gordon.jackson@east 1998-09-04

I have done a bit more digging into this and, despite the issues that we are
seeing with the Intersolv libs (old or new), I believe that there has been some
form of regression between Beta4 and the current FCS builds (all the way back
to FCS-A. In a nutshell, when the JDBC test harness is run with the -native
switch, then the app hangs. In default (green threads) mode, depending on the
version of the OS (2.5 or 2.6) and the version of the Intersolv libs, the app
will either complete ok, hang, or (recently with the new libs) core dump.

I will be glad to assist in reproducing this.

================================================================================
Regression: yes
Build found on: JDK1.2FCS-G
Build last worked: JDK1.2Beta4

Recently, we were delivered new libraries from Intersolv, to
fix red bug 4144874 which was reported against the JDK1.2Beta4 build.

These new libraries do indeed fix the problem reported in the
JDK1.2Beta4 build. However, in the JDK1.2FCS-G build I am encountering
regressions.

The following table summarizes the test results:
Build New Libs native mode non-native mode
----- -------------- ----------- ---------------
Beta4 yes OK OK
FCS-g yes hang core dump
FCS-g no 4144874 OK

So, using the Beta4 build with the new libs we were fine.

Attachments:
The log files for the Beta4 run (native mode) to show that the original
bug is fixed. The log files the the FCS-g run(native mode and non-native mode) plus the core dump file.

The new libs from Intersolv.