Slow performance while doing "Run command" - **SOLVED**

[Hans-Ole Kaae]

Hi,

A couple of weeks ago I asked the list for help, re.
the very slow behaviour of the 'run command' construct.

While calling this problem 'Solved' is a bit exaggerated,
I do appreciate the many great ideas and comments on this
subject.

I will especially thank Mr. Tim Snyder of Boston Scientific
for providing the solution we chose to implement. Since this
really improves performance and therefore is of general
interest - I enclose parts of Mr. Snyders reply:

"The best way to sort the data in QUICK is not to sort it.
This seems to be a paradox but I assure you that with HP's
Image database you can accomplish your goal using sorted chains.
If the data is sorted in the appropriate order then when you
access it via the corresponding path in QUICK it will appear
in the proper order with no additional processing required.
This technique is most appropriate when dealing with very
static data otherwise the overhead of maintaining the sort
sequence may offset the benefit of efficient retrieval."


Let's hope, that Cognos one day improves the performance of
the 'run command' - and while waiting, this construct will
prove efficient, even though the use of sorted items in
Image databases in general should be minimized, as we
learned many years ago...

If you would like to try this out - the changes to your
databases are minimal - mail me directly, if you need a
copy of the Schema we used.

With kind regards,

/Hans

Hans-Ole Kaae, IT-Consultant, ScanConsult·IT-Partners ApS
E-mail: hok@scanconsult.dk


-----------------------------------------------------------------
Hi,

This goes for PH 8.19 C3 on a HP 3000/987, MPE/iX 5.5 with PP5.

While checking out some old programs, I found that a normal way
of sorting in Quick in this particular system is as follows:

1) Purge a temporary file
2) Build a temporary file
3) Put the sort-statements in this temporary file
4) Purge the outfile file (temporary)
5) Build the output file (temporary)
5) Setting up a file-equation for the output file
6) Invoking sort - via CI.PUB.SYS
7) A little clean-up and housekeeping...

All using the RUN COMMAND construct.

This runs very, very slow, indeed.

Does anyone have a good idea how to improve performance
radically on this?

I guess a few of the statements could be avioded, tuned etc. but
what I go for is really cutting down the many seconds, the program
is working and the user is waiting.

Just tried to put all the handling in a separate 3GL-program.
Runs perfectly fast - but as soon as PowerHouse is calling the
program via RUN COMMAND, the process again slows dramatically.

I guess the real culprit is the RUN COMMAND construct - but it's
not easy to avoid in this context.

Of course we could re-design all the screens using Quiz and sorted
subfiles; but I suppose it should be possible to implement all this
in Qdesign using sort-intrinsics and a DO EXTERNAL construct, maybe?

All suggestions are most welcome, either thru The List or mailed
directly.

TIA!

With kind regards,

/Hans

Hans-Ole Kaae, IT-Consultant, ScanConsult·IT-Partners ApS
E-mail: hok@scanconsult.dk


= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Subscribe: "subscribe powerh-l" in message body to majordomo@lists.swau.edu
Unsubscribe: "unsubscribe powerh-l" in message to majordomo@lists.swau.edu
powerh-l@lists.swau.edu is gatewayed one-way to bit.listserv.powerh-l
This list is closed, thus to post to the list, you must be a subscriber.