Object caching

[Florian Müller <fm...@opentext.com>]

Hi,

I have just re-introduced caching into the OpenCMIS client API. While caching objects by id makes perfect sense, I'm not so sure about caching by path. Objects could have been deleted, recreated, moved, etc. and the path actually points now to a completely different object. I assume that could be a problem for some applications.

Should we cache by path at all? Should the path cache have an expiration date to minimize the impact? Any opinions?


- Florian

RE: Object caching

["Klevenz, Stephan" <st...@sap.com>]

Hi Florian,

I prefer to get cache support for both id and path. It is always an issue that cache runs out of sync and at least the delete case is also an issue for the cache by id. 

In my opinion it is not necessary to handle the path cache different from the id cache. 

Applications have to be aware of cache behavior and to prevent conflicts they have to avoid concurrent writes to the objects or clear and update the cache before write. If a conflict cannot be avoided then an application has to handle the error case.

Regards,
Stephan

-----Original Message-----
From: Florian Müller [mailto:fmueller@opentext.com] 
Sent: Dienstag, 9. März 2010 13:54
To: chemistry-dev@incubator.apache.org
Subject: Object caching

Hi,

I have just re-introduced caching into the OpenCMIS client API. While caching objects by id makes perfect sense, I'm not so sure about caching by path. Objects could have been deleted, recreated, moved, etc. and the path actually points now to a completely different object. I assume that could be a problem for some applications.

Should we cache by path at all? Should the path cache have an expiration date to minimize the impact? Any opinions?


- Florian