Lina,

Note that there my be multiple path values, but only a single path value for a 
given object since paths are stored as a set. The guarantee is provided by set 
semantics. We do not particularly care about duplicates across multiple objects.

- Alex.

> On Apr 19, 2017, at 1:09 PM, Na Li <[email protected]> wrote:
> 
> Sasha,
> 
> For now, the path entry in DB is owned by each Authz object. 
> 
> The DB change Kalyan made does not guarantee that with the same path value, 
> there is only one instance of such path entry in DB in 
> https://issues.apache.org/jira/browse/SENTRY-1629 
> <https://issues.apache.org/jira/browse/SENTRY-1629> (He will address this 
> later). Therefore, there could be two or more entries that have the same path 
> value. So Hao has to change the equals() to find the path instance owned by 
> authz Object.
> 
> 
> 
> Thanks,
> 
> Lina
> 
> On Wed, Apr 19, 2017 at 2:05 PM, Hao Hao <[email protected] 
> <mailto:[email protected]>> wrote:
> Hi Alex,
> 
> I changed MPath class to include pathID to have correct MPath equality
> comparison. Since the original equals() definition only compares 'path'
> which for different MPath object it may have the same value.
> 
> For pathID (primary key) assignment I change it to be auto-increment by
> specifying value-strategy to be "increment ".
> 
> Please let me know if you think any of the change is not reasonable. Thanks!
> 
> Best,
> Hao
> 
> On Wed, Apr 19, 2017 at 11:37 AM, Alexander Kolbasov <[email protected] 
> <mailto:[email protected]>>
> wrote:
> 
> > Hao,
> >
> > Can you clarify the changes you propose for MPath class and related
> > package.jdo changes for SENTRY-1587?
> >
> > You suggest changing the identity type from "database" to "application",
> > but the pathId is not initialized in constructor and not assigned anywhere.
> > What is your intention here?
> >
> > Also you changed the equals() method to include ID. What kind of equality
> > do you require - equality by path value or equality by identity? What is
> > your thinking here?
> >
> > This is in regards to the code review https://reviews.apache.org/r/58412 
> > <https://reviews.apache.org/r/58412>.
> >
> > Thanks,
> >
> > - Alex
> >
> 

Reply via email to