In the scene below is spline is defined and then put into an array. When 
it is pulled out of the array it always is a linear spline, instead of 
the original 'curvy' spline. This used not to be the case as one of the 
meshmaker macro's depends on doing this since 3.5.

I cannot trace back where this behaviour originated as I only have 3.8 
alpha's and beta's here and all show this behaviour. (running win10) 


---%<------%<------%<---
#version 3.8;

global_settings {assumed_gamma 1.0}
camera {location <0,3.5,-12> look_at <0,1.5,0> angle 40}
light_source {<500,500,-500> rgb <0.8,0.9,1>}

#declare A1=spline {
   cubic_spline
   -0.5, < 1,-1, 0>
    0.0, < 1, 0, 0>
    0.5, < 2, 1, 0>
    1.0, < 1, 2, 0>
    1.5, < 1, 3, 0>
}

#declare An=array[1]{
   spline{A1},
}

#declare M=500;
union {
  #declare Spl=spline{An[0]};
  #for(I,0,50)
   sphere {Spl(I/50), 0.04 no_shadow pigment {rgb 1}}
  #end
}
---%<------%<------%<---


-- 
https://ingoogni.nl

Post a reply to this message

On 7/15/21 5:54 AM, ingo wrote:
> In the scene below is spline is defined and then put into an array. When
> it is pulled out of the array it always is a linear spline, instead of
> the original 'curvy' spline. This used not to be the case as one of the
> meshmaker macro's depends on doing this since 3.5.
> 
> I cannot trace back where this behaviour originated as I only have 3.8
> alpha's and beta's here and all show this behaviour. (running win10)
> 
> 
> ---%<------%<------%<---
> #version 3.8;
> 
> global_settings {assumed_gamma 1.0}
> camera {location <0,3.5,-12> look_at <0,1.5,0> angle 40}
> light_source {<500,500,-500> rgb <0.8,0.9,1>}
> 
> #declare A1=spline {
>     cubic_spline
>     -0.5, < 1,-1, 0>
>      0.0, < 1, 0, 0>
>      0.5, < 2, 1, 0>
>      1.0, < 1, 2, 0>
>      1.5, < 1, 3, 0>
> }
> 
> #declare An=array[1]{
>     spline{A1},
> }
> 
> #declare M=500;
> union {
>    #declare Spl=spline{An[0]};
>    #for(I,0,50)
>     sphere {Spl(I/50), 0.04 no_shadow pigment {rgb 1}}
>    #end
> }
> ---%<------%<------%<---
> 
> 

Looks like a problem with spline{} block parsing when passing a spline 
id. The current Parse_Spline() code expects the spline type token to 
exist otherwise it assumes the intent was a linear spline.

A fix is to add some dynamic_cast 'type checking' to:

GenericSpline *Parser::Parse_Spline()

in parser_expressions.cpp so it looks something like:

     if (!New)
     {
         if (Old)
         {
             if (dynamic_cast<LinearSpline *>(Old)!=nullptr)
                 New = new LinearSpline(*Old);
             else if (dynamic_cast<QuadraticSpline *>(Old)!=nullptr)
                 New = new QuadraticSpline(*Old);
             else if (dynamic_cast<CatmullRomSpline *>(Old)!=nullptr)
                 New = new CatmullRomSpline(*Old);
             else if (dynamic_cast<NaturalSpline *>(Old)!=nullptr)
                 New = new NaturalSpline(*Old);
             else
                 throw POV_EXCEPTION_STRING("Internal Error. Pointer to 
existing spline not of a known type.");
//          New = new LinearSpline(*Old); // <-- what is done currently.
         }
         else
             New = new LinearSpline();
     }

Bill P.

Post a reply to this message

Am 15.07.2021 um 13:29 schrieb William F Pokorny:

> Looks like a problem with spline{} block parsing when passing a spline 
> id. The current Parse_Spline() code expects the spline type token to 
> exist otherwise it assumes the intent was a linear spline.
> 
> A fix is to add some dynamic_cast 'type checking' to:
> 
> GenericSpline *Parser::Parse_Spline()

Fortunately, there's a more elegant solution: The `GenericSpline` class 
has a virtual member function, `Clone()`, that is implemented by each 
subclass to return a newly constructed copy of whatever the actual type 
of the object is.

So the code in question can remain as simple as:

     if (!New)
     {
         if (Old)
             New = Old->Clone();
         else
             New = new LinearSpline();
     }

Will you do the honors, or shall I?

Post a reply to this message

On 7/15/21 4:56 PM, clipka wrote:
> Am 15.07.2021 um 13:29 schrieb William F Pokorny:
> 
>> Looks like a problem with spline{} block parsing when passing a spline 
>> id. The current Parse_Spline() code expects the spline type token to 
>> exist otherwise it assumes the intent was a linear spline.
>>
>> A fix is to add some dynamic_cast 'type checking' to:
>>
>> GenericSpline *Parser::Parse_Spline()
> 
> Fortunately, there's a more elegant solution: The `GenericSpline` class 
> has a virtual member function, `Clone()`, that is implemented by each 
> subclass to return a newly constructed copy of whatever the actual type 
> of the object is.
> 
> So the code in question can remain as simple as:
> 
>      if (!New)
>      {
>          if (Old)
>              New = Old->Clone();
>          else
>              New = new LinearSpline();
>      }
> 
> Will you do the honors, or shall I?

Please go ahead.

I'll be updating my own pile of code to your cleaner fix - thank you.

Bill P.

Post a reply to this message

>in news:60f1546d$1@news.povray.org William F Pokorny wrote:
>> On 7/15/21 4:56 PM, clipka wrote:

Thanks.

-- 
https://ingoogni.nl

Post a reply to this message