map, record 24 from the key
file, the two processing tables that are still left, and the automatic
index definitions.)

That would probably zip down to only a few K and include nothing that
would violate the NDA.

--
+-------------------------+--------------------+----------------------------
-+
| Kenneth J. Brody        | www.hvcomputer.com |
|
| kenbrody at spamcop.net | www.fptech.com     | #include <std_disclaimer.h>
|
+-------------------------+--------------------+----------------------------
-+