Hi George,

If I understood correctly, the module compiles correctly. When you
start/run the graph it complains it can not find/'see' the function.
This could be related to default visibility of the objects. If you look at
the code for your block my_adder_ff, there should be a line similar to
    class *TUTORIAL_API *my_adder_ff { ....
The TUTORIAL_API (or whatever it is called) is a macro defined in api.h and
eventually evaluates (for GNUC) to
     __attribute__ ((visibility ("default")))
which ensures visibility to other modules. You can check the built library
for your module using nm command, e.g.(I'm just guessing library name
libTutorial.so)
    $ nm -C libTutorial.so | grep -i my_add
If the function is 'qualified' with t (lower case) it will not be visible.
If it is qualified with uppercase 'T', then it is visible

An easy thing to try to fix is to add to the function declaration in the
header file
    float *TUTORIAL_API *my_add(float, float)
and rebuild.

If curious about the visibility attribute, there is nice bit of an
explanation in
https://stackoverflow.com/questions/52719364/how-to-use-the-attribute-visibilitydefault

Hope that made sense
Cheers
Boris

On Thu, 21 Oct 2021 at 08:02, George Edwards <[email protected]> wrote:

> Hi Vasil,
>
> Thank you very much!
>
> I have never used the Git Repository so I am setting up one and will post
> my code once that is done. I will let you know when it is posted.
>
> Again, thank you so much for your offer.
>
> Regards,
> George
>
> On Wed, Oct 20, 2021 at 1:19 AM Vasil Velichkov <[email protected]>
> wrote:
>
>> Hi George,
>>
>> On 19/10/2021 16.55, George Edwards wrote:
>> > Below is a description of what I did:
>> > ....
>>
>> What you did is correct and should be enough.
>>
>> > When I compiled and run, it failed because it could not see the function
>> > my_add.
>>
>> What is the exact error message? How did you run it? Can you provide the
>> full source code of this test module (as a git repo or a tarball file on
>> some file sharing website).
>>
>> Regards,
>> Vasil
>>
>>

Reply via email to