We have a computational intensive background non-visual app which is run on a windows server. It was written and compiled with PowerBuilder 2017 LTR.
Largely it reads data from a SQL Server database in the network via a datastore, processes the data and writes new data out to the same database engine from another datastore. Usually we run as many of these processes as possible on the server. Typically in a few hours of processing it generates more than 100 million records in the target database.
The app has been running for years on Server 2012R2 and now upgraded the OS to 2019.
It seems to run faster on 2019 than on 2012r2 on the same hardware.
We started testing the same apps compiled under PB2019 LTR and immediately noticed a reduction in compute utilization and overall thruput.
I am investigating the cause.
As we are fairly new to PB 2019 we wonder if the was a trick we were missing.
On 2017 LTR we had to include the following in the pb.ini to maximize thruput.
[datastore behavior]
usehwnd=no
is a similar feature needed in 2019 LTR and how is it used.
Has anyone else compared background computing like this between the two versions?
But QA remarks: "Duplicate entries for:
GenerateWSAssembliesOnCompile=YES
debug_ws_metadata=YES / 1
in [Data Window] section".