Migration issue

Daniel J. Luke dluke at geeklair.net
Fri Jan 13 14:45:10 CET 2017

On Jan 13, 2017, at 8:26 AM, Adam Dershowitz <dersh at alum.mit.edu> wrote:
> I uninstalled all ports, and then tried to just install my requested ports, with variants, that I wanted.  Almost all were just default and none of these were +universal.  But, when it got to install wine-devel, it proceeded to install a whole bunch of +universal ports.  

yes, wine and wine-devel are i386 only (the portfile has a comment with a link to http://www.winehq.org/pipermail/wine-devel/2014-February/103074.html)

> The very strange thing is that it was a different set from on my prior machine (the original source for myports.txt) and on the new machine (where things were working, but there were a whole bunch of universal ports, that I didn’t expect were necessary).  For example, it built and installed a universal version of ghostscript.  And, it tried and failed to build a universal version of texlive-bin.  So, just before I uninstalled everything I had wine-devel installed, but it hadn’t needed texlive-bin, or ghostcript (or a whole bunch of others) to be universal.  But, now when I build in a different order, it does, and it fails.

The dependency engine in MacPorts doesn't really handle variants, so I always expect everything that does magic with variants to have problems like this (the difference in +universal ports probably just depends on the order that things were installed, when they're being pulled in as a dependency of something that is +universal, they get +universal when they might not get it if they were installed already).

> So, it seems that the order is critical to install certain ports, both in terms of which dependencies end up being +universal, and if the build can succeed at all!  

if it doesn't build, it's a bug that we'll want to fix.

> And, now I don’t know how to get wine, or wine-devel to install.  Any ideas?

we would need to see your failing build log to help.
Daniel J. Luke

More information about the macports-users mailing list