breakage from r99712

Jack Howarth howarth at bromo.med.uc.edu
Sat Nov 17 12:50:15 PST 2012


On Sat, Nov 17, 2012 at 02:43:49PM -0600, Ryan Schmidt wrote:
> 
> On Nov 17, 2012, at 14:40, Jack Howarth wrote:
> 
> > On Sat, Nov 17, 2012 at 02:13:51PM -0600, Ryan Schmidt wrote:
> >> 
> >> On Nov 17, 2012, at 10:48, Jack Howarth wrote:
> >> 
> >>> It should have at the very least
> >>> converted the corefoundation variant into a no-corefoundation variant
> >> 
> >> We do not use variants whose names begin with "no" anymore. We did that before MacPorts 2.0, because the old flat registry could only remember enabled variants, not disabled variants. But the new sqlite registry, which is the only registry format used with MacPorts 2.0 and later, remembers both. Any variants you find that are still named with a "no" prefix are relics and should be renamed / converted.
> >> 
> >> From the user's perspective, instead of enabling a no_foo variant:
> >> 
> >> sudo port install example +no_foo
> >> 
> >> They should disable a foo variant:
> >> 
> >> sudo port install example -foo
> > 
> > The problem is that this is getting absurdly complex for the average user who
> > just want to install pymol. They have to puzzle out (without documentation) that
> > 'sudo port install tk +x11' is required to have the pymol graphics window open
> > and 'sudo port install tcl -corefoundation' is required for the pymol tk gui
> > to open. Not exactly a user friendly packaging environment.
> 
> Yes. I don't disagree; it should be made to "just work" for users. I just wanted to give you a side explanation on variant naming.

I guess in the near term we could modify the pymol startup script to
test if the +x11 variant of tk is active as well as the non-variant
of tcl and exit with appropriate error messages if they aren't. Not
a particular desirable fix but at least it would avoid having the
user lists swamped with the same breakage reports on pymol.
                Jack
> 


More information about the macports-dev mailing list