opensc does not need openct, please remove the requirement from ebuild. openct will cause problems to pcsc-lite too. Reproducible: Always Steps to Reproduce: 1. emerge opensc -p [ebuild N ] dev-libs/openct-0.5.0 [ebuild N ] dev-libs/opensc-0.9.6
should openct be an optional dependency pulled in by a user flag? like this? RDEPEND=.... pcsc-lite? ( sys-apps/pcsc-lite ) !pcsc-lite? ( openct? ( >=dev-libs/openct-0.5.0 ) )"
step, any ideas with regard to comment #1?
if it lets you emerge only the opensc, then this is OK.
!pcsc-lite? ( openct? ( >=dev-libs/openct-0.5.0 ) )" is wrong I think, openct does not conflict pcsc-lite. opensc is doing well with both openct and pcsc-lite installed.
opensc provides a number of generic pkcs#11 utilities that could (in theory) be useful to a lot of different smartcard/token implementations. these include, but are not limited to openct, pcsc-lite, opencryptoki and maybe athena-scs (pcsc-ase) or the aladdin sdk. so I think the user should decide if opensc is to be merged with openct or pcsc-lite support or simply merged as-is and only used to maintain an opencryptoki/whatever token that provides a module that can be loaded by pkcs11-tool.
Created attachment 96810 [details, diff] opensc-0.11.0-r1.ebuild.diff I agree. I've added a new local use flag opensc so people can choose if they want such dependency. In order to be backward compatible opensc use flag may be set by default, but I am not sure that it should. It is better that user will set the correct underline drivers. I've also removed much of the dependencies that were incorrect for the modular opensc distribution. And introduce currect ones. There is a problem with sparc and alpha dependency, since pcsc-lite is not on ~ARCH.
(In reply to comment #6) > Created an attachment (id=96810) [edit] > opensc-0.11.0-r1.ebuild.diff > > I agree. > I've added a new local use flag opensc so people can choose if they want such > dependency. > In order to be backward compatible opensc use flag may be set by default, but I > am not sure that it should. It is better that user will set the correct > underline drivers. Good. Thanks for the patch. Is it possible to force the compile process towards the underlying drivers rather than relying on dependencies? > I've also removed much of the dependencies that were incorrect for the modular > opensc distribution. And introduce currect ones. > > There is a problem with sparc and alpha dependency, since pcsc-lite is not on > ~ARCH. > Can you create a testing bug for the sparc and alpha teams and make it block this bug?
(In reply to comment #7) > Good. Thanks for the patch. Is it possible to force the compile process towards > the underlying drivers rather than relying on dependencies? No. The configure uses pkgconfig in order to do the detection. > > There is a problem with sparc and alpha dependency, since pcsc-lite is not on > > ~ARCH. > > > Can you create a testing bug for the sparc and alpha teams and make it block > this bug? The problem is on current ebuilds too... So this is not new to this change. Maybe first commit this fix? We can mask out pcsc-lite USE for alpha and sparc, then ask them to check it out.
sounds good - done.