small fix for for the ebuild to add libgssapi the mit version don't provide some functions.
Created attachment 88655 [details, diff] mit-krb5-1.4.3-r1.ebuild-libgssapi.patch
(In reply to comment #1) > Created an attachment (id=88655) [edit] > mit-krb5-1.4.3-r1.ebuild-libgssapi.patch > the reason for this patch ist a dependency tree: libgssapi --> mit-krb5 --> librpcsecgss --> nfs-utils
I think the dependency for this in mit-krb5 should be removed, since they don't have much with each other to do. This libgssapi is a simple "switch" library, which is currently (AFAIK) only used by nfs-utils and its libraries.
Can you please explain me which problem are you trying to solve? We do not have libgssapi/librpcsecgss in the tree.
(In reply to comment #4) > Can you please explain me which problem are you trying to solve? We do not have > libgssapi/librpcsecgss in the tree. > i want to add full nfs4 support to gentoo this requires that the system has a kerberos 5 server running to make this possible librpcsecgss is a need dependency. as the mit krb 1.4 dose not provide all the needed functions mit krb schould depend om libgssapi or we could wait until mit krb 1.5 witch also provides the needed functions for librpcsecgss heimdal 0.7.x has a libgssapi implementation that already provides all needet functions mit krb only provides a spezial libgssapi optimized for krb5 the better way would be extend mits version with the needet functions. or let mit krb compile against libgssapi package??? adding libgssapi as a dependency to mit-krb5 solves the problem in a ugly way but it works(tm) wfg Mario
I think the right way to do this (provided that Heimdal really does make libgssapi unnecessary -- I haven't seen that myself, but then again, I haven't tried either), would be to make librpcsecgss depend upon a virtual, which resolves to either libgssapi or kth-krb (which would also block each other).
(In reply to comment #6) > I think the right way to do this (provided that Heimdal really does make > libgssapi unnecessary -- I haven't seen that myself, but then again, I haven't > tried either), would be to make librpcsecgss depend upon a virtual, which > resolves to either libgssapi or kth-krb (which would also block each other). > I realy think you are right. ok lets add virtuals to heimdal and libgssapi.
This bug is invalid.
Closing with the correct resolution.