overriding variant universal

[Ryan Schmidt]

On Feb 10, 2009, at 09:07, Jeremy Lavergne wrote:

> I'm presently working with a portfile that has lots of values  
> defined by the submitted --- mostly set to defaults.  It also seems  
> that all dependents are missing (though apple's built-in stuff will  
> end up satisfying it's needs, it seems).
>
> http://trac.macports.org/ticket/17900
>
> Anywho, the real problem I'm having is that he has the universal  
> variant overridden.  According to the guide, this is a no-no unless  
> the port won't build.  Is this another instance of where the  
> submitted created the defaults?  Any insight from those experiened  
> would be appreciated!

Looks like you've already made progress on this in the ticket and  
noted that this port uses cmake. The default universal variant is  
designed for autoconf/automake ports and does not work with cmake, so  
that's probably why the portfile was submitted with an overridden  
universal variant. You're probably right that the cmake portgroup  
could be used instead, which already provides an overridden universal  
variant.

The ruby variant claims to build a ruby extension, but does not  
declare a dependency on a ruby port. How does this work, and where  
does the extension go?

The demos variant says it installs a demo... Does the demo take a lot  
of time to build or is it very large? If not, remove the variant and  
make the port always build and install the demo. Fewer variants /  
maximal functionality up front is better; disk space is cheap. :)