#emerge maxima [...] LISPTYPE=cmucl ; export LISPTYPE ;\ CMUCL=lisp ; export CMUCL ;\ ../lisp-utils/maxima-run-lisp -i "../lisp-utils/defsystem" \ -x '(funcall (intern "OPERATE-ON-SYSTEM" :mk) "maxima" :compile :verbose t)' && \ ../lisp-utils/maxima-run-lisp -i "../lisp-utils/defsystem" \ -x '(funcall (intern "OPERATE-ON-SYSTEM" :mk) "maxima" :load :verbose t)' \ -d binary-cmucl/maxima ../lisp-utils/maxima-run-lisp: line 366: exec: lisp: not found make[1]: *** [binary-cmucl/maxima.core] Error 127 make[1]: Leaving directory `/var/tmp/portage/maxima-5.9.0-r2/work/maxima-5.9.0/src' make: *** [all-recursive] Error 1 !!! ERROR: app-sci/maxima-5.9.0-r2 failed. !!! Function src_compile, Line 41, Exitcode 2 !!! (no error message) I guess the problem here is, that the lisp environment got changed to cmucl some time ago. But cmucl does not emerge on ppc, see bug 49398. I don't know how to fix this, since at the moment maxima does not work with the formerly used clisp, either.
I was wrong, version 5.9.0, which uses clisp 2.30-r1 on my ~ppc system, emerged just fine. 5.9.0-r2 still doesn't work, of course.
A few things needed to be changed for this ebuild. On non-x86 systems, the version of lisp required has to be clisp, the other two do not build on non-x86 systems. In addition, the current ppc stable version of clisp does not compile with gcc 3.4, however, the latest x86 stable version compiles correctly. This then exposes a problem with the latest version of maxima, so the version should be bumped. I suppose we can wait until the 5.9.1 version, but I've fixed the ebuild to work with the 5.9.0.9_rc1 release. As I don't really know much about the package, I haven't tested it, but it does compile fine.
Created attachment 40099 [details] Updated maxima ebuild
Created attachment 40100 [details] Updated emacs patch for the new version of Maxima
Added cmucl to ppc-stacked-profile use.mask. Could the sci-herd look over the ebuild and do the appropriate updates?
The sci herd as put the 5.9.1 version of maxima into portage, which is great, but unfortunately, the ebuild only depends on lisp if you have clisp in your useflags. On x86, if there isn't a lisp selected, it defaults to cmucl, other arches need a default as well, probably clisp.
Apologies for not paying attention to this bug much sooner - I only have a working x86 box and so tend to not think much about other platforms. It would seem to raise some thorny issues - will the checks for USE flags need to check for current platform as well? Hmm. Did 5.9.1 + lisp updates clear up the other problems? A lot of this stuff is trick and best left to upstream, which unfortunately can be a tad slow at times.
I think this is fixed in 5.9.1. If that is not the case, feel free to reopen.
