: We are having a vanishing problem with the runtime
: version of our Visual Age application. The app comes
: up and we open another screen via a menu. The screen
: is displayed and a few queries are done, and then
: BOOM, the whole VA app is gone without leaving a
: walkback or any clue whatsover as to why or where the
: problem lives.

: Has anyone out there in VA land witnessed such a
: problem?

: +aShiM+

We have seen the same thing but only when really stressing an app.  I had
suspected our problem was a multi-thread problem.  Are you doing any of that?

Todd Meadows
Sr Programmer - Architecture
H.B. Fuller Company

This is an indication of the VM abending.

If you are using odbc try loading all threads/odbc patches from
http://www.software.ibm.com/software/ad/visfixes.html.

I traced VM abends to a method in AbtThreadsApp for deleting a thread.
This method was being called by odbc when disconnecting a connection.

The VM abended when I had multiple connections and wanted to terminate one.

The funny thing is that when I put a call to the transcript in that method,
the VM
did not abend. I assume writing to the transcript performed some low-level
synchronization
that averted the problem.

Also, the abend happened on NT4.0 and not on Win95. Because version 3.0a is
not
officially supported on NT4.0, we are hoping that the new version will
resolve this problem.

Regards,

Faisal

We had the meltdown problem in our shop, and IBM said that there was a
problem with the garbage collection trying to clean up objects that were
no longer there.  This happens when you save the image when a database
connection is still up. If you are doing a database connection, you can
see if it's up from the Organizer (Options, Database Connections). At
first, your development environment will not show any problems, but each
time you restart VisualAge, the meltdown will happen sooner and sooner,
until it dies at the first garbage collection cycle.

This problem will be exported into your runtime, which will melt down as
well.

The fix for us was to change, in AbtIbmSqlda,
allocateMemoryForRow:copyData:

comment out both places where you see abtAddToBeFinalized

and in AbtRecordStructureBaseApp, in OSObject, within the category
AbtRec-API, the method
abtMoveToOSMemory

comment out abtAddToBeFinalized.

Of course, if you're already getting the meltdown, then your development
image is already corrupted.  Before making these changes, get a fresh
image.

Under OS2, this meltdown shows as a SYS3175, but under Windows NT or
Windows 95, it just disappears.
--
The condition upon which God hath given liberty
to man is eternal vigilance;
which condition if he break,
servitude is at once the consequence of his crime,
and the punishment of his guilt.

John Philpot Curran (1750-1817)
Irish lawyer, politician.
"The Right of Election of the Lord Mayor of Dublin,"
speech, 10 July 1790, Dublin.

You can get to the FTP site with fixes by exploring the VisualAge
URL in my sig below ...

--

           VisualAge: <http://www.software.ibm.com/software/ad/visage/>
                                           Java: Bean there, done that.