R-mAr produces an empty, useless debuginfo package; if that's expected, adding a %define debug_package %{nil} would get rid of it. Also, some of the installed files contain buildroot traces, which generally may cause security problems. I don't know a thing about R, so I don't know if that's the case. You can reproduce this by installing the fedora-rpmdevtools package and making sure your ~/.rpmmacros has /usr/lib/rpm/check-buildroot in %__arch_install_post (see also fedora-buildrpmtree). + /usr/lib/rpm/check-buildroot Binary file /var/tmp/R-mAr-1.1-3-buildroot-scop/usr/lib/R/library/mAr/Meta/hsearch.rds matches Found '/var/tmp/R-mAr-1.1-3-buildroot-scop' in installed files; aborting error: Bad exit status from /var/tmp/rpm-tmp.73072 (%install)
(In reply to comment #0) > R-mAr produces an empty, useless debuginfo package; if that's expected, adding a > %define debug_package %{nil} would get rid of it. I am surprised as well. :-) I did not expect this to happen. :-) > Also, some of the installed files contain buildroot traces, which generally may > cause security problems. I don't know a thing about R, so I don't know if > that's the case. You can reproduce this by installing the fedora-rpmdevtools > package and making sure your ~/.rpmmacros has /usr/lib/rpm/check-buildroot in > %__arch_install_post (see also fedora-buildrpmtree). > > + /usr/lib/rpm/check-buildroot > Binary file > /var/tmp/R-mAr-1.1-3-buildroot-scop/usr/lib/R/library/mAr/Meta/hsearch.rds matches > Found '/var/tmp/R-mAr-1.1-3-buildroot-scop' in installed files; aborting > error: Bad exit status from /var/tmp/rpm-tmp.73072 (%install) Thanks for the tip. I will examine it further to determine if this is "feature" of this R-package or if it happens for other R packages, since I intend to submit more like this for review in Extras.
(In reply to comment #0) > R-mAr produces an empty, useless debuginfo package; if that's expected, adding a > %define debug_package %{nil} would get rid of it. Done in 1.1-4 already built. > Also, some of the installed files contain buildroot traces, which generally may > cause security problems. I don't know a thing about R, so I don't know if > that's the case. You can reproduce this by installing the fedora-rpmdevtools > package and making sure your ~/.rpmmacros has /usr/lib/rpm/check-buildroot in > %__arch_install_post (see also fedora-buildrpmtree). > > + /usr/lib/rpm/check-buildroot > Binary file > /var/tmp/R-mAr-1.1-3-buildroot-scop/usr/lib/R/library/mAr/Meta/hsearch.rds matches > Found '/var/tmp/R-mAr-1.1-3-buildroot-scop' in installed files; aborting > error: Bad exit status from /var/tmp/rpm-tmp.73072 (%install) This happens for every R package and it is a consequence of the way as R installs its packages. This is an upstream issue and it is not related with FE packaging.
(In reply to comment #2) > > Found '/var/tmp/R-mAr-1.1-3-buildroot-scop' in installed files; aborting > > error: Bad exit status from /var/tmp/rpm-tmp.73072 (%install) > > This happens for every R package and it is a consequence of the way as R > installs its packages. This is an upstream issue and it is not related with > FE packaging. Does that mean that it's harmless, or does R potentially do something funny with the paths, like load modules or something from those dirs? Loading them from /var/tmp would be a big security hole.
As far as I understand there is not any problem: https://stat.ethz.ch/pipermail/r-help/2006-February/086069.html I have asked this question in R-devel as it seems that all rpms, and I would expect debian as well, suffer from this.
If they are truely harmless, we should add an exception for the relevant files (*.rds ?) to /usr/lib/rpm/check-buildroot like there already exists for *.pyo, *.pyc, *.elc and .packlist. But that requires a definitive confirmation from someone who can say with confidence that it's the right thing to do.
They are harmless, see http://thread.gmane.org/gmane.comp.lang.r.devel/7069 FWIW I intend to answer some of those messages. :-) It looks like that that problem will go away in the next release that should be release soon (a couple of months). So I propose to ignore this for the moment, since as soon as the new version is released all packages will be rebuild. Is this OK with you?
Yep, works for me.