Z.rw.fragment moves the entire temp file, making a copy, and then (when
frag is finished) the temp file is moved back.

This is why a fragment causes such a performance penalty to a report
that has a largish sort file.

I highly suspect that the five second report has no temp sort file, or a
very small one (like one record), and the five minute report has a 
Larger temp sort file.

The solution is to SEG to the PHA directory, open prefixes to the PHA
database and dictionaries, and run %PHA.MAR.allergies, then
SEG back to the original directory and restore the prefixes.

You have opened a task with the RW group.  They are going to be in a
bind to offer you a solution because they have some kind of
semi-official position that fragments are the only approved way to cross
applications.  

A creative solution would be for them to write you a Z or MIS utility
that would fetch allergies, that would keep the Segment crossing and
prefix opening and restoring code on the NPR Programmer's side of
things.

I doubt that this problem is due to the 5.5 update, unless your general
response time has slowed and that has caused the performance sapping
behavior of fragments to become more of an issue.



Joe Cocuzzo
Vice President
NPR Services
Iatric Systems, Inc.
Phone/Fax: (978) 805-4115
Email: [EMAIL PROTECTED]
Web: www.iatric.com


-----Original Message-----
From: [email protected] [mailto:[EMAIL PROTECTED] On Behalf
Of Herb Bromenshenkel
Sent: Wednesday, March 21, 2007 12:34 PM
To: [email protected]
Subject: [MEDITECH-L] NPR Slowness with Fragments (depending on
initialreport) post 5.5

We have the most bizarre issue since 5.5 update.

We have two reports (both call the exact same fragment), both supply the
exact same info for the select (verified by %Z.ddc)

I run one report - the fragment runs fast (5 seconds) (this report is
much more complex than the 2nd) I run the other report - easily a minute

I can even hard code the fragment with the select (passing @urn from
ADM) and run it by itself. Runs about 5 seconds.
I use the "slow" report to run the hard coded fragment, it hits the
fragment right away, but takes well over a minute to process it (even
though the select is already coded).

Other FYI: The fragment is pulling the patient's allergies using
meditech's mar allergy routine:
%PHA.MAR.allergies(patient,70,"",1,"",2)


I called Meditech's NPR and since have opened a task.  They said the
slowness has been reported before but didn't have a solution.

Has anyone else seen this issue?  Any suggestions would be greatly
appreciated. I just can't fathom why the same fragment (with hard coded
selects) would run so much slower when called from one report versus
another. 

Herb Bromenshenkel RPh/Clinical Analyst
1300 Anne Street NW
Bemidji, MN 55601
Voice: 218-333-5527
Fax: 218-333-5889
[EMAIL PROTECTED]





=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
To subscribe or unsubscribe to the meditech-l, visit MTUsers.COM.

To check the status of the meditech-l, visit MTUsers.NET.

For help, email [EMAIL PROTECTED]
______________________________________
meditech-l mailing list
[email protected]
http://mtusers.com/mailman/listinfo/meditech-l

=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=*=
To subscribe or unsubscribe to the meditech-l, visit MTUsers.NET.

To check the status of the meditech-l, visit MTUsers.NET.

For help, email [EMAIL PROTECTED]

Visit the MTUsers WikiPedia at MTUsers.NET/mwiki
______________________________________
meditech-l mailing list
[email protected]
http://mtusers.com/mailman/listinfo/meditech-l

Reply via email to