OpenMotif on OSX El Capitan

It has been really annoying for over a month now: I have some old analysis codes that are Motif-based and after the OS X El Capitan upgrade last month I could not get them to compile properly.

The problem started when I tried to upgrade macports (I count on their database for old codes/libraries support instead of fink or homebrew). Then all went ok till the point I asked for openmotif libraries. They are based on gcc48 and present a persistive dependency during installation. After some digging I found that problem originated in one of the openmotif dependencies -the package ifeffit- that asks for gcc48. This version of gcc is older, but there is likely a bug in macports/Xcode7 that keeps it alive due to such dependency constraints, as in ifeffit. The flag +gfortran49 after the sudo port upgrade ifeffit, saved the day and had openmotif fully upgraded at last.

However, when I ran my analysis code, a new problem appeared: the new SIP policy by Apple that prevents anyone (even the superuser) to access some “critical” directories of the system, including /usr

The latest update had killed all symbolic links for home-compiled dylib files that had been placed under /usr. There were my openmotif libs as well! A solution was nowhere to be seen whatever I tried. After some extensive reading online I found some advice in a Brazilian site having similar problems but for the popular plotting/fitting package Grace (xmgrace). Here is the link (update: new working link).

I decided to follow the steps suggested, circumventing the SIP. I booted in recovery mode, disabled SIP, rebooted, created my symbolic links as needed, rebooted again in recovery, re-enabled SIP and after a last reboot, here I am reporting on a successful operation with the patient being completely healthy. My analysis code (under X11) works fine, all looks ok.

This SIP thing is really a problem. It is the first time I cursed Apple on their operating system configurations. After 12+ years of continuously working on OS X! Let’s see if another problem will pop up in the near future. Till then, I have plenty of spectra to analyze.