Processing of port freecad fails

Ryan Schmidt ryandesign at macports.org
Fri Jan 9 11:44:00 PST 2015


On Jan 9, 2015, at 1:34 PM, Sean Farley wrote:
> 
> Ryan Schmidt writes:
> 
>> On Jan 9, 2015, at 1:16 PM, Marius Schamschula wrote:
>>> 
>>> freecad used to declare +gcc48 as the default variant until  https://trac.macports.org/changeset/130949 was committed. Given that this change set deleted fortran related default variants for a number of packages, we will likely hear about these in the near future.
>> 
>> Actually, in that revision, setting the default compiler variant was moved from the individual portfiles to the compilers portgroup.
>> 
>> Then in r130953 the default was changed from gcc48 to gcc49. But what about the ports like freecad that don't have a gcc49 variant?
> 
> Welp, that's unintended. In cases like these, I imagined portfile
> authors would override and add their own default variant. I'm a little
> busy now, so if someone feels encouraged they can add it back.
> 
> A longer term fix would be to test for gcc49 and work backwards to
> gcc48, etc.
> 
> Is there a reason why +gfortran was left out?

My guess is that a gcc49 variant could be added to freecad and that it would work fine. I guess gcc49 wasn't stable when freecad's default of gcc48 was chosen. But if you're leaving it up to each port to decide which variants to provide, then you have to either let each port decide which of them is the default, or else have much more intelligent logic in the portgroup to pick the "best" one as the default, and not make assumptions about which variants exist.




More information about the macports-users mailing list