[49158] trunk/dports/_resources/port1.0/group/muniversal-1.0.tcl

[Rainer Müller]

On 2009-04-04 21:56, Marcus Calhoun-Lopez wrote:
> On Apr 4, 2009, at 3:31 PM, toby at macports.org wrote:
>> Message
>> More fixes to handle recent base change.
>> Also, remove insanity from muniversal_get_arch_flag
>> Modified Paths
>> 	• trunk/dports/_resources/port1.0/group/muniversal-1.0.tcl
>>
> ...
>>  proc muniversal_get_arch_flag {arch} {
>>  -    global os.arch
>>  -    # Prefer -m to -arch
>> -    set archf "-arch ${arch}"
>> -    if { ${os.arch}=="i386" && ${arch}=="i386" } {
>> -        set archf -m32
>> -    } elseif { ${os.arch}=="i386" && ${arch}=="x86_64" } {
>> -        set archf -m64
>> -    } elseif { ${os.arch}=="powerpc" && ${arch}=="ppc" } {
>> -        set archf -m32
>> -    } elseif { ${os.arch}=="powerpc" && ${arch}=="ppc64" } {
>> -        set archf -m64
>> -    }
>> -    return ${archf}
>> +    return "-arch ${arch}"
>> }
>  
> Could you please explain why this change was made?
> Why was the old way "insanity?"
> 
> The old way helped with endian issues.
> See http://thread.gmane.org/gmane.os.apple.macports.devel/6859/focus=6862
> 
> It also made fortran use possible.
> gfortran seems to work with -m64 but does not recognize -arch x86_64.
> 
> It also precludes the use of gcc-mp-* for building 32/64-bit universals.

Although -arch seems to be easier, you have a valid point.

By the way, what is the current status of muniversal? I see you are
converting more and more ports to use it. Is it ready yet to be
integrated into base?

Rainer