In the scripts i'm using, kdebindings has always been almost last on the list.
so in compiling the latest 4.1.85 tarballs, kdeedu failed on something PyKDE related, and so if i moved kdebindings to build and install before kdeedu, everything is good.

is this safe behaviour to have kde application suites dependent on a bindings set ?

If something is not present on your system then CMake shouldn't try and build + install it. If it doesn't do this then it is a bug.

However modules are allowed to depend on other modules, so long as building it without it is possible ( apart from support / libs / base )

As far as I can tell when we packaged kdeedu 4.1.85 for Kubuntu there was not a need for a dependency on pykde4. (Everything built fine, CMake didn't complain about any software not building due to missing dependencies) We do, however, build-depend on libboost-python-dev, so maybe that got pulled in when you installed the development header packages needed to build kdebindings for your distribution?

bcooksley wrote:If something is not present on your system then CMake shouldn't try and build + install it. If it doesn't do this then it is a bug.

However modules are allowed to depend on other modules, so long as building it without it is possible ( apart from support / libs / base )

this could be why, because i was compiling on a system that i had kde-4.1.82 on currently, i may have seen that libs/headers were there, but it failed on some undefined references.

i almost seems that one could go through and compile everything, install it, then recompile for good measure to get the desired robust packages.

vb