> review needsfixing
>
> Is this a merge for both PluginManagerService and PackageManagerService?
> If so, could we merge PluginManagerService separately, and then merge
> PackageManagerService after that has landed?
There is already a pending merge for PluginManagerService, this merge depends on it.
> You've added Do.Linux.JoliCloud/Makefile.in to bzr. That's not very
> useful :).
Whoops. Removed.
> In Do.Platform.Linux.JoliCloud/src/Do.Platform/Do.Platform.Linux/Do.Platform.Linux.JoliCloud/PackageManagerService.cs:
>
> I don't much like the magical Daemon property, particularly since you
> need to document it with the comment
> """
> // this call will instantiate the daemon, as well as make sure we also
> got a DBus object
> """
> in Initialize (). Is there any particular reason that Initialise can't
> initialise the daemon explicitly?
I did it this way to avoid the duplication of adding the event, etc. It seems cleaner to me.
> In MaybeStartDaemon (), is it possibly to ask DBus to wait for an owner
> of BusName rather than just sleeping for 5 seconds? I'm not sure if it
> is, but I'd find it cleaner.
I don't think it's possible, this is how Banshee does it, but I'm pretty sure I can safely remove that sleep call if it'd make you feel better.
> In Do.Platform/src/Do.Platform/AbstractPackageManagerService.cs:
> MaybePluginForPackage (string): It might be worthwhile extending the
> plugin API to allow a plugin itself to provide a list of packages that
> it supports, as well as this magic name/description matching.
I had mentioned this in irc, perhaps in the plugin manifest adding an element/attribute of keywords that could be used. This could also be handy in the search.
> What happens if multiple addins support a new package?
We take the first.
> This comment from PluginManagerService.cs seems wrong?
> /// If this class loads, we have a serious plugin because that probably
> means we have no plugin manager.
This is a comment for the Plugin Manager merge, but I'll fix it there
> review needsfixing rvice and PackageManagerS ervice? rvice separately, and then merge ervice after that has landed?
>
> Is this a merge for both PluginManagerSe
> If so, could we merge PluginManagerSe
> PackageManagerS
There is already a pending merge for PluginManagerSe rvice, this merge depends on it.
> You've added Do.Linux. JoliCloud/ Makefile. in to bzr. That's not very
> useful :).
Whoops. Removed.
> In Do.Platform. Linux.JoliCloud /src/Do. Platform/ Do.Platform. Linux/Do. Platform. Linux.JoliCloud /PackageManager Service. cs:
>
> I don't much like the magical Daemon property, particularly since you
> need to document it with the comment
> """
> // this call will instantiate the daemon, as well as make sure we also
> got a DBus object
> """
> in Initialize (). Is there any particular reason that Initialise can't
> initialise the daemon explicitly?
I did it this way to avoid the duplication of adding the event, etc. It seems cleaner to me.
> In MaybeStartDaemon (), is it possibly to ask DBus to wait for an owner
> of BusName rather than just sleeping for 5 seconds? I'm not sure if it
> is, but I'd find it cleaner.
I don't think it's possible, this is how Banshee does it, but I'm pretty sure I can safely remove that sleep call if it'd make you feel better.
> In Do.Platform/ src/Do. Platform/ AbstractPackage ManagerService. cs: ackage (string): It might be worthwhile extending the
> MaybePluginForP
> plugin API to allow a plugin itself to provide a list of packages that
> it supports, as well as this magic name/description matching.
I had mentioned this in irc, perhaps in the plugin manifest adding an element/attribute of keywords that could be used. This could also be handy in the search.
> What happens if multiple addins support a new package?
We take the first.
> This comment from PluginManagerSe rvice.cs seems wrong?
> /// If this class loads, we have a serious plugin because that probably
> means we have no plugin manager.
This is a comment for the Plugin Manager merge, but I'll fix it there