@sifmelcara, thanks for your PR! By analyzing the history of the files in this pull request, we identified @changlinli, @ehmry and @lverns to be potential reviewers.

change ${qt5.qmake}/bin/qmake" to $NIX_QT5_TMP/bin/qmake", since this is the correct path to find qmake binary.

Thanks for catching this. Referring directly to the store path for qmake was never correct.

removes nativeBuildInputs = [ qt5.qmake ];, which was added by @ttuegel . This change makes rstudio built again. Although I have no idea why removing qt5.qmake from nativeBuildInputs will fix the problem and I am not very sure if this is the correct fix (maybe this breaks cross compiling or something?).

Including the qt5.qmake setup hook is required for qmake to find Qt during the build. I would be very surprised if the build succeeds without it. However, some builds only use qmake to find certain paths and not to do the actual build; possibly that is happening here. If it works, then it works.

However, some builds only use qmake to find certain paths and not to do the actual build; possibly that is happening here.

The strange thing is that rstudio's cmake file only want QT_QMAKE_EXECUTABLE to be set to some value and don't care if there is actually a qmake executable in the given path.
However, since it builds fine and its rstudio executable runs correctly, I think it can be kept as is.

Edit: I mean, the second change in this commit is not necessary to make rstudio build and work correctly.

Thank god, it's running again and it's a reasonably current release. Just tested the build with your changes as well and encountered no problems.

I think this PR is ready to be merged, since rstudio had successfully been built on travis and also @ciil reports this PR solves the problem.