Please add USE "udev" and move sys-power/upower and sys-fs/udisks from pkg_postinst to RDEPEND. x11-misc/spacefm, x11-libs/libfm (pcmanfm), media-sound/qmmp and many other packages have udev USE for udisks. Using einfo instead of USE flag seems illogical for me.
Or even PDEPEND...
No You miss the point of D,RDEPEND. These variables are there for optional runtime/build time dependencies that affect how the package is built. Razor-qt does not depend on upower of udisks. In other words, if you build razor-qt with "udev" or "-udev" the final binary would be the same because the source code does not change whether you have these packages installed. So the point of adding a use flag for an optional runtime dependency is moot. So this bug is invalid. All the ebuilds you mentioned are using udev in src_configure phase to build optional udev support in the package. This is not the case with razor-qt though
Disagree. DEPEND is related to src_configure, but RDEPEND and PDEPEND are not. These two variables are not about how you build binaries, but how you use them, how they run and work. Your argument conflicts with PDEPEND existence and RDEPEND definition: >These variables are there for optional runtime/build time dependencies that affect how the package is built. http://devmanual.gentoo.org/general-concepts/dependencies/index.html : "The RDEPEND ebuild variable should specify any dependencies which are required at runtime. This includes libraries (when dynamically linked), any data packages and (for interpreted languages) the relevant interpreter." A lot packages have DEPEND="1 2 3"\ RDEPEND="$DEPEND 4 5 6"\ PDEPEND="7 8 9" If you think it's wrong, why you do not open bugs and create TRACKER for them? Without udisks and upower users are not able to do a lot important for DE tasks, which are expected to work out of the box. Also what about python packages? Their "binaries" are not compiled, behavior often changes according to packages found at startup.
(In reply to comment #3) > Your argument conflicts with PDEPEND existence and RDEPEND definition: > > http://devmanual.gentoo.org/general-concepts/dependencies/index.html : > "The RDEPEND ebuild variable should specify any dependencies which are required > at runtime. This includes libraries (when dynamically linked), any data > packages and (for interpreted languages) the relevant interpreter." Please read the linked page carefully, and note the usage of the words "required" and "must". In our case these dependencies are completely optional: razorqt will function without crashes if these additional packages are not installed. It will just offer some additional (but very useful) functionality in case these extra packages are present on the system. But if we introduce a useflag, then this change, and any subsequent toggling of the useflag by the user, will force the package to be rebuilt, which is completely unnecessary. This is why it is Gentoo policy not to introduce useflags for such optional packages, and only add dependencies that need to be present at build time or that affect the basic functionality of a package (i.e. the package doesn't work at all without the runtime dependency). If you wish to see a change in this policy, then please bring the discussion to the gentoo-dev mailinglist. This bug report is not the place for that.
Seems this bug should be marked fixed, because both deps are now in RDEPEND for razorqt-base/razorqt-power, razorqt-base/razorqt-autosuspend and razorqt-base/razorqt-panel.
