Board index » cppbuilder » De{*word*81} dies : ? (

De{*word*81} dies : ? (

Has anyone else had problems with the de{*word*81} bombing builder
completely?

I am running my program from the environment, and an access violation
exception gets raised ( guess I have a bug in there somewhere!).

In BCB3 at this point, I would hover the mouse over my variables,
looking for one with a dodgy value, like a pointer set to NULL for
instance.  However, if I put the mouse over any variable today (note, no
hovering!) I suddenly find myself looking at the Windows desktop.

This will need narrowing down a lot before filing a bug report, I can't
just say 'when project X fouls up, it kills the de{*word*81} just when I
need it most', I need a simple set of circustances to cause it.

Otherwise, I'll sit on it unreported.  It's always possible I am doing
something utterly foul and killing Windows, although I very much doubt
it, I don't write code that smart...

AlisdairM

 

Re:De{*word*81} dies : ? (


Quote
Alisdair Meredith <Alisdair.Mered...@BenettonFormula.com> wrote in message

news:36F90508.BC24169E@BenettonFormula.com...

Quote
> Has anyone else had problems with the de{*word*81} bombing builder
> completely?

Yes!

It was quite common in the place i worked (using BCB3). We could not
identify why it was happening... on some machines it didn't happen (or
rarely happened) and on others it was quite common. I suggest installing the
patches for BCB3 (if you didn't install them by now).

Btw: i hear BCB4 dont have this problem..at least i did not encounter any
posts regarding such problem in BCB4.

Quote

> I am running my program from the environment, and an access violation
> exception gets raised ( guess I have a bug in there somewhere!).

> In BCB3 at this point, I would hover the mouse over my variables,
> looking for one with a dodgy value, like a pointer set to NULL for
> instance.  However, if I put the mouse over any variable today (note, no
> hovering!) I suddenly find myself looking at the Windows desktop.

> This will need narrowing down a lot before filing a bug report, I can't
> just say 'when project X fouls up, it kills the de{*word*81} just when I
> need it most', I need a simple set of circustances to cause it.

> Otherwise, I'll sit on it unreported.  It's always possible I am doing
> something utterly foul and killing Windows, although I very much doubt
> it, I don't write code that smart...

> AlisdairM

Yoav
y...@netune.net

Re:De{*word*81} dies : ? (


Quote
Alisdair Meredith wrote:
> Has anyone else had problems with the de{*word*81} bombing builder
> completely?

> I am running my program from the environment, and an access violation
> exception gets raised ( guess I have a bug in there somewhere!).

> In BCB3 at this point, I would hover the mouse over my variables,
> looking for one with a dodgy value, like a pointer set to NULL for
> instance.  However, if I put the mouse over any variable today (note, no
> hovering!) I suddenly find myself looking at the Windows desktop.

> This will need narrowing down a lot before filing a bug report, I can't
> just say 'when project X fouls up, it kills the de{*word*81} just when I
> need it most', I need a simple set of circustances to cause it.

> Otherwise, I'll sit on it unreported.  It's always possible I am doing
> something utterly foul and killing Windows, although I very much doubt
> it, I don't write code that smart...

> AlisdairM

Yes,
See my post "De{*word*81} Crash (was .... "
The problem (in my case appeared to be the length of a classname (8
characters))

Bye,
Cees

Re:De{*word*81} dies : ? (


Quote
Alisdair Meredith wrote:
> [...]
> In BCB3 at this point, I would hover the mouse over my variables,
> looking for one with a dodgy value, like a pointer set to NULL for
> instance.  However, if I put the mouse over any variable today (note, no
> hovering!) I suddenly find myself looking at the Windows desktop.

I've had this problem with BCB3. You are probably using templates,
aren't you? And you can watch call stack safely, but none of the other
debug windows? If yes, then there is a very good chance that BCB4 will
solve your problems (it solved mine).

FYI, I disabled tooltip evaluation when I was working with BCB3, just to
move mouse safely 8-(

Regards,
Igor.

Re:De{*word*81} dies : ? (


: Alisdair Meredith <Alisdair.Mered...@BenettonFormula.com> wrote:

Quote
>Has anyone else had problems with the de{*word*81} bombing builder
>completely?

I only know of one from Cees...

Quote
>I am running my program from the environment, and an access violation
>exception gets raised ( guess I have a bug in there somewhere!).

Not necessarily... Is that an AV in *your* code? I know that the BCB4
de{*word*81} itself AVs on some "interesting" floating point code under
some special circumstances in the de{*word*81} environment, too.

Do you use the SCL (STL)? Is that crash possibly related to templates?

First action: does the IDE disappear also if you use the Standard VGA
graphic drivers?

Second action - isolate your bug, keep it and cut out around it until
you get both the original AV and the IDE crashing. This is a tedious
task, I know.

--
Stefan Hoffmeister      http://www.econos.de/
Due to time constraints I cannot give free face-to-face advice.
Please do apply judgement when sending email.

Re:De{*word*81} dies : ? (


Quote
Yoav Gelbfish wrote:
> Yes!

> It was quite common in the place i worked (using BCB3). We could not
> identify why it was happening... on some machines it didn't happen (or
> rarely happened) and on others it was quite common. I suggest installing the
> patches for BCB3 (if you didn't install them by now).

> Btw: i hear BCB4 dont have this problem..at least i did not encounter any
> posts regarding such problem in BCB4.

Actually, I never had any problems on BCB3 with this, I only got this
problem today with BCB4!

That's why I'm looking for a better grasp on the problem.  I think it's
more than just I generated an AV, and the de{*word*81} couldn't cope, but
finding the right set of hoops to make it jump through before it falls
over could be tricky...

AlisdairM

Re:De{*word*81} dies : ? (


Quote
Igor Nossov wrote:
> I've had this problem with BCB3. You are probably using templates,
> aren't you? And you can watch call stack safely, but none of the other
> debug windows? If yes, then there is a very good chance that BCB4 will
> solve your problems (it solved mine).

> FYI, I disabled tooltip evaluation when I was working with BCB3, just to
> move mouse safely 8-(

Actually, BCB3 was incredibly stable for me (although I remember a few
teething problems when first upgrading from V1, I can't remember what
they were, I guess I learned a new way to do things!)

It was annoying not being able to debug templates too well.
I've not had much practice with 4 yet, I think templates by their nature
are complex, but it does 'feel' easier dealing with them.  I think the
compiler errors are easier to track for instance, but I'm too busy
getting on with the job at hand to go back and make direct comparisons :
? )

AlisdairM

Re:De{*word*81} dies : ? (


Quote
Stefan Hoffmeister wrote:
> Not necessarily... Is that an AV in *your* code? I know that the BCB4
> de{*word*81} itself AVs on some "interesting" floating point code under
> some special circumstances in the de{*word*81} environment, too.

Yup, it was my AV.   Finally tracked it down.  I had somehow 'lost' a
file from my project, and when I added it back the project manager
didn't associate it's form with it, and so didn't put it back on the
autocreate list.  Suddenly I am trying to play with an un-initialized
pointer, so anything can happen!

This is exactly the sort of thing I found tooltip evaluation would
pinpoint for me in a hurry though, shame it's not as stable in BCB4.  Of
course, I don't know exactly which bit of memory I was overwriting, if
it happened to be BCB itself by chance...

I'm filing this one as a 'curious incident' unless a trend develops.

AlisdairM

Other Threads