[DGD] Broken Pipe? yee gads

[Felix A. Croes]

Michael McKiel <crashnbrn71 at yahoo.ca> wrote:

> Though it doesn't look like there is any help on this
> issue, unless one uses the Kernel Lib itself under
> typecheck 2. I will restate a few things that I
> thought were quite clear.
> 1/ Broken Pipe was received before GDB was used.

What you saw was a "broken pipe" message in a different program.
This is completely unrelated to the "broken pipe" condition that
you observe when running DGD within a debugger.


> 2/ I didn't realize it was "Broken Pipe" as when I
> telnetted in via TF 4. all I got was an instant
> Connection Closed, without seeing the WelcomeScreen.
> 3/ When TF 5 was used to TELNET, it was shown to be
> "Broken Pipe" - not just Connection Closed.
> 4/ GDB was used to see if any internal error could be
> found. And it just basically told me that it was a
> Broken Pipe. 

This has been explained to you in two previous postings.


> 5/ Search on the web on broken pipes: attempts to send
> a write to a closed descriptor. - not much help there.
>
> Go to the mailing list, not much help here. And a wee
> bit oh sarcasm :P

What I did was follow your assumptions through to obviously
nonsensical conclusions, in the hope that you would see that they
were <faulty> assumptions.  That was after you had already been
told exactly what was wrong with those assumptions.  Really, you
are not being smart about this. :)


> > It doesn't cause broken pipes.  What it does cause
> is > a typechecking error in a critical driver.c
> function > somewhere where no errors are allowed. 
>
> And if it's such a critical function, where an error
> isn't allowed, why does it compile and run without any
> compile or runtime errors? 

It does not run without runtime errors!  It is a fatal runtime error
that shuts it down.

Regards,
Dworkin
_________________________________________________________________
List config page:  http://list.imaginary.com/mailman/listinfo/dgd