[DAIS-WG] Interop Document and Next Steps.
Mario Antonioletti
mario at epcc.ed.ac.uk
Fri Nov 13 10:47:44 CST 2009
Hi,
We appear to have gone through the public comment phase and
received comments from two people - one from Isao and the other from
Elias. See:
http://www.ggf.org/gf/docs/comment.php?id=308
Elias' comments require addressing, mainly:-
- In case of GetSQLResponseItem, maybe it should be clarified that the
items are supposed to return in order, i.e. first rowsets, then
update counts, etc
Point 24 on page 32 of the interop document proposes an update to
the proposed WS-DAIR recommendation to specify an order in which
these elements SHOULD be returned.
- In case of GetTuples, the AccessMode will give an error if already
the required rows have been accessed. It was not clear how this
condition would be produced on the test.
Test 16 invokes SQLRowset::GetTuples to get 1 tuple at position 1
(the second tuple available, since the first tuple is at position
0). Test 17 then attempts to get the first tuple at position 0:
If the SQLRowset supports Random AccessMode then Test 17 expects
the first tuple to be returned.
If the SQLRowset supports Forward AccessMode then Test 17 expects
an InvalidPositionFault to be generated.
- Finally, without knowing the details of the AMGA implementation, is
consistency expected with the OGSA-DAI one in the case that an
SQLResponse is destroyed and consequently all the child SQLRowsets
are destroyed as well?
The WS-DAI specifications do not specify what happens when
associated derived data resources are destroyed
(ChildSensitiveToParent and ParentSensitiveToParent give an
indication of whether changes in the dervied data resources
are written through to the parent or vice versa but this
becomes irrelevant if the parent or child resource is not
there anymore). See 5.3.3 in the WS-DAI spec:
When an internally managed data resource is destroyed all of its
associated data MAY be destroyed, and the resources used MAY be
reclaimed. The internal behavior depends upon the data access
service implementation but from the consumers perspective the data
resource is no longer available following this operation.
Hence there is no reason for AMGA to be consistent in behaviour to
what OGSA-DAI does.
Are people happy with these answers?
In addition Yan Zhang has noted that in Figures 1, 2 and 3 the return
type SQLRowSet should in fact be a DatasetData. I would like to add
this as a proposed change to the WS-DAIR document. We only had it for
Figure 1 - point 19 on p31 for changes to the WS-DAIR document. It
should also apply to Figures 2 and 3.
Are people happy with these answers?
We then need to check that Elias is happy with our response and we put
the document in for publication. Apparently, according to GFD.152
section 4.5.4, an external reviewer then needs to be appointed by the
area directors to assess whether the documents should go on to become
OGF recommendations - the process is not quite what I thought it was
but there you have it,
Mario
+-----------------------------------------------------------------------+
|Mario Antonioletti:EPCC,JCMB,The King's Buildings,Edinburgh EH9 3JZ. |
|Tel:0131 650 5141|mario at epcc.ed.ac.uk|http://www.epcc.ed.ac.uk/~mario/ |
+-----------------------------------------------------------------------+
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
More information about the dais-wg
mailing list