It turns out I hardcoded the output path that qt's tarball extracts.
But that path is versioned (4.8.5 for example).
As I've already merged x-updates on my own system, my qt version was
different (4.8.4 vs 4.8.5). Made the path-guessing more flexible, so
now it should work with any 4.8.*
And name the desktop file "eagle.desktop", not "Eagle.desktop". The user
facing application name is still "Eagle"; it has nothing to do with the
name of the desktop file.
Upstream insists on using private qt headers.
We do not want nixpkgs' qt to export those.
So I provided a small hack to take them directly from qt's source tarball.
I made sure everything uses the normal system qt and headers, except for the
1 .so file (qt_hack) that needs these private headers.
Because of this, there is barely any increate in size or buildtime.
dramatically speeds up my boot time because it was the last
service (for me) that depended on udev-settle.service
udev-settle isn't needed for modern system initialization but some
oldschool services (mdadm/lvm/cryptsetup) depend on it so they can
just enumerate devices instead of having to react to changes
dynamically. In NixOS these things are usually already taken care of
during stage 1 (early ramdisk) if you use them.
Running 1 test suites...
Test suite hoogle-test: RUNNING...
hoogle-test: datadir/testdata.txt: openFile: does not exist (No such file or directory)
Test suite hoogle-test: FAIL
Test suite logged to: dist/test/hoogle-4.2.19-hoogle-test.log
0 of 1 test suites (0 of 1 test cases) passed.
This commit also fixes an issue where pkgconfig was only added as a
dependency when gtk support was enabled. This made ./configure unable
to find other libraries (libtiff, libxml2, gnutls, and others).
