Hi there, Just a note about the modification I just finished on the transformation service on both trunk and 5.1-branch. I refactored the API to use Nuxeo Core API blobs instead of the SerializableInputStream from JBoss remoting which appeared to be extremely bad in term of memory consumption while dealing with large binaries. As well, I hooked up Apache POI for fulltext extraction while indexing documents for speed sake.
Pay attention if you got plugins implementation. You might have to
change a bit your implementation. It's rather easy. though Feel free to
ask questions on how to fix up this if you got any doubts. It might be
the case on client code invoking transformation. Note, exceptionally I
did NOT keep BBB on this because I didn't want anybody to still use the
old API which was crashing the JVM with OutOfMemoryError in cas of too
large binaries to transform.
As well, check your docModifier contributions. The way the doc modifier
is still implemented was taking advantage of a bug in the transformer
plugins options forward. If you don't want to be in trouble for sure
just use the same name for the transformer and the underlying plugin
defined in the chain if you are un such a simple case. Same here, do not
hesitate to ask any questions if you got some.
The good news is that Nuxeo5 now supports injection of really large
blobs such as 60Mo Word documents on 5.1 branch as well :)
Cheers,
J.
--
Julien Anguenot | Nuxeo R&D (Paris, France)
Open Source ECM - http://www.nuxeo.com
Nuxeo 5 : http://www.nuxeo.org
Mobile: +33 (0) 6 72 57 57 66
signature.asc
Description: OpenPGP digital signature
_______________________________________________ ECM mailing list [email protected] http://lists.nuxeo.com/mailman/listinfo/ecm
