I already mentioned this issue in https://bugs.gentoo.org/show_bug.cgi?id=506006#c6 (and possibly in other bugs): ENABLE_GTK3 takes yes/no instead of 1/0. I did not notice that this change was missed because I used my own ebuild. --- uzbl-9999.ebuild.org 2014-12-19 09:28:07.631513998 +0100 +++ uzbl-9999.ebuild 2014-12-19 09:28:43.076511606 +0100 @@ -126,7 +126,7 @@ src_compile() { [[ ${PV} == 9999 ]] && gtk_var='ENABLE_GTK3' || gtk_var='USE_GTK3' - emake PREFIX="${PREFIX}" ${gtk_var}=$(use gtk3 && echo 1 || echo 0) + emake PREFIX="${PREFIX}" ${gtk_var}=$(use gtk3 && echo yes || echo no) } src_install() {
(In reply to Franz Fellner from comment #0) > + emake PREFIX="${PREFIX}" ${gtk_var}=$(use gtk3 && echo yes || echo Or use the newfangled $(usex gtk3 yes no '' '')
(In reply to Jeroen Roovers from comment #1) > (In reply to Franz Fellner from comment #0) > > + emake PREFIX="${PREFIX}" ${gtk_var}=$(use gtk3 && echo yes || echo > > Or use the newfangled $(usex gtk3 yes no '' '') Probably, yes. But it would still require extra code for the non-live-ebuild, as they expect "1/0". IMHO the best solution would be to simply patch the non-live-sources to use ENABLE_GTK3, make it optional and use yes/no. Should be straight forward. If this package still has a maintainer and he will accept such a patch I will create it. Otherwise I will (have to) use my own ebuild, again.
There is still a maintainer (a proxy-maintainer as a matter of fact). Please attach your patch, I'll apply it.
Created attachment 392106 [details, diff] uzbl-2012.05.14_gtk3_no_autodep.patch --- uzbl-2012.05.14-r1.ebuild.org 2014-12-20 19:46:37.059945813 +0100 +++ uzbl-2012.05.14-r1.ebuild 2014-12-20 20:09:31.079853087 +0100 @@ -97,14 +97,15 @@ } src_prepare() { + if [ ${PV} != 9999 ] + then + epatch "${FILESDIR}/${P}_gtk3_no_autodep.patch" + fi + # remove -ggdb sed -i 's/-ggdb //g' Makefile || die '-ggdb removal sed failed' - # make gtk3 configurable - sed -r 's:^(USE_GTK3) = (.*):\1?=\2:' -i Makefile || - die 'Makefile sed for gtk3 failed' - # specify python version python_fix_shebang bin/uzbl-tabbed || die 'Fix shebang failed' @@ -125,8 +126,7 @@ } src_compile() { - [[ ${PV} == 9999 ]] && gtk_var='ENABLE_GTK3' || gtk_var='USE_GTK3' - emake PREFIX="${PREFIX}" ${gtk_var}=$(use gtk3 && echo 1 || echo 0) + emake PREFIX="${PREFIX}" ENABLE_GTK3=$(usex gtk3 yes no '' '') } src_install() { === I tested both 2012.05.14-r1 and 9999 with gtk3 enabled and disabled. Patch applies only to non-live, USE="-gtk3" emerge --nodeps fails - which is what I expect as I don't have webkit-gtk:2 installed - my laptop simply doesn't have the power to emerge yet another webkit just for fun/testing ;) USE="gtk3" compiles fine.
*** Bug 533148 has been marked as a duplicate of this bug. ***
Created attachment 399848 [details] uzbl-9999.ebuild (with dev-python/six) New version of the ebuild. Changelog: - fix bug 533006 (thanks to Franz Fellner) - remove -j1 in install as upstream fixed 351 - add dev-python/six in RDEPEND, as it fail with the latest release Same ebuild for both 9999 and 2012.05.14-r1.
It took real 132m10.817s to emerge net-libs/webkit-gtk-2.4.8-r200::gentoo but it was all worth it for it took and all of real 0m4.437s to build uzbl-2012.05.14-r2.ebuild. Unless someone has any doubt or reason not to, the ebuild and patch appear good to commit.
uzbl is out of tree now.