Re: swig/python: making a nested class inside templated class accessible by using declaration

--===============6294498250691489912==
Content-Type: multipart/alternative; boundary="0000000000005823fa061152e4a8"

--0000000000005823fa061152e4a8
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Mon, 1 Jan 2024 at 20:35, Claudius Schn=C3=B6rr <[email protected]> =
wrote:

> Subject corrected -> 'accessible'
>
> -------------------------------------------
>
> Hello,
>
> after swig release 4.2.0 has been submitted, and improvements in
> 'using-declarations' have been stated, I run again a test from some years
> ago.
>
> I tested whether a nested class 'B' inside a templated class 'A' can be
> made accessible by a global using statement.
> There is no longer a syntax error with swig-4.2.0, however it still
> doesn't work.
> 1) Can someone quickly check whether I have overseen something?
>
> The other way would be to make the nested class global and declare it as =
a
> nested class by a local using-declaration in A. This works, but requires
> bigger recodings in the C++-library I wrap to python.
> 2) is there a timeline for when python support for nested template classe=
s
> can be expected?
>
>
> -------------------------------------------------------------------------=
------------------------------------------------------------------------
> NestedTemplate.h:
> -------------------------
>
> #include <iostream>
> using namespace std;
>
> template <typename Tnum>
> class A {
> public:
>     class B {   //nested class
>     public:
>         B() { cout << "A::B() called" << endl; };
>     };
>     A() { cout << "A() called" << endl; };
> };
>
> template <typename Tnum> using Bnest =3D typename A<Tnum>::B;
>
>
> -------------------------------------------------------------------------=
------------------------------------------------------------------------
>
> NestedTemplate.i:
> ------------------------
>
> %module NestedTemplate
>
> %feature ("flatnested",1)  A::B;
>
> %header %{
> #include "NestedTemplate.h"
> %}
> %include "NestedTemplate.h"
>
> // explicit instantiations
>
> Rename the wrapped nested class as SWIG calls the generated nested class
just B.
%rename(A_B_i) A<int>::B;


> %template(A_i)       A<int>;       //ok
>
> //this syntax would be perfect, however not supported yet
> //%template(A_i_B)     A<int>::B;  //swig syntax error
>
> //via using alias:
> %template(A_B_i)  Bnest<int>;       //with swig-4.2.0, no syntax error
>
The %template above doesn't do anything as this is not what should be
instantiated. The one and only thing to instantiate is the template A<int>
and that has already been instantiated above. I guess it ought to warn or
error out though. The instantiation named A_i generates the nested B class,
named simply as B, though.

I'd raise a bug report in Github about the nested class being called A
instead of say A_B_i though by default. I think that might be a bug.

William

--0000000000005823fa061152e4a8
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div dir=3D"ltr"><br></div><br><div class=3D"gmail_quote">=
<div dir=3D"ltr" class=3D"gmail_attr">On Mon, 1 Jan 2024 at 20:35, Claudius=
 Schn=C3=B6rr &lt;<a href=3D"mailto:[email protected]">schnoerr@mailzon=
e.com</a>&gt; wrote:<br></div><blockquote class=3D"gmail_quote" style=3D"ma=
rgin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:=
1ex">

 =20
   =20
 =20
  <div>
    <p>Subject corrected -&gt; &#39;accessible&#39;</p>
    <p>-------------------------------------------<br>
    </p>
    <p>Hello, <br>
    </p>
    <p>after swig release 4.2.0 has been submitted, and improvements in
      &#39;using-declarations&#39; have been stated, I run again a test fro=
m
      some years ago.</p>
    <p>I tested whether a nested class &#39;B&#39; inside a templated class=
 &#39;A&#39;
      can be made accessible by a global using statement.<br>
      There is no longer a syntax error with swig-4.2.0, however it
      still doesn&#39;t work.<br>
      1) Can someone quickly check whether I have overseen something?<br>
    </p>
    <p>The other way would be to make the nested class global and
      declare it as a nested class by a local using-declaration in A.
      This works, but requires bigger recodings in the C++-library I
      wrap to python.<br>
      2) is there a timeline for when python support for nested template
      classes can be expected?<br>
    </p>
    <p>--------------------------------------------------------------------=
---------------------------------------------------------------------------=
--<br>
      NestedTemplate.h:<br>
      -------------------------<br>
    </p>
    <p>#include &lt;iostream&gt;<br>
      using namespace std;<br>
      <br>
      template &lt;typename Tnum&gt;<br>
      class A {<br>
      public:<br>
      =C2=A0=C2=A0=C2=A0 class B {=C2=A0=C2=A0 //nested class<br>
      =C2=A0=C2=A0=C2=A0 public:<br>
      =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 B() { cout &lt;&lt; &quot;=
A::B() called&quot; &lt;&lt; endl; };<br>
      =C2=A0=C2=A0=C2=A0 };<br>
      =C2=A0=C2=A0=C2=A0 A() { cout &lt;&lt; &quot;A() called&quot; &lt;&lt=
; endl; };<br>
      };<br>
      <br>
      <font color=3D"#0000ff">template &lt;typename Tnum&gt; using Bnest =
=3D
        typename A&lt;Tnum&gt;::B;</font><br>
    </p>
    <p>--------------------------------------------------------------------=
---------------------------------------------------------------------------=
--</p>
    <p>NestedTemplate.i:<br>
      ------------------------<br>
    </p>
    <p>%module NestedTemplate<br>
      <br>
      %feature (&quot;flatnested&quot;,1)=C2=A0 A::B;<br>
      <br>
      %header %{<br>
      #include &quot;NestedTemplate.h&quot;<br>
      %}<br>
      %include &quot;NestedTemplate.h&quot;<br>
      <br>
      // explicit instantiations<br>
      <br></p></div></blockquote><div>Rename the wrapped nested class as SW=
IG calls the generated nested class just B.<br></div><div>%rename(A_B_i) A&=
lt;int&gt;::B;<br>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"ma=
rgin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:=
1ex"><div><p>
      %template(A_i)=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 A&lt;int&gt;;=C2=
=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 //ok<br>
      <br>
      //this syntax would be perfect, however not supported yet<br>
      //%template(A_i_B)=C2=A0=C2=A0=C2=A0=C2=A0 A&lt;int&gt;::B;=C2=A0 //s=
wig syntax error<br>
      <br>
      //via using alias:<br>
      <font color=3D"#0000ff">%template(A_B_i)=C2=A0 Bnest&lt;int&gt;;=C2=
=A0</font>=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0
      //with swig-4.2.0, no syntax error<br></p></div></blockquote><div>The=
 %template above doesn&#39;t do anything as this is not what should be inst=
antiated. The one and only thing to instantiate is the template A&lt;int&gt=
; and that has already been instantiated above.  I guess it ought to warn o=
r error out though. The instantiation named A_i generates the nested B clas=
s, named simply as B, though.<br></div><div>=C2=A0</div>I&#39;d raise a bug=
 report in Github about the nested class being called A instead of say A_B_=
i though by default. I think that might be a bug.=C2=A0</div><div class=3D"=
gmail_quote"><br></div><div class=3D"gmail_quote">William<br></div></div>

--0000000000005823fa061152e4a8--


--===============6294498250691489912==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline


--===============6294498250691489912==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
Swig-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/swig-user

--===============6294498250691489912==--