CTD bug cause overwriting of process memory

[ptomasini]

 Posted by:  Tomasini Paolo 

Attached a sample code with a very strange and dangerous bug (tested
on CTD 3.1-PTF4 and CTD 2005).

1. Put a breakpoint on SAM_User message handler in cFTableClass.Col_1.

2. Run application and when breakpoint is reached try to access to
variable m_sVariable from "Expressions" debug window...

3. You obtain a compile error ==> the variable is no really accessible
(why?, THIS IS A FIRST BUG) but CTD doesn't log an error or stop
application (try to step over). This cause an internal memory
overwrite of other CTD variables (I've discovered and proved that
after days of debug)!!! This is a second VERY DANGEROUS BUG.....

Try another situation:

Move the declaration of column Col_1 from class cFTableClass to Object
cFFormClass.Table.
Repeat the test.
Now the variable is correctly accessible: This is the expected
behavior in the first case too.

I've posted this message to Gupta log-a-bug service too...
Can somebody try to reproduce the bug on CTD 2005.1?
Any suggestion or workaround without rewriting the class hierarchy?

Regards,
Paolo Tomasini

[ptomasini]

 Posted by:  Paolo Tomasini 

[Unknown user]

 Posted by:  Unknown 

#85338
Inconsistence with the RUNTIME vs DEBUGGER when assigning an instance
variable and its evaluation in the debugger.

[Unknown user]

 Posted by:  Unknown 

# 85507
Memory overwritten in an application having a specific class hierarchy when
setting an independent instance variable

COMMENTS
As stated please see attached testcase and readme for reproduction. This is
related to defect #85338. Setting an independent instance variable
overwrites n other variable when checked in the debugger... The APP is a
TD310 application, the problem shows with both TD2005 and TD2005.1

[ptomasini]

 Posted by:  Tomasini Paolo 

After 1 year I'm still waiting these bugs are solved....
I expected gupta to be more reactive and serious!!!

Paolo

[Dave Rabelink]

Have you received a defect number ?

[ptomasini]

 Posted by:  Tomasini Paolo 

The two defect numbers are in the previous messages: #85338 - #95507.

Waiting.....
Paolo

[Jim McNamara]

 Posted by:  Jim McNamara 

Bugs that are 'show stoppers' (cause mission critical applications to fail)
go to the top of the priority list.

Bugs that multiple users report that cannot be worked around are next.

If you are the only user that has reported a defect, it will never come to
the top of the priority list unless you make a business case, in private and
with respect, to the people who manage the maintenance of the software.
(Suren, Charlie, & company)

They don't have the time and resources to chase down bugs that no one really
cares about.

[Ingo Pohl]

 Posted by:  Ingo Pohl 

I have this behavior also...

So if i want a bugfree TD, i have to reclaim every known Bug???
My thoughts were that if a bug is reported it will be solved sooner or
later.
But now i assume that the community has to report the same bug 3 or 4
times???

In my oppinion all Bugs concerning the Debugger should be solved rather
quickly, since this is the Tool to search for errors...

Also this Bug has the capacity to be undiscovered for a while since it is so
special that many users could run over it but didnt recognize it so far.

I really dont want to riot but maybe this bug should be reported by everyone
here in this forum since the debugger is one of the most essential ways to
guarantee a functional Programm for the customers buying our Product...
And even if the bug is really exotic it has to be solved!
So @ALL: Report these known Bugs #85338 and #85507

Just because im intrested in:
How many known Bugs are on the List?
How can i see the List of known Bugs?
Will there be a PTF 5 for TD 3.1?

[Mirko]

 Posted by:  Mirko BONANNO 

Ok Ingo,
if you report mines I will report yours (lol) !!!

More seriously :
I totaly agree with you ;
very often we suspect a bug on this NG BUT if someone tells us that it is a
known bug we do not (re) log-a-bug it.

Are we supposed to re-log Known bug in order to make it clinb in the
priority list ?

I wish an official answer from Gupta

Mirko

BTW & AFAIK Charlie is not working for Gupta Anymore (sigh!!!)

[Jim McNamara]

 Posted by:  Jim McNamara 

You missed my point. If you have a bug that you think is important YOU need
to follow up. You can't just log-a-bug and expect it to go to the top of
the priority list, especially if no one, including you, is making it an
important matter.

The wrong way to handle it is to log your bug, wait a year, then come back
to a public forum and shout 'STILL WAITING...'.

You are a software developer. And I think I can be quite sure that your
software has bugs. Do you fix them all immediately when someone brings them
to your attention? If you do then you have the luxury too much time on your
hands. Most of us fix the bugs that: 1. are impacting the mission; or 2.
the users are screaming about. Many bugs never get fixed because hardly
anyone notices or cares.

If you want your bug noticed or cared about you need to follow-up, in a
polite and business like fashion, with e-mail, snail mail, or a phone call
to start a dialogue that will get the priority of your issue elevated. I've
been there, done that - with very satisfactory results.

BTW, AFAIK TD 3.1 has been out of support for over a year so if the issue
still exists in versions still in engineering support you'll need to
upgrade, or at least bring your license maintenance agreement current to see
the benefit of any forth coming fix.

Just my 2 cents worth...

[ptomasini]

 Posted by:  Tomasini Paolo 

I completly disagree with Jim:

- The reported bug is present in TD 2005.1, the last version...
- The reported bug is very hidden and dangerous: the case I reported is
probably a particular case in which i could observe the effect of bug. I
(and you) cannot exclude the same effect occours in other situations... You
should be happy your customers make bug-discovery for you (but this is true
for all programmers...). This should be on of the success-keys of a
developer community!
- Every Data-Corruption bug is always an high-priority bug. For me and for
my customers... perhaps for you isn't it...
- Gupta doesn't offer a real tool to monitor/track the work-in-progress on
each ACCEPTED bug report. The only way for us is send a mail and wait
indefinitely (and without any contact) until we find our "Bugs #nnnnn" in a
release note... if we are lucky...
- One year ago I overcome the bug modifing the class hierarchy and not
adding others members variables. few days ago adding a boolean member
variable to one of my classes I crash again with the bug... And i spent a
lot of time (days...) in these situations.

So you can't answer me: Is a "stupid" problem because is only a your
problem!

Regards
Paolo Tomasini

[Jim McNamara]

 Posted by:  Jim McNamara 

Charlie has responded to a couple of posts this week.

[Jim McNamara]

 Posted by:  Jim McNamara 

You still missed my point. Let me try this again -- a bit more bluntly.

You need to contact Gupta directly. Bitching in the news group will only
get you a cold shoulder and a reputation of being a complainer.

[ptomasini]

 Posted by:  Tomasini Paolo 

I'm trying to collaborate with you and gupta!!!!
I don't like to be a complainer: i'm only trying to explain to you (and to
other community members) my difficulties and my point of view.
I don't find this is scandalous.
Perhaps you don't like these arguments be public...

Please tell me how to contact you or gupta directly.
Do you need some info to work on bug reported? Ask me!!!

Paolo