To handle this, I will need to make some extra patches to my WIP sagemath package, but would like to know if there is really a requirement to name those this way... A very good share of current "sage -testall" failures are caused by a message like: Exception raised: [...] RuntimeError: Error executing "poly.x -fv" for a polytope sequence! Output: /bin/sh: poly.x: command not found as the binary is actually installed as: $ rpm -ql palp | grep poly.x /usr/bin/palp-poly.x I suggest a new package not renaming binaries, and also installing the arxiv.org documentation pdf for palp; not sure how to make automated tests on it tough, due to the way the cgi handles the requests for the file, see the new spec... Suggested new package: Spec URL: http://fedorapeople.org/~pcpa/palp.spec SRPM URL: http://fedorapeople.org/~pcpa/palp-1.1-7.fc18.src.rpm
The rationale is here: https://fedoraproject.org/wiki/Packaging:Conflicts#Potential_Conflicting_Files There are only 4 binaries: $ repoquery -ql palp | grep bin /usr/bin/palp-class.x /usr/bin/palp-cws.x /usr/bin/palp-nef.x /usr/bin/palp-poly.x The names seem very short / generic, hence the palp- prefix. We could easily update the package to create a libexec directory with symlinks to the binaries, e.g.: /usr/libexec/palp/class.x -> /usr/bin/palp-class.x ... This would allow a user (or dependent package) to set PATH="$PATH:/usr/libexec/palp" and use the short binary names. Do you think the names are unique enough?
I think they are unique enough, and would prefer to have less sagemath patches (already have way too much to adapt from sage release to sage release). But could also add /usr/libexec/palp/ first to $PATH in the sage script. Also, only people installing sagemath should also install palp anyway, but I still think there should not exist any other package with name conflicts.
I just asked for commit access to palp so I can do a more complete help on the issue :-)
New rawhide palp package corrects the problem.