Examples: Report-Algebra: Ensure all logs are opend in dtype=numpy mode so...

Examples: Report-Algebra: Ensure all logs are opend in dtype=numpy mode so current merge/add/etc. operations work.

darshan-util/pydarshan/darshan/experimental/aggregators/merge.py

@@ -31,7 +31,7 @@ def merge(self, other, reduce_first=False):
         nr.provenance_reports[self.filename] = None
         nr.provenance_reports[other.filename] = None
 
-        nr.provenance_log.append(("add", self, other, datetime.datetime.now()))
+        nr.provenance_graph.append(("add", self, other, datetime.datetime.now()))
 
 
     # update metadata helper

darshan-util/pydarshan/requirements_dev.txt

@@ -12,6 +12,7 @@ cffi
 numpy
 pandas
 matplotlib
+jupyterlab
 
 pytest==3.8.2
 pytest-runner==4.2

darshan-util/pydarshan/tests/test_report_copy.py

+## Check if copy.copy is sufficient (ok on this level.. but!)
+#id(r1.records['POSIX']) == id(combined.records['POSIX'])
+## False means: The dictionary holding POSIX records are not the same (this does not extend to referenced values (here log-records)!)
+## Expect false
+#
+#
+## Check if copy.copy is sufficient (potential conflict for this level?)
+## The question has to be, if we assume people to modify a log record
+## I'd tend to say no, and document that with a warning somewhere
+#id(r1.records['POSIX'][0]) == id(combined.records['POSIX'][0])
+## Expect false