[ckan-dev] API Search does not return deleted packages

[David Read]

On 20 January 2012 09:42, Philipp Lämmel
<philipp.laemmel at fokus.fraunhofer.de> wrote:
> Hey David,
>
> Our use case is somehow similar to an additional draft state. But in more
> detail, our use case is the following:
>
> Our platform separates two types of users : readers and so called package
> owners. Package owners are allowed to add or edit datasets. Additionally,
> package owners have the possibility to 'hide' a  (faulty) dataset and
> publish it later again. This is achieved by setting the state of this
> particular dataset to 'deleted' / 'active'.

Thanks very much for explaining your use case.

We use deleted as a more permanent change. Rather like a file system,
no-one in reality uses the wastebasket as a temporary storage area,
although technically it can be used for that.

I think using the permissions model would be more appropriate to stop
people seeing these particular datasets temporarily.

> We have to different API calls for readers and package owners: readers use
> an API call without a key (that's why they don't see deleted packages) and
> data owners used a call with an API key (therefore they saw their own
> 'deleted' packages, too).
>
> On CKAN 1.5.1 this won't work anymore, since package owners don't see
> deleted packages and can't republish them.
>
> A note on an additional draft state: This would be an elegant solution to
> this use case. But I do think, that you have to reinsert a distinction
> between a call with or without an API key.

There is still a distinction - the user is identified, and what is
listed is affected by permissions. For example, she will probably have
the 'admin' permission for datasets she created. This means that she
can remove the 'read' permission for other people and still see it in
her list. This sounds like it suits your use case?

If you want all the 'package_owners' to have access to all the
'hidden' packages too, then you could create a 'package_owners' role
and give it edit/admin permissions for all the datasets.

I note that we don't have any API calls to change the permissions -
currently just the web API. That is a serious omission which I'll look
at now. So if you wanted to 'hide' a dataset using permissions then
the only current way is with the Web UI until this is implemented.
There's a chance we can get this into 1.5.2 which is about 2/3 weeks
away. http://trac.ckan.org/ticket/1688

What I think would also be useful is if we extended the package_list
API call to take a 'state' parameter, so that deleted (or 'draft' when
we have this) datasets could be exposed in the API.
http://trac.ckan.org/ticket/1689

David
>
> Best,
> Philipp
>
>
> On 01/18/2012 05:52 PM, David Read wrote:
>>
>> Philipp,
>>
>> Yes this changed because it was confusing lots of people. We decided
>> that deleted packages should be hived off into a 'trash can' area for
>> administrators, rather than in the main list, search results etc..
>> That's the way it's viewed in the front-end. Admittedly the trash can
>> is not available (yet) in the API - no-one's asked for it yet but
>> maybe this is what you're after.
>>
>> The whole concept of deleted packages is something we've gradually
>> developed, as it's difficult to get right.
>>
>> Do tell us what you're using hidden packages for. Here are my guesses:
>>
>> * a sort of 'draft' state? We envision using the state field for this,
>> but work needs to be done to achieve this - perhaps you can help spec
>> it out.
>>
>> * Or is the intention to keep these datasets private to particular
>> people? Maybe try removing the 'read' permission for the average user?
>>
>> David
>>
>> On 18 January 2012 16:36, Philipp Lämmel
>> <philipp.laemmel at fokus.fraunhofer.de>  wrote:
>>>
>>> Hey all,
>>>
>>> I've updated my CKAN to 1.5.1 and noticed that there is no difference
>>> anymore in the output of the call
>>>
>>> http://<ckan-instance>/api/rest/dataset
>>>
>>> for package owners (with API-Key) or anonymous users.
>>>
>>> In CKAN 1.4.2 this call showed all deleted packages for package owners,
>>> too.
>>> This was quite handy for us, because we used this for hiding datasets.
>>>
>>> So my question is (rather short and simple): Why did you drop this
>>> distinction?
>>>
>>> Thanks and best regards,
>>> Philipp
>>>
>>>
>>>
>>> _______________________________________________
>>> ckan-dev mailing list
>>> ckan-dev at lists.okfn.org
>>> http://lists.okfn.org/mailman/listinfo/ckan-dev
>>
>>
>> _______________________________________________
>> ckan-dev mailing list
>> ckan-dev at lists.okfn.org
>> http://lists.okfn.org/mailman/listinfo/ckan-dev
>
>
>
> _______________________________________________
> ckan-dev mailing list
> ckan-dev at lists.okfn.org
> http://lists.okfn.org/mailman/listinfo/ckan-dev