I'm in the terminal, you will see it in the log Reproducible: Always
Created attachment 432618 [details] dependancies info
Created attachment 432622 [details] emerge --info
Did you try building ipython with USE="-notebook"? Quote: * Note that circular dependencies can often be avoided by temporarily * disabling USE flags that trigger optional dependencies.
(In reply to Sven E. from comment #3) > Did you try building ipython with USE="-notebook"? > > Quote: > * Note that circular dependencies can often be avoided by temporarily > * disabling USE flags that trigger optional dependencies. yes, but portage just require that I add the use "notebook" again, so there is no way to go.
The circular dependency is not due to sage but interdependency of the various ipython ebuilds. This has been reported before. A number of optional dependencies of ipython should be PDEPEND instead of RDEPEND. I already got it fixed for qtconsole and smp. notebook has to be added to this list, and probably nbconvert- not sure about ipywidgets. And the fix shouldn't just be applied to the greatest and latest since it often affects stable too. A workaround for you is to emerge ipython (alone) without the notebook useflag. Once this is done you can re-add the notebook useflag and emerge sage and it should work.
Actually after re-reading the log and the ebuild, ipywidgets needs to go to PDEPEND as well since it is depending on the notebook useflag.
(In reply to Francois Bissey from comment #5) > The circular dependency is not due to sage but interdependency of the > various ipython ebuilds. > > This has been reported before. A number of optional dependencies of ipython > should be PDEPEND instead of RDEPEND. I already got it fixed for qtconsole > and smp. notebook has to be added to this list, and probably nbconvert- not > sure about ipywidgets. > > And the fix shouldn't just be applied to the greatest and latest since it > often affects stable too. > > A workaround for you is to emerge ipython (alone) without the notebook > useflag. Once this is done you can re-add the notebook useflag and emerge > sage and it should work. thank you very much! if there is a dependency circulation, build one of the dependencies first is a way to break out the loop, I got it! by building the ipython first without use 'notebook', then I have the chance to fetch sage, but several blocks is there and I have to add '~amd64' to some packages, like blas, lapack, and gsl, and then I am asked to add another '~amd64' manually , the block was break! and I'm now building sagemath. indeed, with different consideration, there is no uniform way to handle the dependences, gentoo is there for us to apply our own choice, with or without manual sometimes. Thanks a lot!
sorry to modify the status so early, and so sorry to find that the sci-libs/lapack-reference and iml can not pass the compilation, failed at pre-compile or build time, eselect cblas or blas shows nothing.
Created attachment 433744 [details] sci-libs/lapack-reference-3.6.0-r1 build log
Created attachment 433748 [details] iml build log
First off, this is a different issue so new bug. Second, your first issue is not resolved as far as I can tell. I gave you a work around but there is a clear solution. I just don't have the privileges to fix it (my own fault really, people have tried to get me to submit my quizzes as far back as 2009). This bug has been mis-triaged because of its title, suggesting it is a sage issue when it is a main tree issue. Regarding your new issue, output of eix -I blas eselect blas list
(In reply to Francois Bissey from comment #11) > First off, this is a different issue so new bug. > sorry, I have realize it, post it here is for the case that somebody may want to have a try, just like me, and he may need or provide some information. > Second, your first issue is not resolved as far as I can tell. I gave you a > work around but there is a clear solution. I just don't have the privileges > to fix it (my own fault really, people have tried to get me to submit my > quizzes as far back as 2009). > Indeed, though not fully understand where the problem is, but if the dependency can be solved in a way like this, then the official method to deal with dependencies need to be renovated. The package like ipython should have considered the dependencies more carefully, and it seems that it's not the time for me to go now as I must deal with some more emergent things. > This bug has been mis-triaged because of its title, suggesting it is a sage > issue when it is a main tree issue. > I'll modify the title to a more proper one. and I have no idea on how to deal with this bug now, maybe you can help to modify the status. > Regarding your new issue, output of > eix -I blas > eselect blas list Now I have returned back to gsl-1.14, unable to install sage with the dependencies circulation. before I install eix, with gsl-2.1, there is no item for blas and cblas with eselect. I have decide not to hang on here as it's not the right time for me. Thank you so much!
now it's resolved! first apply this: sci-libs/gsl ~amd64 virtual/blas ~amd64 virtual/cblas ~amd64 virtual/lapack ~amd64 then I'll be reqiured to add: # required by virtual/blas-2.1-r5::science # required by dev-libs/igraph-0.7.1-r1::gentoo # required by sci-mathematics/rw-0.7::gentoo[executable] # required by sci-mathematics/sage-7.1::sage-on-gentoo # required by sci-mathematics/sage-notebook-0.11.7::sage-on-gentoo =sci-libs/blas-reference-20151113-r1 ~amd64 # required by virtual/cblas-2.0-r4::science # required by dev-python/numpy-1.10.4::gentoo[lapack] # required by media-gfx/inkscape-0.91-r1::gentoo # required by @selected # required by @world (argument) =sci-libs/cblas-reference-20151113-r3 ~amd64 # required by virtual/lapack-3.6::science # required by sci-libs/linbox-1.3.2-r2::sage-on-gentoo # required by sci-mathematics/sage-7.1::sage-on-gentoo # required by sci-mathematics/sage-notebook-0.11.7::sage-on-gentoo =sci-libs/lapack-reference-3.6.0-r1 ~amd64 then I can begin building sage, but it fails when building sci-libs/lapack-reference, hang on.. this morning I'm trying to fight for the last time (for recent months) to get it done, eselect blas shows nothing and I think maybe another packages need to be installed, so I began to focus on blas, found this one: https://wiki.gentoo.org/wiki/User_talk:Houseofsuns seems to be the case, but doesn't work,, no blas with eselect still, and soon I found atlas, and tried to build it, but failed with sci-libs/atlas-3.11.38-r1(not knowing why), after find this: https://github.com/gentoo-science/sci/issues/583 I tried sci-libs/atlas-3.10.2, it takes more than 3 hours to build but finally it works, then rebuild numpy for matplotlib, things seems to be fine, at least till now.
