Hallo! I'm attaching ebuild for ffnet 0.5. ffnet is a python package for feed-forward neural networks. It's very fast (partially written in fortran) and have some very nice features like automatic normalization of data or genetic algorithm based optimization. ffnet depends on numpy, scipy (which are present in portage) and also networkx which is currentlly bug #120794 Greetings Marek
Created attachment 103957 [details] ffnet-0.5.ebuild
*** Bug 158037 has been marked as a duplicate of this bug. ***
Created attachment 104372 [details] ffnet-0.5.ebuild Modified ebuild added: previous one limited possible fortran compilers to g77 only. Now user can set up system variable F2PY_FC with proper f2py fortran compiler name, otherwise gnu compilers will be used. I don't know if this is good solution. Maybe fortran.eclass should be used here in some way...?
Created attachment 104373 [details] ffnet-0.5.ebuild Modified ebuild added: previous one limited possible fortran compilers to g77 only. Now user can set up system variable F2PY_FC with proper f2py fortran compiler name, otherwise gnu compilers will be used. I don't know if this is good solution. Maybe fortran.eclass should be used here in some way...?
Created attachment 114229 [details] Ebuild for ffnet 0.6 version New version 0.6 of ffnet is released. Slight change in ebuild is made to include documentation distributed with ffnet sources.
New version 0.6.1 of ffnet is released - it's enough to rename ffnet-0.6.ebuild to ffnet-0.6.1.ebuild in order to install it.
Created attachment 208491 [details] ebuild for ffnet-0.6.2 New release of ffnet. Now all dependencies of ffnet are in portage tree (numpy, scipy, networkx).
hallo! is there a way to speed up inclusion of fnet to portage tree? greetings m.
I cleaned it up and imported in the science overlay. Not so sure how useful it is.
thanks! ebuild works for me however there is a typo in HOMEPAGE: there is: http://ffnet.soureceforge.net should be: http://ffnet.sourceforge.net br marek
In the science overlay. Closing.