testing node type management

[Julian Reschke <ju...@gmx.de>]

Julian Reschke wrote:
> Hi,
> 
> according to <https://issues.apache.org/jira/browse/JCR-1608>, 
> Jackrabbit (trunk) currently only supports creating workspaces, but not 
> deleting them.
> 
> The spec allows UnsupportedOperationException on both create and delete.
> 
> So what should the TCK do? Discover the repository feature, attempt 
> creation, then deletion, but not report an error if the latter fails? 
> That may be tricky because it would cause an ever-growing number of 
> workspaces being generated every time the TCK is run...
> 
> BR, Julian
> ...

And, related to my previous question:

<https://issues.apache.org/jira/browse/JCR-1591> indicates that only 
registration, but not removal is supported.

Looking at the Javadoc for the unregister methods, I don't see this case 
considered. So, should we consider this a missing feature of the RI?

And, again, how should the TCK deal with it?

BR, Julian

Re: testing node type management

[Julian Reschke <ju...@gmx.de>]

Marcel Reutegger wrote:
> Hi,
> 
> On Tue, Jul 7, 2009 at 15:41, Julian Reschke<ju...@gmx.de> wrote:
>> <https://issues.apache.org/jira/browse/JCR-1591> indicates that only
>> registration, but not removal is supported.
>>
>> Looking at the Javadoc for the unregister methods, I don't see this case
>> considered. So, should we consider this a missing feature of the RI?
> 
> to me, it seems the spec is not clear enough about unregistering node
> types. e.g. what happens if there still is content using that node
> type? AFAICS the spec only talks about collisions when updating a node

I think we can't require repositories to support removal of a node type 
that is in use. We may want to specify the exact exception to be thrown.

> ...

BR, Julian

Re: testing node type management

[Marcel Reutegger <ma...@day.com>]

Hi,

On Tue, Jul 7, 2009 at 15:41, Julian Reschke<ju...@gmx.de> wrote:
> <https://issues.apache.org/jira/browse/JCR-1591> indicates that only
> registration, but not removal is supported.
>
> Looking at the Javadoc for the unregister methods, I don't see this case
> considered. So, should we consider this a missing feature of the RI?

to me, it seems the spec is not clear enough about unregistering node
types. e.g. what happens if there still is content using that node
type? AFAICS the spec only talks about collisions when updating a node
type. and for that case there's also a repository descriptor:
NODE_TYPE_MANAGEMENT_UPDATE_IN_USE_SUPORTED. at least in the spec. I
just saw that it is missing in the Repository interface :-/
(https://jsr-283.dev.java.net/issues/show_bug.cgi?id=793). I think for
unregistration there should be something similar.

regards
 marcel

> And, again, how should the TCK deal with it?
>
> BR, Julian
>
>
>
>
>