Re: multi-arch and breaking ``maintenance only'' targets (eg a29k)

[Andrew Cagney <ac131313@cygnus.com>]

No one commented.  I'm going to send out an announcement indicating that 
I intend removing a29k.

		Andrew


> Hello,
> 
> The a29k target currently builds but isn't being maintained.  Post 5.1 it will either need to be multi-arched or will find its self dieing a slow death.  That death, however, may come sooner than some might think.
> 
> I'm about to post a patch that eliminates the macro HOST_BYTE_ORDER from GDB.  Unfortunatly the a29k target has code like:
> 
>     #ifdef TARGET_BYTE_ORDER == HOST_BYTE_ORDER
>     #define MD(X)
>         .....
>     #else
>     #define MD(X)
>         .....
>     #endif
> 
> and that will no longer work.  Consequently, I intend adding a #error ... to the a29k so that it is clearly broken.
> 
> With this comes a definite shift in policy.  Up until now developers were to expected to keep non-multi-arch targets compiling.  If the above goes through, it will mean that non-multi-arch targets that are not being maintained can be broken on a case by case basis.
> 
> enjoy,
> Andrew