Re: gplot.txt

[Daniel J Sebald]

On 03/05/2016 07:02 PM, Ben Abbott wrote:
> > On Mar 5, 2016, at 7:06 PM, Ben Abbott<address@hidden>  wrote:
> >
> > > On Mar 5, 2016, at 1:54 AM, Daniel J Sebald<address@hidden>  wrote:
> > >
> > > On 03/04/2016 09:40 AM, Daniel J Sebald wrote:
> > > > On 03/04/2016 07:44 AM, Ben Abbott wrote:
> > > [snip]
> > > > > Dan's patch didn't resolve the speye segfault. I'll move on to the
> > > > > changesets isolated by bisecting.
> > > >
> > > > OK, thanks Ben. I will follow up with more. Go ahead with bisecting, and
> > > > I'll approach just from looking at the code.
> > > >
> > > > Dan
> > >
> > > Ben,
> > >
> > > Here's another small patch for you to apply.  It's meant to just hunt around 
> > > for where the problem might be--prints out the value of the rep pointer before 
> > > and after the object creation and copy.  Here is what I see when running this:
> > >
> > > octave:1>  speye(10);
> > > Before object creation and copy:  rep = 24093664
> > > After object creation and copy:  rep = 24099376
> > > Before object creation and copy:  rep = 24096528
> > > After object creation and copy:  rep = 24099808
> > >
> > > I'm curious how this action works with your compiler.
> > >
> > > Dan
> > > <octave-mac_speye_fail-djs_2016mar04.diff>
> >
> > Sorry for the delay. Below is what I get …
> >
> > > > speye(10)
> > Before object creation and copy:  rep = 140417476007088
> > panic: Segmentation fault: 11 -- stopping myself...
> > attempting to save variables to 'octave-workspace'...
> > save to 'octave-workspace' complete
> > Segmentation fault: 11
> >
> > Ben
>
> Just in case the seg-fault’s I see while building the docs is related ...
>
>    GEN      doc/interpreter/splinefit3.eps
> Before object creation and copy:  rep = 140389001213056
> After object creation and copy:  rep = 140389001213056
> Before object creation and copy:  rep = 140389005748384
> After object creation and copy:  rep = 140389005748384
> Before object creation and copy:  rep = 140389005748384
> After object creation and copy:  rep = 140389005748384
> Before object creation and copy:  rep = 140389005770816
> After object creation and copy:  rep = 140389005770816
> Before object creation and copy:  rep = 140389007043568
> After object creation and copy:  rep = 140389007043568
> Before object creation and copy:  rep = 140389007051872
> After object creation and copy:  rep = 140389007051872
> Before object creation and copy:  rep = 140389007043568
> After object creation and copy:  rep = 140389007043568
> Before object creation and copy:  rep = 140389007051872
> After object creation and copy:  rep = 140389007051872
> panic: Segmentation fault: 11 -- stopping myself...
> attempting to save variables to 'octave-workspace'...
> save to 'octave-workspace' complete
> octave exited with signal 11
>
> Ben

I would conclude that the problem above isn't because of the as_array() 
function in this case.  I say that because all the index code above has 
an "After" after the "Before".  Also notice that the addresses are the 
same before and after in all cases.  For the speye(10) example, I found 
that the address Before and After change.  In other words, for all the 
cases above, this test fails:

  if (rep->idx_class () != class_vector)
    *this = idx_vector (as_array (), extent (0));

and the idx_vector() function is not called.

So your second example here is a different issue, but it may be the same 
kind of issue.  Rik will know how extensive this type of OCTAVE_NORETURN 
discrepancy between base class and inheriting classes is.

Dan