Re: [Chicken-users] segfault is not enough

[Joerg F. Wittenberger]

Felix Winkelmann <address@hidden> writes:

> Joerg F. Wittenberger wrote:
> > Hi Felix,
> > I do have a problem to isolate the problem.  I just have been able to
> > fix one instance of the segfault problem by rearranging the
> > definitions in the source code.
> > Now I have:
> > (define rfc-822-time-string-format "~a, ~d ~b ~Y ~H:~M:~S ~z")
> > (define (rfc-822-timestring t)
> >   (srfi19:date->string t rfc-822-time-string-format))
> > (define (logerr . args)
> >   (display (rfc-822-timestring (current-date (timezone-offset)))
> >            (current-error-port))
> >   (apply format (current-error-port) args)
> >   (flush-output-port (current-error-port)))
> > Before the `rfc-822-timestring' was defined in the same file a few
> > more lines after `logerr'.  A test use of logerr appeared immedately
> > after it's definition.
> > Does this info already help?
> >
> 
> Hm. No. I assume `flush-output-port' is defined, right?

Right.  A macro.

> Are you sure you are not using anything like -unsafe,
> (declare (unsafe)), (declare (no-XXx-checks ...)) or
> -optimize-level 3?

The flags where in the first mail.  Besides that I intented to have a
cond-expand with those declarations one day, no, just checked again.

> If nothing helps, could you send me the C-file (one that
> reproduces the segfault)?

I still hope that I'd eventually have a reasonable sized test case.

I tried to reproduce the problem with just one source file.  No
success.  I tried to split it into units as the original code is.
Works a well.

At the moment I could send package 2M Scheme code for you.  Do you
want that?

best regards

/Jörg

-- 
The worst of harm may often result from the best of intentions.