variant or platform

[Ryan Schmidt]

On Jul 10, 2007, at 18:45, Yves de Champlain wrote:

> I'm in a normalizing mood.
>
> I found out that
>
> 87 ports use variant darwin
> 10 ports use variant puredarwin
> 10 ports use variant freebsd
>  3 ports use variant powerpc
>  3 ports use variant x86

Don't forget 2 ports that use "variant macosx". Oops, one of those is  
mine...


> Now, that should be "platform", shouldn't it ?
>
> Should I go on and change everything ?

All except "variant x86", probably yes...

As for "variant x86", the correct syntax would be "platform i386",  
wouldn't it? I don't see any ports with "platform x86", and nobody  
with an Intel Mac is currently automatically getting a variant called  
"x86", as far as I know. And since those users are not complaining, I  
must conclude that whatever is in the x86 variant is not in fact  
necessary for Intel Macs, and may in fact be harmful. So each of the  
x86 variants should be specifically examined and only retained if  
found to still be necessary. I mean, up until not too long ago, the  
only way to use MacPorts on an x86 processor was to use an OS other  
than Mac OS X, right?

The "variant x86" ports are freetype1, python23 and llvm-gcc4. I sure  
don't understand what's going on in the powerpc and x86 variants of  
llvm-gcc4, why that's necessary. (Cc'ing maintainer.) As for  
python23, the x86 variant only does "configure.args-append --disable- 
toolbox-glue", which is the same thing it does for the puredarwin  
variant, which makes it sound like it would be a bad idea for this to  
be changed to "platform i386" since that would disable the toolbox  
glue on Intel Macs, where the glue should probably remain. And  
freetype1 is obviously old and unmaintained and no ports depend on  
it, so maybe we should just delete that port.