Switch enable-multilib to default

[Chris Johns]

Ralf Corsepius wrote:
> On Thu, 2008-02-21 at 14:47 +1100, Chris Johns wrote:
>> Ralf Corsepius wrote:
>>> The purpose of this step is to encourage people to build RTEMS
>>> multilib'ed, to extend exposure of multilibs and to shake out
>>> multilib-related issues early in RTEMS development process.
>>> (ATM, we have a pretty poor footprint on such issues)
>>>
>> I think this change is a good idea although it does increase build time.
> It doesn't. All it does is forcing users who want to stay with the
> "one-tree"/"rebuild everything per-bsp" to pass --disable-multilib to
> the toplevel configure.

Yes. I should have said by default.

> 
> For users who work on several BSPs at the same time (using several
> multilib variants), or users who work on a BSP or applications (those
> who don't change cpukit), almost nothing changes.
> 
> To the contrary, it speeds up "per-BSP compliation time".
> 

I agree and welcome the evolution of the cpukit and the ability to one day 
have RTEMS installed via an RPM.

>> I am concerned about the current state of the m68k target given it will not 
>> build with the --enable-multilib option due to GCC PR32307. Does this mean we 
>> will not be able to ship a 4.9 release until it is fixed ?
> AFAICT, it seems to be fixed in gcc-4.3.x ;)

Ok.

Regards
Chris