[M3devel] cm3 -O (cm3cg -O3)

[Tony Hosking]

Hmm.  Not sure exactly where to put -O in the boot scripts...

On 25 May 2009, at 12:46, Jay wrote:

>
> -O is not interpreted as -O3 currently.
> One place to twiddle is
> m3-sys/cminstall/src/config-no-install/Unix.common
>
> proc m3_backend(source, object, optimize, debug) is
> ...
> <    return try_exec (m3back, args, source, "-o", object)
>>   return try_exec (m3back, "-O3", args, source, "-o", object)
> end
>
> I realize that's not ideal.
>
> I think if you:
> a) restore that so optimize => -O3
>
> b) and then maybe, I'm just looking/guessing just now
>
> scripts/regression/defs.sh
>
>  BUILDSCRIPT="./scripts/do-cm3-${BSET}.sh"
> ..
>      OMIT_GCC=yes ./scripts/do-cm3-core.sh realclean || exit 1
> ..
>      OMIT_GCC=yes ./scripts/do-cm3-core.sh realclean || exit 1
>
> probably omit one of them -- the one that uses the older compiler.
> Adding -O to these should get through with Olaf's recent changes.
>
>
> - Jay
>
>
> ----------------------------------------
>> From: hosking at cs.purdue.edu
>> To: hosking at cs.purdue.edu
>> Date: Mon, 25 May 2009 11:22:45 +1000
>> CC: m3devel at elegosoft.com
>> Subject: Re: [M3devel] cm3 -O (cm3cg -O3)
>>
>> PS Where do I look in the Tinderbox configurations to have things
>> build cm3 -O? I might be good to make that the default.
>>
>> On 25 May 2009, at 11:19, Tony Hosking wrote:
>>
>>> I've successfully bootstrapped with -O now for
>>>
>>> AMD64_DARWIN,
>>> PPC_DARWIN,
>>> I386_DARWIN, and
>>> SOLgnu,
>>>
>>> so I think the gcc-based backend is now in much better shape.
>>>
>>> The only common target missing here is x86/amd64 linux.
>>>
>>> I'll try to get around to those sometime soon.
>>