jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Craig Schaefer (JIRA)" <j...@apache.org>
Subject [jira] Updated: (OCM-4) requestObjectCache not cleared in ObjectContentManager.getObjectIterator()
Date Fri, 19 Jun 2009 15:54:07 GMT

     [ https://issues.apache.org/jira/browse/OCM-4?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Craig Schaefer updated OCM-4:
-----------------------------

    Attachment: OCM-4.patch

Two main changes:

1. The getObjectIterator(...):Iterator methods now return an ObjectIterator with their own
local object converter and therefore their own ObjectCache instance.  Therefore the global
cache belonging to the ObjecContentMangerInstance is not influenced as the iterator is used
at some unknown time in the future.

2. refresh() now always clears teh ObjectCache belonging to the instance of ObjectContentManagerImpl
as otherwise there is no fallback possible.

Note
1. This patch also contains the related fixes from OCM-34.
2. Possible improvement would be to requestCache.clear() in finally blocks in all occurances,
espeically as a get could fail and leave the cache dirty.

> requestObjectCache not cleared in ObjectContentManager.getObjectIterator()
> --------------------------------------------------------------------------
>
>                 Key: OCM-4
>                 URL: https://issues.apache.org/jira/browse/OCM-4
>             Project: Jackrabbit OCM
>          Issue Type: Bug
>            Reporter: David Franke
>         Attachments: OCM-4.patch
>
>
> I found this bug using an old snapshot version (1.5-ASF666297), but I had a look at the
sources in the trunk and the implementation has not changed, so I assume it still exists.
> This bug only applies to the two methods Iterator getObjectIterator(...). Collection
getObjects(...) works.
> The method 
> public Object getObject(Session session, String path)
> puts objects in the RequestObjectCache. If an object is already cached it is retrieved
from the cache.
> if (requestObjectCache.isCached(path))
>  {
>         return requestObjectCache.getObject(path);
> }
> In ObjectContentManager.getObjects() this cache is cleared after retrieving the object.
> In ObjectContentManager.getObjectIterator() this is never cleared.
> This makes problems if I make several searches with the same ObjectContentManager while
the data in the repository changes.
> Example:
> UserSearch searches for all user with a given name. 
> An XPath is created like this:  //(*, xy:user)[@name='givenName']
> It always keeps the same instance of ObjectContenManager to call the search until the
session expires:
> userIterator = ocm.getObjectIterator(query,Query.XPATH);
> ...
> In the meantime a user may change his name. When I now search for this new name, the
user is found, but the old user object will be returned (with old name). There is no way to
clear the cache manually. ocm.refresh(false) does not help in this case.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message