removed already deprecated prepare_debian_package.cmake, which has just been added and will unnecessarily cause conflicts in the anyway deprecated way to create debian packages

Martin Killenberg authored 8 years ago and Martin Killenberg committed 8 years ago

use readelf instead of readlink to extract the DOOCS version and do not strip the system-dependent part of the version (e.g. -xenial1), since it is needed to identify the version correctly.

undid the changes to use pbuilder in make_debian_package.sh. This conceptualluy does not work and will completely be changed in future.

added prepare_debian_package.cmake (currently just a copy of the version in the DOOCSwrappers library)

Warning: this commit will create conflicts, but it is anyway needed to adapt the projects using this templates due to the last commits.

fixed that the configuration for loop for replacing the DEBVERSION and SOVERSION in the debian_from_template was running on the wrong (non-existing) directory

- replace DEBVERSION in file names in the debian_from_template directory
This is needed, since the debian version is not yet known in cmake.

added cmake script to set the version number cmake variables as needed by other scripts of the project-template

Added the new getDebianBuildVersion script and use it inside make_debian_package.sh.in to obtain the version
IMPORTANT: Projects importing this commit must edit the (newly created) DebianPackaging.config file!

- Quotes around input variables when doing string replace avoid existing semicolons from being deleted.
- The project target library itself was missing in the linker flags.

Revert "Capture only coverage data for the current project in the first place, instead of extracting it from all coverage data in a second step. This seems to be needed as extraction spuriously might fail in some cases."

This reverts commit ce14dcd0.

Reason for the revert: The change did not work on Ubuntu 12.04 due to an old version of lcov.

Capture only coverage data for the current project in the first place, instead of extracting it from all coverage data in a second step. This seems to be needed as extraction spuriously might fail in some cases.