pquery --restrict-revdep-pkgs wrongly matches some deps

[remi]

# pquery --restrict-revdep-pkgs dev-cpp/gtkmm:2 --print-revdep dev-cpp/gtkmm | cut -f 1,6 -d " "
dev-cpp/gtkglextmm-1.0.1 =dev-cpp/gtkmm-2.2*
dev-cpp/libglademm-2.0.1 =dev-cpp/gtkmm-2.2*
dev-db/mysql-gui-tools-5.0_p12-r2 dev-cpp/gtkmm:2.4  <<<< wrong !
games-board/freedoko-0.7.3 dev-cpp/gtkmm
media-sound/alsa-patch-bay-1.0.0-r1 =dev-cpp/gtkmm-2.2*
media-sound/alsa-patch-bay-1.0.0-r2 =dev-cpp/gtkmm-2.2*
net-p2p/lince-0.99.4 dev-cpp/gtkmm
x11-misc/fluxbg-0.7 =dev-cpp/gtkmm-2.2*
x11-misc/periodic-calendar-2.1 >=dev-cpp/gtkmm-2.2.0

[ferringb]

dev-db/mysql-gui-tools-5.0_p12-r2 dev-cpp/gtkmm:2.4 <<<< wrong ! Correct, actually, via =dev-cpp/gtkmm-2.2.12 ; recall that --restrict-revdep-pkgs does a look of the specified atom (dev-cpp/gtkmm:2 in this case), looking for anything that *can* depend on that version- 'dev-cpp/gtkmm' can.

Closing, but suspect you'll want the behaviour tweaked, so feel free to reopen with suggestions please.

[remi]

Trac works again, yay! :) So here's a real use case for the record :

# pquery --raw --restrict-revdep-pkgs dev-cpp/gtkmm:1.2 --print-revdep dev-cpp/gtkmm | cut -f 1,6 -d " " | sed -e "s/ /\t\t/"
dev-db/mysql-gui-tools-5.0_p12-r2               dev-cpp/gtkmm:2.4   <<<<< wrong
games-board/freedoko-0.7.3              dev-cpp/gtkmm
media-sound/amsynth-1.0.0               =dev-cpp/gtkmm-1.2*
net-misc/cisco-aironet-client-utils-2.1         <=dev-cpp/gtkmm-2.0
net-p2p/btg-0.9.7               dev-cpp/gtkmm:2.4                   <<<<< wrong
net-p2p/lince-0.99.4            dev-cpp/gtkmm:2.4                   <<<<< wrong
net-print/omni-0.9.1            >=dev-cpp/gtkmm-1.2.5
net-print/omni-0.9.2            >=dev-cpp/gtkmm-1.2.5

Notice how I asked for the 1.2 slot :)

And for the first example I gave, ":2" should have meant "just slot 2" and not "slot 2 and slot 2.4".

[ferringb]

pquery implementation of that feature daftly was knocking it down to a versioned_atom; same issue existed for repo/use atoms beyond slot atoms.

Fixed via 3519 of my branch, will hit integration shortly.