@@ 173,7 173,7 @@ def migrate_add_diffstart_diffend(engine
     from tshistory import util
     print(f'add columns `diffstart` and `diffend` to {namespace}.revision')
 
-    def addattributes(cn, tablename):
+    def migrated(cn, tablename):
         sql = (
             f"select exists (select 1 "
             f" from information_schema.columns "

          
@@ 182,11 182,10 @@ def migrate_add_diffstart_diffend(engine
             f" column_name='diffstart'"
             f")"
         )
+        migrated = cn.execute(sql).scalar()
+        return migrated
 
-        migrated = cn.execute(sql).scalar()
-        if migrated:
-            return False
-
+    def addattributes(cn, tablename):
         cn.execute(
             f'alter table "{namespace}.revision"."{tablename}" '
             f'add column diffstart timestamptz'

          
@@ 267,7 266,8 @@ def migrate_add_diffstart_diffend(engine
             )
 
         if delete:
-            print(f'{pid}: revs to delete:', ','.join(x['idate'].isoformat() for x in delete))
+            print(f'{pid}: revs to delete:', ','.join(x['idate'].isoformat()
+                                                      for x in delete))
             sql = (
                 f'delete from "{namespace}.revision"."{tablename}" '
                 f'where id = %(csid)s'

          
@@ 276,11 276,20 @@ def migrate_add_diffstart_diffend(engine
 
     # main
     tsh = tshclass(namespace)
-    names = [
-        name
-        for name, kind in tsh.list_series(engine).items()
-        if kind == 'primary'
-    ]
+    with engine.begin() as cn:
+        cn.cache = {'series_tablename': {}}
+        allnames = {
+            name: tsh._series_to_tablename(cn, name)
+            for name in tsh.list_series(engine).keys()
+        }
+        names = [
+            name
+            for name in allnames
+            if allnames[name] is not None
+            and not migrated(cn, allnames[name])
+        ]
+
+    print(f'{len(names)} series to migrate.')
     cpus = 1 if sys.platform == 'win32' else int(multiprocessing.cpu_count() / 2)
     chunked = listchunks(names, int(cpus))
 

          
@@ 290,18 299,11 @@ def migrate_add_diffstart_diffend(engine
         pid = os.getpid()
         engine = create_engine(url)
         for name in names:
-            print(f'{pid}: migrating `{name}`')
-
             with engine.begin() as cn:
                 cn.cache = {'series_tablename': {}}
                 tablename = tsh._series_to_tablename(cn, name)
-
-                if tablename is None: # not a primary
-                    continue
-
-                if not addattributes(cn, tablename):
-                    print(f'{pid} ... already migrated, skipping')
-                    continue
+                print(f'{pid}: migrating `{name}`')
+                addattributes(cn, tablename)
                 populatedata(pid, cn, name, tablename)
                 finalizeattributes(cn, tablename)
 

          

@@ 1090,24 1090,25 @@ def diffs(cn, tsh, name, tablename, from
         items.reverse()
         # non-appends will cause costly recomputations
         # we might want to find a middle ground
-        _cache.clear()
+        if len(_cache) > 100:
+            for ci in list(_cache.keys())[:20]:
+                _cache.pop(ci)
         out = _cache[top] = patch(
             prev,
             sto._chunks_to_ts(items)
         )
         return out
 
-    def buildseries(parent):
-        top = parent
+    def buildseries(top):
         items = []
 
+        parent = top
         while parent in chunks:
             grandpa, chunk = chunks[parent]
             items.append(chunk)  # bytes
             if grandpa in _cache:
                 # found a known parent: let's patch and remember
-                cached = _cache.pop(grandpa)
-                return patched(top, cached, items)
+                return patched(top, _cache[grandpa], items)
             parent = grandpa
 
         # initial revision or full new series