[ckan-dev] Urgent: org changes clarification

[Sean Hammond]

> from
> 
> https://raw.github.com/okfn/ckan/c49c3d7a6e6a635ceecd796b55a679f519242830/doc/organizations_and_groups.rst
> 
> 
> * **Anyone** can see a list of all an organization's _public_ datasets
> * **Anyone** can see a list of all a group's datasets (groups can't
> have private datasets, they're all public)
> 
> This is contradictory/broken logic
> 
> options
> 
> a) any dataset can be in a group but viewers of the group only see
> datasets they can see
> b) any private datasets are never seen as groups datasets even though
> they may be in the group
> c) if a dataset becomes private it is removed from ALL groups it is in
> (REGARDLESS of any rights they have to ANY of those groups)
> 
> 
> I vote for (b) what do people think?

I see what you mean. (a) is simpler than (b), because if a dataset
belongs to some groups and then it gets made private it simply still
belongs to those groups, but people won't see it unless they're a member
of its organization. Then if it gets made public again, it just
continues belonging to the same groups but people can see it now. Would
be the same for tags, from the user's point of view.

Whereas with (b) when someone makes a dataset private you have to remove
it from any groups it's in, then if they make it public again, do you
add it back to the groups or not?

> Also on the bigger issue can we agree that datasets organization
> owners can do things like remove them from groups regardless of group
> ownerships and do we make this more explicit?

Hmm. I think only group members should be able to add things to groups,
although organization members would be able to 'hide' datasets from
groups by making them private.

I'm not sure whether organization members should be allowed to remove a
dataset from a group. On the one hand they are supposed to be able to
edit the datasets in their organization so they would expect to be able
to edit the groups. On the other hand, if they are not a member of a
group they are not supposed to be able to add and remove datasets from
the group. I'm leaning towards they cannot add or remove a dataset from
a group unless they're a member of that group.

But there's potential trouble when someone could create a group with a
stupid name and add a lot of datasets to it, then the stupid name
appears under Groups on the dataset's page and the members of the
dataset's organization can't get rid of it.

Anyway, I think the plan is to implement organizations as described in
the spec but to leave groups as they are currently (and not try to make
them match the spec) and to remove groups from the frontend and replace
them with organizations. So CKAN 2.0 will have only organizations and
not groups, although in the backend the groups feature would still be
there and could be finished off at some point. That's what we agreed
right?