|
Re: Deleting stored records [message #654645 is a reply to message #654599] |
Wed, 16 February 2011 14:51 |
Eclipse User |
|
|
|
Originally posted by: juergen.schumacher.attensity.com
Hi,
Am 16.02.2011, 14:15 Uhr, schrieb Andrej Rosenheinrich <andrej.rosenh
> Or would the obvious solution be to add a pipelet at the end of the =
> workflow that will do this job?
Yes I think so. At least I don't have a better idea currently.
Regards,
J=C3=BCrgen
|
|
|
|
Re: Deleting stored records [message #656235 is a reply to message #655448] |
Thu, 24 February 2011 16:04 |
Eclipse User |
|
|
|
Originally posted by: juergen.schumacher.attensity.com
Am 21.02.2011, 09:16 Uhr, schrieb Andrej Rosenheinrich =
<andrej.rosenheinrich@unister-gmbh.de>:
> Thanks for your answer, J=C3=BCrgen, but I have to ask one more questi=
on on =
> this topic. In the BPEL workflow I have the ids of the records. How ca=
n =
> i get the ids from the attachments I need to delete them in the binary=
=
> store? In my understanding these are not the record ids, right?
> Another question would be if deleting a record (removeRecord(Id id))ha=
s =
> an effect on the binary store or if it only deletes the record from th=
e =
> record store?
In a pipelet you should be able to remove attachments by using
blackboard.removeAttachment(Id, String attachmentName). This
should delete the file on disk, too. The RecordStorage does not
know anything about the BinaryStorage, so removeRecord(Id) on the =
RecordStorage
should have no effect on the BinaryStorage. There should probably be a
removeRecord(Id) method on the blackboard that coordinates these things.=
Regards,
J=C3=BCrgen.
|
|
|
|
|
Re: Deleting stored records [message #658480 is a reply to message #657600] |
Tue, 08 March 2011 16:39 |
Eclipse User |
|
|
|
Originally posted by: juergen.schumacher.attensity.com
Hi,
Am 03.03.2011, 13:49 Uhr, schrieb Andrej Rosenheinrich =
<andrej.rosenheinrich@unister-gmbh.de>:
> My question would be what is happening after the last pipelet in the =
> BPEL workflow is processed. And why the hell is the Recordfilter invok=
ed =
> after the workflow is finished? Or, what would be really bad, are =
> RecordIds not unique so that deleting one record might affect records =
in =
> another workerthread?
Sorry for the late answer, I somehow did not see this message ...
I think the behaviour is correct. Your pipelet just should not return th=
e =
IDs of
records it has deleted, because the WorkflowProcessor must write =
informatin about the returned
IDs into the output BPEL variable of the invocation. The workflow is not=
=
finished after the last
invoke but only after the <reply> element that selects the variable to =
return as the result of
the complete workflow. This cannot be done anymore if the record does no=
t =
exist anymore on the
blackboard and in the storages, of course. So it should work if you just=
=
return a new empty ID
array at the end of your pipelet (assumed that it deleted all input =
records). It would be nice
if you could test this, I'm not completely sure, if the processor can wo=
rk =
with an empty result
array. However, if not it should be quite easy to fix.
Cheers,
J=C3=BCrgen.
|
|
|
|
Powered by
FUDForum. Page generated in 0.04202 seconds