Please for explanation why it is bad head for bug which block this one. This package was anounced as broken in old bug 267776 i will mask and treeclean it in 30 days if noone step up to fix it.
Hi Tomáš, PLEASE do NOT treeclean paraview just because it uses cmake directly. Paraview as well as vtk are very complex scientific packages which use cmake in a fairly complex way (as a matter of fact they are both developed by kitware which also develops cmake). Both packages are used by a lot of people. I have not had time for an in depth look at the cmake-utils.eclass yet but I suspect quite a few of the things paraview/vtk need are not yet covered in the eclass at all - I might be wrong though. Porting paraview and vtk to the cmake.eclass is on my to do list but will be a lot of work and require even more testing for both of which I don't have time right now so this will have to wait until later this year. If you have time to help do some of the porting work or check how the paraview/vtk ebuild will fit in the framework of the eclass I'd certainly appreciate it. If you have any concerns please don't hesitate to get in touch. Thanks, Markus
Sorry for the double post - bugzilla seems confused right now. Markus
(In reply to comment #3) > Sorry for the double post - bugzilla seems confused > right now. > > Markus > No prob :] Well my concern is that eclass should cover everything. Cmake itself is compiled using cmake eclass. So it would be kinda weird if this two would need better treatement. From what i can see it is easy to migrate it really. Just take look on kdelibs, they have same level of complexity, or soprano library, it uses also stupid variable names as vtk. Cheers Tomas
(In reply to comment #4) > No prob :] > > Well my concern is that eclass should cover everything. Cmake itself is > compiled using cmake eclass. So it would be kinda weird if this two would need > better treatement. > > You're probably right D) Anyway, I'll promise I'll look at it as soon as I find a stretch of time where I can dig into the eclass and make sure I don't break anything paraview wise. The next two months will be packed for me due to work/travel but once that's done I should be able to give this a serious go. Thanks for your patience. cheers, Markus
Hi, just a quick word - as discussed with bicatali and scarabeus I am tackling this as a side effect of trying to get the overview application to really install. I have also reworked quite a number of things in the process [Most of those plugins now configure and compile - I just leave you ffmpeg :) ]. I left comments all over the place it is some wild ride. I am expecting to upload a new ebuild in the science overlay later tonight for general consumption, once I am sure it really works at last and I have documented those pesky use flags.
Created attachment 213058 [details] ebuild proted to cmake-utils.eclass I am attaching an "OverView" (and gl2ps) less version of the ebuild I posted in the science overlay. It has also been cleaned a little bit around the edge. The main advantage of this ebuild is it doesn't depend on anything unstable making it a good candidate for stabilization as well. It also take care of bug #273117 .
Dont depend on cmake, its handled by eclass. This is plain wrong: dohtml "${D}/usr/bin/about.html" && rm -f "${D}/usr/bin/about.html" \ || die "Failed to move about.html into doc dir" Also just place cmake-utils as last in inherit order and drop src_compile. New syntax for cmakelists the current one will be deprecated: mycmakeargs+=( "argument" "argument" "argument" )
(In reply to comment #8) > This is plain wrong: > dohtml "${D}/usr/bin/about.html" && rm -f "${D}/usr/bin/about.html" \ > || die "Failed to move about.html into doc dir" Fixing that will require patching cmake files somewhere, it is generated and of course there is a rule to install it somewhere. Or we could just remove it.
Created attachment 213136 [details] OK ported to the newest cmake-utils eclass and with an added patch.
Created attachment 213137 [details, diff] patch for detecting nvidia-cg-toolkit on Gentoo
Created attachment 213161 [details] fix the about.html thing patch for it following OK it was easier to fix the about.html issue than I expected.
Created attachment 213163 [details, diff] patch to avoid installing about.hml in /usr/bin
Thanks, Francois. I hope I can have a look at this over the holidays a bit. Keep up the good work :) cheers, Markus
Ok guys, this is last direct cmake user in main tree, so chop chop :D
(In reply to comment #15) > Ok guys, this is last direct cmake user in main tree, so chop chop :D > I am embarassed :( Hopefully, I'll get this done this week. Thanks for your patience. cheers, Markus
I haven't checked it yet but we probably can as well revbump to version 3.6.2, which came out last week, at the same time. The release notes mentioned a new plugin but I haven't seen a config option for it. Other than that it is a maintenance release so no real changes.
I've just added a paraview-3.6.2 ebuild which has all of these fixes. Thanks to everybody, but in particularly Francois for their great help. Best, Markus
