Index: /branches/eam_branches/ipp-20121219/ippToPsps/jython/setupScratchDb.py
===================================================================
--- /branches/eam_branches/ipp-20121219/ippToPsps/jython/setupScratchDb.py	(revision 35072)
+++ /branches/eam_branches/ipp-20121219/ippToPsps/jython/setupScratchDb.py	(revision 35073)
@@ -14,4 +14,32 @@
 '''
 class SetupScratchDb(IppToPsps):
+
+    ## XXX note that setupScratchDb.py fails if there are no valid
+    ## entries in the config table (or if it is passed a non-existent
+    ## config) This is because ipptopsps.__init__ calls
+    ## checkClientStatus, which calls refreshConfig, which calls
+    ## readConfig, which attempts to select the matching entry from
+    ## the config table.  if the select returns 0 rows, the code
+    ## crashes.  even if it were not to crash (eg, if it returned
+    ## False), checkClientStatus would enter a waiting loop until the
+    ## config existed.
+    
+    ## since a scratchDb is needed to do basically anything, it seems
+    ## like 'setupScratchDb' should not be dependent on the details of
+    ## the configs, etc.
+
+    ## what are the options?  I think this is a problem of poor
+    ## factoring / dependencies, esp for 'config.py'.  there are two
+    ## (or more) aspects to 'config.py' : management of the dynamic
+    ## 'config' information (basically, the regions / dvodbs of
+    ## interest) and manipulation of the static / system configuration
+    ## information.
+
+    ## these concepts should probably be split and the more general
+    ## part (settings & logger) placed in one module and the rest elsewhere.
+
+    ## as for setupScratchDb.py, it looks like it only uses the static
+    ## settings info and the logger functions, so it should not be a
+    ## child of the ipptopsps class
 
     '''