Re: FW: Error compiling gnustep-startup-0.11.0 using gcc 4.0.0

[Richard Frith-Macdonald]

On 2005-06-20 17:43:06 +0100 Adam Fedor <fedor@doc.com> wrote:

> Does this seem like a reasonable solution? Considering the code isn't 
> supposed to be used anyway? Or perhaps we should really make good on our 
> promise to depreciate this code?

I think it's a harmless change.

It's a long time since I looked at the mframe stuff though ... so I'm not 
sure we can simply remove all of it (which would be nice), and my 
inclination is to just leave it in place and accept patches ... until/unless 
someone gets time to look at it carefully.


> -----Original Message-----
>  From: Cay Horstmann [mailto:cay@horstmann.com]
> Sent: Saturday, June 18, 2005 8:32 PM
> To: Adam Fedor
> Subject: Re: Error compiling gnustep-startup-0.11.0 using gcc 4.0.0
>
>
> Hi! I didn't want to join the mailing list for this, but I found a 
> workaround 
> and thought you might be able to make use of it. The problem is around line 
> 1711 in mframe.m. Replacing size (which was an int initialized with 0) with 
> 4
>
> case _C_ARY_B:
>      case _C_STRUCT_B:
>      case _C_UNION_B:
>        {
>          typedef struct {
>            char  val[/* size */ 4];
>          } block;
>          inline block retframe_block(void *rframe)
>          {
>            __builtin_return (rframe);
>          }
>          *(block*)buffer = retframe_block(retframe);
>          break;
>        }
>
> I am sorry to say that I don't understand what I am doing, but I was able 
> to 
> build GNUStep and GWorkspace after this change.
>
> Hope that helps,
>
> Cay