Re: [Bug-gsl] Bug - GSL_FN_FDF_EVAL_F

[Patrick Alken]

I don't fully understand the issue. In gsl_math.h, we have:

135 struct gsl_function_fdf_struct
136 {
137   double (* f) (double x, void * params);
138   double (* df) (double x, void * params);
139   void (* fdf) (double x, void * params, double * f, double * df);
140   void * params;
141 };
142
143 typedef struct gsl_function_fdf_struct gsl_function_fdf ;
144
145 #define GSL_FN_FDF_EVAL_F(FDF,x) (*((FDF)->f))(x,(FDF)->params)

Therefore there is an 'f' field of gsl_function_fdf which means the 
macro should work ok. There is no 'function' field so I don't see how 
your change will work. Also, this macro is used by several modules in 
the roots directory so it seems to be functioning properly

On 02/19/2015 07:06 PM, Edith Viau wrote:
> Hi everyone : )
>
> I think that a small bug was lost in versioning... Correct me if I am
> wrong, and I am most sorry if this issue was already brought up to you - I
> was unable to find previous mentions of this.
>
> In the gsl_math.h file, line 145. We read :
>
> #define GSL_FN_FDF_EVAL_F(FDF,x) (*((FDF)->*f)*)(x,(FDF)->params)
>
>
> It should be :
>
>
> #define GSL_FN_FDF_EVAL_F(FDF,x) (*((FDF)->*function*))(x,(FDF)->params)
>
>
> When I make the change, I am able to use the macro as intended.  The change
> is in bold.
>
>
> Let me know if there is anything or if I should provide a program sample.
>
>
> Have a great day/evening,
>
>
> Edith