Yes, the job script dumps details on libicudata.so.* and they are the same as when the script is run on a non-jenkins Ubuntu machine:
/home/jenkins/jenkins-slave/workspace/UIMA-CPP-SDK/dependencies/lib/libicudata.so.50.2: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, BuildID[sha1]=d2f443b63fcae650e5bebc60dccb9815cd8af6fb, not stripped /home/jenkins/jenkins-slave/workspace/UIMA-CPP-SDK/dependencies/lib/libicudata.so.50: symbolic link to libicudata.so.50.2 /home/jenkins/jenkins-slave/workspace/UIMA-CPP-SDK/dependencies/lib/libicudata.so: symbolic link to libicudata.so.50.2 A previous run dumped the entire tree under "dependencies" and everything expected to be there was indeed there. Eddie On Mon, Nov 18, 2019 at 9:20 AM Roberto Carlos Toledano Gómez < rct...@gmail.com> wrote: > Hello Eddie > You checked that the prefix path is in > /home/jenkins/jenkins-slave/workspace/UIMA-CPP-SDK/dependencies/ ? > Regards > > El lun., 18 nov. 2019 a las 8:12, Richard Eckart de Castilho (< > r...@apache.org>) escribió: > > > On 17. Nov 2019, at 00:39, Eddie Epstein <eaepst...@gmail.com> wrote: > > > > > > > > > test_cas: error while loading shared libraries: libicudata.so.50: > > > cannot open shared object file: No such file or directory > > > > My C/C++ heydays are a few years back, but I wonder if the test might > > somehow be misconfigured to not respect the `--with-icu=$PREFIX` > parameter? > > Maybe try explicitly including the prefix via some `-I$PREFIX`? > > > > Cheers, > > > > -- Richard >