Re: Does anybody has developed a critical application using
Posted in 1997
Santiago,
I have had memory leakage problems using PB5 and Informix. The PB
version 5.02 was apparently compiled using the wrong compilation
flags. Version 5.03 has fixed this problem.
The way to identify this problem is to run "onstat -g ses". If the
total memory and used memory keep increasing for your session, upon
multiple runs of your program, this could be your problem. We had on
program increase from 2meg to 200 meg in around 100 iterations of the
program.
Regards
David Ashby
______________________________ Reply Separator _________________________________
Subject: Does anybody has developed a critical application using PB 4
Author: "Santiago" <crespo@idecnet.com> at WCA-INET
Date: 11/12/97 9:45 PM
Hello
Me and 6 persons have been developing a critical application (working 24
hours x 7 days x 30 users accessing concurently) with:
- PowerBuilder 4 - 5.0
- Windows 3.11, NT and 95
- Informix 7.2
And it has serious PERFORMANCE problems.
The application validates the data inserted calling 5 stored procedures and
this is the problem, because the performance goes down when the application
is working about 1 hour and has inserted about 1.000 records. The SP doesn't
updates data, only reads data.
We have developed a small application that only calls one stored procedure
and the procedure ONLY ejecutes ONE SQL statement. This application has been
tested in four different Informix databases servers and the PROBLEM REMAINS.
As a FINAL solution we have developed C++ applications that only executes de
SP and returns the data to the PB application AVOIDING THAT PB CALL THE SP
and at the end THE APPLICATION WORKS WITHOUT PROBLEMS.
We have tested with about 4-5 diferent PB-Informix DLL, and INET versions
AND IT DOESN`T WORK`S.
SO MY CONCLUSION IS THAT POWERBUILDER (I've been using PB for 5 years, but
no with critical applications) IS NOT A GOOD TOOL TO DEVELOP CRITICAL
APPLICATION. AND IF YOU WANT TO DO THIS IS BETTER TO USE ANOTHER TOOL (C++
or Informix 4GL).
If anybody wants to demostrate me that this is not true, I can send you a
simple sample.
My address is:
Name: Santiago Crespo Calvo
Company: Cia de Sistemas Nekkar.
E-Mail: scc@nekkar.es or crespo@idecnet.com
WWW: www.nekkar.es
Country: SPAIN - MADRID.
TIA