Re: orbitcpp 1.3.8 generated code and gcc 2.95 issue

[Bowie Owens <bowie owens csiro au>]

Alexander Nedotsukov wrote:

> Hi,
>
> In attempt to build new libbonobomm stuff I faced this problem. Gcc 
> 2.95. don't understands contructions like this (it complains about 
> syntax error):
> template<>
>    inline ::Bonobo::Canvas::affine_slice * 
> ::Bonobo::Canvas::affineProps::alloc ()
> {
>    return ::Bonobo::Canvas::affine_alloc ();
> }
>
> Since we already in ::Bonobo::Canvas namespace scope there this can be 
> rewriten as:
> template<>
>    inline ::Bonobo::Canvas::affine_slice * affineProps::alloc ()
> {
>    return ::Bonobo::Canvas::affine_alloc ();
> }
>
> Wich satisfy crafty tool.
>
> I believe this problem was introduced in this commit:
> http://cvs.gnome.org/bonsai/cvsview2.cgi?diff_mode=context&whitespace_mode=show&file=pass_xlate.cc&branch=&root=/cvs/gnome&subdir=/orbitcpp/orbitcpp/idl-compiler&command=DIFF_FRAMESET&rev1=1.22&rev2=1.23 
>
>
> The question is what can be done to resolve the situation right way? I 
> guess the commit above was reasonable and it's may be not safe just to 
> backout it.

As I recall the reason for using the fully qualified name is to deal 
correctly with nested array types. To resolve this properly I think we 
would need to switch from generating code of the form:

namespace X {

Ret Klass::method() { ... }

}

to

Ret X::Klass::method() { ... }

The only hard part is making sure everwhere gets changed over to using 
fully qualified names.

--
Bowie Owens

CSIRO Mathematical & Information Sciences
phone  : +61 3 9545 8055
fax    : +61 3 9545 8080
mobile : 0425 729 875
email  : Bowie Owens csiro au