these days gcc have an update to version 6.4.0, and a new system profile is introduced which recommend a re-emerge @world for my system, however when I decide to emerge -qaeDuNv @world, it will failed with spidermonkey and the resume option does not work Now I'm stuck with a half rebuilt system and unable to move forward .. Reproducible: Always
Created attachment 507818 [details] emerge --info
Created attachment 507820 [details] build.log
I tried to figure out a method to bypass spidermonkey-24.2.0-r4 dependency, as dev-lang/spidermonkey-38.2.1_rc0 can be smoothly built on my system, so long as gjs do not depend on the version 24.2.0(with dev-libs/gjs-1.48.6, it depends on spidermonkey-38.2.1_rc0), I can have my system updated now!
(In reply to rtgiskard from comment #3) > I tried to figure out a method to bypass spidermonkey-24.2.0-r4 dependency, > as dev-lang/spidermonkey-38.2.1_rc0 can be smoothly built on my system, so > long as gjs do not depend on the version 24.2.0(with dev-libs/gjs-1.48.6, it > depends on spidermonkey-38.2.1_rc0), I can have my system updated now! unfortunately gnome-shell-3.22.3-r2 depends on gjs-1.46.0. With gjs-1.48.x it fails at buils with a missing gjs-***** package. And if gnome-shell cannot be rebuild it segfaults at gdm startup in libgtk-3. so there is a serious problem updating a mostly stable gnome to gcc-6.4.0 and the new profile. ervin
Created attachment 508244 [details, diff] spidermonkey-24.2.0-icu-60.patch Patch by me using lots of evil reinterpret_cast's to convince it to compile with dev-libs/icu-60.1 using sys-devel/gcc-7.2.0
(In reply to Mark Wright from comment #5) > Created attachment 508244 [details, diff] [details, diff] > spidermonkey-24.2.0-icu-60.patch > > Patch by me using lots of evil reinterpret_cast's to convince it to compile > with dev-libs/icu-60.1 using sys-devel/gcc-7.2.0 thanks a lot! But I'm now having updated to gcc-7.2 , glibc-2.26-r3, sys-libs/readline-7.0_p3, dev-libs/icu-60.1, dev-lang/spidermonkey-38.2.1_rc0, gnome-base/gnome-3.24.0 .. I finally make it! 😱
(In reply to Ervin Peters from comment #4) > (In reply to rtgiskard from comment #3) > > I tried to figure out a method to bypass spidermonkey-24.2.0-r4 dependency, > > as dev-lang/spidermonkey-38.2.1_rc0 can be smoothly built on my system, so > > long as gjs do not depend on the version 24.2.0(with dev-libs/gjs-1.48.6, it > > depends on spidermonkey-38.2.1_rc0), I can have my system updated now! > > unfortunately gnome-shell-3.22.3-r2 depends on gjs-1.46.0. With gjs-1.48.x > it fails at buils with a missing gjs-***** package. > > And if gnome-shell cannot be rebuild it segfaults at gdm startup in libgtk-3. > > so there is a serious problem updating a mostly stable gnome to gcc-6.4.0 > and the new profile. > > ervin thank you so much! I build dev-libs/gjs-1.48.7 and update to gnome-3.24, till now, I finally get my desktop back, everything seems to be ok... except for gnome-3.24 can not restart gnome-shell without hurting the running applications when using wayland, and it seems it does not read .xprofile as well ..
This feels remotely related to bug 621532 about webkit icu incompatibility. Perhaps we can learn something from solutions proposed there.
Does anyone know whether ICU 59 causes this issue here as well, or whether it is ICU 60 only? I assume the former, in which case this bug here should be blocking bug 618636 aka. icu-59. Someonec with edit permission please add that. From the description of that tracking bug: > UChar typedef in C++ was previously unsigned short int by default > (but optionally char16_t). It is now always char16_t. Some reverse > dependencies need to be changed to use char16_t. Details are described in: > http://site.icu-project.org/download/59#TOC-ICU4C-char16_t This might be worth keeping in mind. Relevant pointers from spidermonkey upstream: https://bugzilla.mozilla.org/show_bug.cgi?id=1353650 https://bugzilla.mozilla.org/attachment.cgi?id=8859673&action=diff Haven't tried backporting that yet.
(In reply to Mark Wright from comment #5) > Created attachment 508244 [details, diff] [details, diff] > spidermonkey-24.2.0-icu-60.patch > > Patch by me using lots of evil reinterpret_cast's to convince it to compile > with dev-libs/icu-60.1 using sys-devel/gcc-7.2.0 Can we use this patch in tree?
