[redland-dev] Subtle error in Redland or Ada bindings

[Dave Beckett]

On 8/5/14, 9:17 AM, Victor Porton wrote:
> In
> https://github.com/vporton/redland-bindings/tree/e0d5aa40de42fb1aad4cecb27e59c8dd566cc84a
> the Ada program src/test/special_test.adb prints '0' as the value of a pointer which is created by raptor_new_uri_from_counted_string() and must not be NULL.
> 
> I suspect that this is an error in Raptor itself not my Ada code.
> 
> Investigating this error is specifically difficult. I will try more, but it is hard.
> 
> It is also weird that replacing
> 
>    URI_1_Handle: constant RDF.Raptor.URI.Handle_Type := Get_Handle(From_String(World, URI_1));
> --     URI_1_Handle: constant RDF.Raptor.URI.Handle_Type :=
> --       C_Raptor_New_Uri_From_Counted_String(Get_Handle(World), To_C(URI_1, Append_Nul=>True), URI_1'Length);
> 
> with equivalent code
> 
> --     URI_1_Handle: constant RDF.Raptor.URI.Handle_Type := Get_Handle(From_String(World, URI_1));
>    URI_1_Handle: constant RDF.Raptor.URI.Handle_Type :=
>      C_Raptor_New_Uri_From_Counted_String(Get_Handle(World), To_C(URI_1, Append_Nul=>True), URI_1'Length);
> 
> eliminates the error. I have no idea why.

I updated the GIT code for raptor_new_uri_from_counted_string() to not
assume that there is a NUL.

Dave