cvs commit: ports/deskutils/gourmet Makefile

[Kirill Ponomarew]

On Tue, Aug 16, 2005 at 02:27:25PM -0400, Adam Weinberger wrote:
> Kirill Ponomarew wrote:
> >krion       2005-08-16 09:14:43 UTC
> >
> >  FreeBSD ports repository
> >
> >  Modified files:
> >    deskutils/gourmet    Makefile 
> >  Log:
> >  Fix dependencies and build.
> >  
> >  Approved by:    portmgr (implicit)
> 
> This won't fix the build at all. Look at the pre-install: target. 
> gourmet requires a python-enabled metakit, and the only way to achieve 
> that is to have WITH_METAKIT_PYTHON=yes defined when building metakit. 
> All you've accomplished here is ensuring that the port is still BROKEN, 
> but in a different way.
> 
> Your change may look better from the standpoint of pointyhat logs, but 
> in reality, gourmet's dependency is more logically on libmk4py.so than 
> on libmk4.so.

I don't know how it would change the functionality of gourmet if
metakit is built without python, but it would make sense to add
NO_PACKAGE into Makefile to prevent cluster from building it until
the problem is resolved.  Feel free to add NO_PACKAGE/RESTRICTED if
you know that package will be unusable for users.

> The real issue lies within metakit. There is no python-enabled metakit 
> slave port, and one cannot choose to enable python based solely upon 
> whether python is already installed (you can only check for 
> ${PYTHON_CMD} after bsd.port.pre.mk, but you can only define WITH_PYTHON 
> *before* bsd.port.pre.mk... python could really use a WANT_PYTHON switch).
> 
> I'd prefer to see this change backed out, but it doesn't make much 
> difference either way. In the future, though, could you please run 
> changes like this by me first? I recognize that you're doing your 
> portmgrial duties here, but surely one of my maintainer-based rights is 
> to challenge changes like this *before* they're committed.

Yeah, you're right, sorry for that.

-Kirill