Hi Christoph, On Fri, Oct 26, 2018 at 06:29:50PM +0200, Christoph Biedl wrote: > Helmut Grohne wrote... > > > The debian policy section 4.6 requires that when build commands fail, > > the build should abort. In case bsd-finger's ./configure aborts, the > > error is ignored and the build attempts to continue potentially > > resulting in a broken package. Adding "set -e; " (as suggested by the > > policy) before the conditional fixes this issue. > > I doubt this is sufficient. There are squences like > > | echo 'no' > | echo 'Compiler '"$CC"' does not exist or cannot compile C; try > another.' > | rm -f __conftest* > + exit > > that rather should exit nonzero as fixed in > > https://sources.debian.org/patches/netkit-tftp/0.17-21/875885.patch/
The link is presently 500, but I agree with the overall idea. > Also I reckon all packages that use a configure "This file was generated > by confgen version 2" are affected from this issue. Do you agree this > applies to > > biff > linux-ftpd > linux-ftpd-ssl > netkit-bootparamd > netkit-ftp > netkit-ftp-ssl > netkit-ntalk > netkit-rsh > netkit-rusers > netkit-rwall > netkit-rwho > netkit-telnet > netkit-telnet-ssl > > as well? > > Greetings from the Karlsruhe BSP, I actually wanted to *de*crease the RC > bug count ... I actually think the issue is even worse than you project it. There also is #911154 (netkit-ntalk misses the generator for configure). I not only agree that the policy 4.6 violation affects all these, but that the dfsg violation likely affects all of these. Now given the dfsg violation, I think the best path forward would be generating these scripts from source (fixing the dfsg violation) and then fixing the generator (fixing the policy 4.6 violation once rather than 13 times). I think we can ensure that the Karlsruhe BSP does not run out of rc bugs now. Helmut
