Update of /cvsroot/cweb/bigdata-rdf/src/java/com/bigdata/rdf
In directory sc8-pr-cvs4.sourceforge.net:/tmp/cvs-serv16041/src/java/com/bigdata/rdf
Modified Files:
TripleStore.java
Log Message:
Updated the IDataService interface and now have client talking over JERI to a data service instance for various btree operations, range count, and range query.
Index: TripleStore.java
===================================================================
RCS file: /cvsroot/cweb/bigdata-rdf/src/java/com/bigdata/rdf/TripleStore.java,v
retrieving revision 1.33
retrieving revision 1.34
diff -C2 -d -r1.33 -r1.34
*** TripleStore.java 19 Apr 2007 19:13:46 -0000 1.33
--- TripleStore.java 20 Apr 2007 16:36:36 -0000 1.34
***************
*** 153,156 ****
--- 153,163 ----
* restart-safe manner even if we "forget" the term-id mapping.
*
+ * @todo Provide a wrapper class for {@link IEntryIterator} that visits
+ * statements, terms, and ids (as appropriate). This will help to isolate
+ * the rdf store from changes to the manner in which keys and values are
+ * encoded in the indices, including whether the term ids can be decoded
+ * from the statement keys, and also will help to create a
+ * (de-)serialization divide for the values in the indices.
+ *
* @todo modify the term identifier assignment mechanism to be compatible with
* the scale-out index partitions (32-bit unique within index partition
***************
*** 186,193 ****
*
* @todo support metadata about the statement, e.g., whether or not it is an
! * inference. consider that we may need to move the triple/quad ids into
* the value in the statement indices since some key compression schemes
! * are not reversable (we depend on reversable keys to extract the term
! * ids for a statement).
*
* @todo Try a variant in which we have metadata linking statements and terms
--- 193,200 ----
*
* @todo support metadata about the statement, e.g., whether or not it is an
! * inference. consider that we may need to move the triple/quad ids into
* the value in the statement indices since some key compression schemes
! * are not reversable (we depend on reversable keys to extract the term
! * ids for a statement).
*
* @todo Try a variant in which we have metadata linking statements and terms
***************
*** 233,240 ****
*
* @todo provide option for closing aspects of the entire store vs just a single
! * context in a quad store. For example, in an open web and internet scale
! * kb it is unlikely that you would want to have all harvested ontologies
! * closed against all the data. however, that might make more sense in a
! * more controlled setting.
*
* @author <a href="mailto:tho...@us...">Bryan Thompson</a>
--- 240,255 ----
*
* @todo provide option for closing aspects of the entire store vs just a single
! * context in a quad store vs just a "document" before it is loaded into a
! * triple store (but with the term identifiers of the triple store). For
! * example, in an open web and internet scale kb it is unlikely that you
! * would want to have all harvested ontologies closed against all the
! * data. however, that might make more sense in a more controlled setting.
! *
! * @todo provide a mechanism to make document loading robust to client failure.
! * When loads are unisolated, a client failure can result in the
! * statements being loaded into only a subset of the statement indices.
! * robust load would require a means for undo or redo of failed loads. a
! * loaded based on map/reduce would naturally provide a robust mechanism
! * using a redo model.
*
* @author <a href="mailto:tho...@us...">Bryan Thompson</a>
|
