EDIT

Andreas Müller wrote:

The fix works so far that search result in TaxEditor allow only opening in TaxEditor now. However, an orphaned taxon as described in #5616 still shows the wrong symbol (black for accepted).
Is there also another algorithm that determines orphaned in the search results?

The editor uses the uuidAndTitleCache object and for a misapplied name the orphaned flag is set to false, so the used image is the one for a not orphaned taxon -> black square.

Also orphaned synonyms are not found in TaxEditor taxon bulk editor anymore.

There is a problem with partial and proparte. If there is no values for these attibutes, an exception is thrown because a boolean attribute can not be null.
If I replace NULL with 0 then all synonyms and taxa are found. We need to add this to the update script, if it is not included already.

Katja, can you have a look on this? If necessary create a new ticket.

Katja Luther wrote:

Andreas Müller wrote:

The fix works so far that search result in TaxEditor allow only opening in TaxEditor now. However, an orphaned taxon as described in #5616 still shows the wrong symbol (black for accepted).
Is there also another algorithm that determines orphaned in the search results?

The editor uses the uuidAndTitleCache object and for a misapplied name the orphaned flag is set to false, so the used image is the one for a not orphaned taxon -> black square.

But it is NOT a misapplied name as far as I remember but only has concept relations. Therefore the "Open in" works correctly now.

Katja Luther wrote:

Andreas Müller wrote:

Also orphaned synonyms are not found in TaxEditor taxon bulk editor anymore.

There is a problem with partial and proparte. If there is no values for these attibutes, an exception is thrown because a boolean attribute can not be null.
If I replace NULL with 0 then all synonyms and taxa are found. We need to add this to the update script, if it is not included already.

This should already be fixed in the update script. However, the test server was updated before this fix. I thought I did update all test server databases manually. If not, please let me know where you found this problem. Definitely it is not on the example database I mentioned above (cdm_rem_conf_am). I tested with

SELECT *
FROM TaxonBase
WHERE DTYPE = 'Synonym' AND (proParte IS NULL OR partial IS NULL)

Also, if this is the problem, the exception should be thrown (and shown to the user) and not be swallowed by the system.

Andreas Müller wrote:

Katja Luther wrote:

Andreas Müller wrote:

The fix works so far that search result in TaxEditor allow only opening in TaxEditor now. However, an orphaned taxon as described in #5616 still shows the wrong symbol (black for accepted).
Is there also another algorithm that determines orphaned in the search results?

The editor uses the uuidAndTitleCache object and for a misapplied name the orphaned flag is set to false, so the used image is the one for a not orphaned taxon -> black square.

But it is NOT a misapplied name as far as I remember but only has concept relations. Therefore the "Open in" works correctly now.

The open in works correctly now. The wrong symbol for taxa with conceptrelationships/misapplied names is because the orphaned flag is set by the hql query not by the isOrphaned() method.

Katja Luther wrote:

The open in works correctly now. The wrong symbol for taxa with conceptrelationships/misapplied names is because the orphaned flag is set by the hql query not by the isOrphaned() method.

This is what I expected. So the critical question is here, how we can keep the java based definition of "isOrphaned" in sync with any hql based methods. Or should we simply leave a warning in the according javadoc that some hql also needs to be adapted.
Also exact tests may help for both methods.

Andreas Müller wrote:

Katja Luther wrote:

The open in works correctly now. The wrong symbol for taxa with conceptrelationships/misapplied names is because the orphaned flag is set by the hql query not by the isOrphaned() method.

This is what I expected. So the critical question is here, how we can keep the java based definition of "isOrphaned" in sync with any hql based methods. Or should we simply leave a warning in the according javadoc that some hql also needs to be adapted.
Also exact tests may help for both methods.

I found this http://stackoverflow.com/questions/25382840/loading-transient-field-of-jpa-entity,
I tried to add a transient attribute isOrphaned and add a method which should be called after loading the entity (with PostLoad annotation).
But the first test did not work. But I think this could be a solution to keep it in sync.

Katja Luther wrote:

Andreas Müller wrote:

Katja Luther wrote:

The open in works correctly now. The wrong symbol for taxa with conceptrelationships/misapplied names is because the orphaned flag is set by the hql query not by the isOrphaned() method.

This is what I expected. So the critical question is here, how we can keep the java based definition of "isOrphaned" in sync with any hql based methods. Or should we simply leave a warning in the according javadoc that some hql also needs to be adapted.
Also exact tests may help for both methods.

I found this http://stackoverflow.com/questions/25382840/loading-transient-field-of-jpa-entity,
I tried to add a transient attribute isOrphaned and add a method which should be called after loading the entity (with PostLoad annotation).
But the first test did not work. But I think this could be a solution to keep it in sync.

This might be a solution but not very nice as it runs the check whenever a taxon is loaded so the performance is going down. We may think about caching this data. Once cache updating works better this might be a solution to keep it more or less in sync and also to retrieve the data directly from database which makes some queries faster (e.g. if you don't want to show orphaned taxa at all)

Katja, can you please extract the open issues to a new ticket and close this one?