Hi Tom,
_p_jar is not set because the blob object is new and has never been
stored yet. This might be because a new blob object has been created
instead of an existing one being updated.
I can only guess, but I believe collective.indexing pushes the
indexing action into the second half of the two phase commit phase
where zope did not commit the changes yet. It could help if you let
solr index stuff asynchronous, or maybe there is a way to change the
indexing behaviour of collective.indexing. (I did not check the docs,
maybe it is).
Best regards,
Patrick
On Thu, Sep 13, 2012 at 8:30 AM, Tom Gross <[email protected]> wrote:
> Hi all
>
> I do indexing of my Plone data with Solr and enhanced collective.solr with
> Tika-support for extracting and indexing binary objects [1]. I use the
> streaming-handler [2]) for feeding the binary data into Solr. In my setup
> Solr and Plone are NOT on the same machine.
> If I upload a file everything works ok but if I change the binary content of
> an existing file the blob is not committed and the '/tmp/...'-path is
> streamed, which obviously does not exist on the remote Solr machine
> (blobstorage is mounted via CIFS). The _p_jar-Attribute of the blob-Object
> in plone.app.blob.field never seems to be set.
>
> As a workaround a commit (a savepoint is for some strange reasons not
> sufficient) in Archetypes.BaseObject._processForm does the trick.
>
> Anyone knows a better way for achieving this? Anything I missed?
>
> Thanks
> :Tom
>
> [1]:
> https://github.com/tomgross/collective.solr/blob/master/src/collective/solr/indexer.py#L98
> [2]: http://wiki.apache.org/solr/ContentStream
>
> _______________________________________________
> Product-Developers mailing list
> [email protected]
> https://lists.plone.org/mailman/listinfo/plone-product-developers
_______________________________________________
Product-Developers mailing list
[email protected]
https://lists.plone.org/mailman/listinfo/plone-product-developers
