On Tue Nov 01, 2022 at 09:55:08AM +0100, Antoine Jacoutot wrote:
> On Mon, Oct 31, 2022 at 07:16:49PM +0100, Rafael Sadowski wrote:
> > On Mon Oct 31, 2022 at 02:36:32PM +0000, Stuart Henderson wrote:
> > > On 2022/10/30 17:30, Rafael Sadowski wrote:
> > > > I add a conflict between gsignond and signond becouse both install
> > > > share/dbus-1/services/com.google.code.AccountsSSO.SingleSignOn.service
> > >
> > > This is going to make things difficult for anything depending on them..
> > >
> >
> > Fixed by removing com.google.code.AccountsSSO.SingleSignOn.service from
> > gsignond because it ships with com.google.code.AccountsSSO.gSingleSignOn.service.
> > ^
> > Now signond comes with com.google.code.AccountsSSO.SingleSignOn.service and gsignond
> > with com.google.code.AccountsSSO.gSingleSignOn.service.
>
> Hi Rafael.
>
> Are we sure we want to go this way?
After your question I would say NO. It seems that no other Linux
distributions ship both. Most ship signond but not gsignond.
https://repology.org/project/signond/versions
https://repology.org/project/gsignond/versions
I'd want to do it that way, too. The libs can both coexist.
>
> It seems they both provide the same DBus interface and service so they should
> conflict. Because what you really depend on is the interface/service.
> gSingleSignOn.service and SingleSignOn.service do *not* provide the same
> interfaces (see dbus/interfaces/ in gsignond), so if you remove
> com.google.code.AccountsSSO.SingleSignOn.service from it, it will not export
> these interfaces anymore.
>
> For dependant ports dependency can be handled with using each where needed.
> (just an example, not tested)
> e.g. RUN_DEPENDS += gsignond-*|signond-*:net/accounts-sso/gsignond
>
> Did you ask upstream about the reason of this conflict maybe?
> Although they don't seem very active.
>
> --
> Antoine
>
No comments:
Post a Comment