Hi, I just made a bug report for Fedora maxima package where I attached a patch I have been keeping in Mandriva for quite some time. It is required to allow sagemath work properly with maxima using clisp as backend.
The fedora bug report is at https://bugzilla.redhat.com/show_bug.cgi?id=837142
But I am afraid the patch is not fully complete (in the sense of maxima's --disable-readline option), and I just workaround it differently in my sagemath package for gcl with this patch http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages/cooker/sagemath/current/SOURCES/sage-4.8-maxima.patch?view=markup that I will adapt to fedora; note the "":lisp #+gcl (progn (si:readline-off) (setf *error-output* (open "/dev/stderr" :direction :output)..."" in the patch...
Thanks
The redhat patch for clisp seems reasonable. The mandriva patch seems to be unreachable.
Why are you setting *error-output* to stderr for gcl? Why is that needed?
Sorry about link. Previous one pointed to "head" commit, but I renamed the file. This should work
http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages/cooker/sagemath/current/SOURCES/sage-maxima.patch?revision=822115&view=markup
The patch sets *error-output* to stderr because by default it is aliased to *terminal-io* (probably due to the pexpect python code allocating a pty where it talks to maxima), and that causes sagemath to dead lock due to failing to parse maxima/gcl output as stdout and stderr are in the same fd. Note that I maintain these patches in my Mandriva, and now my (not yet "official") Fedora sagemath package. Upstream sagemath supports only the ecl maxima backend (that is now built in Fedora maxima since this patch http://pkgs.fedoraproject.org/cgit/maxima.git/commit/?id=ea0b7f2ee66bbc08de07c83217ac44790147c581\)