diff --git a/pdfrw/findobjs.py b/pdfrw/findobjs.py
index f19ebdf..67d33a0 100644
--- a/pdfrw/findobjs.py
+++ b/pdfrw/findobjs.py
@@ -8,7 +8,6 @@
 '''
 
 from .objects import PdfDict, PdfArray, PdfName
-from .pdfwriter import user_fmt
 
 
 def find_objects(source, valid_types=(PdfName.XObject, None),
@@ -81,7 +80,7 @@ def wrap_object(obj, width, margin):
         iw, ih = float(obj.Width), float(obj.Height)
         ch = 1.0 * cw / iw * ih
         height = ch + margin[1] + margin[3]
-        p = tuple(user_fmt(x) for x in (cw, ch, xoffset, yoffset))
+        p = tuple(('%.9f' % x).rstrip('0').rstrip('.') for x in (cw, ch, xoffset, yoffset))
         contents.stream = fmt % p
         resources = PdfDict(XObject=PdfDict(MyImage=obj))
         mbox = PdfArray((0, 0, width, height))
diff --git a/pdfrw/objects/pdfname.py b/pdfrw/objects/pdfname.py
index 1fdf5b5..28a1464 100644
--- a/pdfrw/objects/pdfname.py
+++ b/pdfrw/objects/pdfname.py
@@ -23,6 +23,7 @@ class BasePdfName(str):
     '''
 
     indirect = False
+    encoded = None
 
     whitespace = '\x00 \t\f\r\n'
     delimiters = '()<>{}[]/%'
diff --git a/pdfrw/pdfwriter.py b/pdfrw/pdfwriter.py
index a7d4af4..3c887ba 100755
--- a/pdfrw/pdfwriter.py
+++ b/pdfrw/pdfwriter.py
@@ -137,11 +137,11 @@ def format_obj(obj):
                 elif isinstance(obj, PdfDict):
                     if compress and obj.stream:
                         do_compress([obj])
-                    pairs = sorted((x, y, getattr(x, 'encoded', x))
+                    pairs = sorted((getattr(x, 'encoded', None) or x, y)
                                    for (x, y) in obj.iteritems())
                     myarray = []
-                    for key, value, encoding in pairs:
-                        myarray.append(encoding)
+                    for key, value in pairs:
+                        myarray.append(key)
                         myarray.append(add(value))
                     result = format_array(myarray, '<<%s>>')
                     stream = obj.stream
@@ -155,7 +155,7 @@ def format_obj(obj):
             # We assume that an object with an indirect
             # attribute knows how to represent itself to us.
             if hasattr(obj, 'indirect'):
-                return str(getattr(obj, 'encoded', obj))
+                return str(getattr(obj, 'encoded', None) or obj)
             return user_fmt(obj)
 
     def format_deferred():
diff --git a/pdfrw/toreportlab.py b/pdfrw/toreportlab.py
index 9f77d26..3434fbf 100644
--- a/pdfrw/toreportlab.py
+++ b/pdfrw/toreportlab.py
@@ -108,7 +108,7 @@ def _makearray(rldoc, pdfobj):
 def _makestr(rldoc, pdfobj):
     assert isinstance(pdfobj, (float, int, str)), repr(pdfobj)
     # TODO: Add fix for float like in pdfwriter
-    return str(getattr(pdfobj, 'encoded', pdfobj))
+    return str(getattr(pdfobj, 'encoded', None) or pdfobj)
 
 
 def makerl_recurse(rldoc, pdfobj):
diff --git a/tests/expected.txt b/tests/expected.txt
index 8c6c5f1..b1b7cca 100644
--- a/tests/expected.txt
+++ b/tests/expected.txt
@@ -11,7 +11,7 @@ examples/subset_b1c400de699af29ea3f1983bb26870ab_1-3_5      880a9578197130273ccb
 examples/unspread_d711b74110eefb4e9e6bf1a5bea16bfe          780a9abe26a9de0b5b95ee22c4835e4b
 
 examples/cat_b1c400de699af29ea3f1983bb26870ab_06c86654f9a77e82f9adaa0086fc391c          62bb9b746ff5932d3f1b88942d36a81d
-examples/rotate_707e3e2d17cbe9ec2273414b3b63f333_270_1-4_7-8_10-50_52-56                841c980dfadf2cc47ad86e4649ca69b6
+examples/rotate_707e3e2d17cbe9ec2273414b3b63f333_270_1-4_7-8_10-50_52-56                7633ba56641115050ba098ecbef8d331
 examples/watermark_b1c400de699af29ea3f1983bb26870ab_06c86654f9a77e82f9adaa0086fc391c    fe2330d42b3bfc06212415f295752f0e
 examples/watermark_b1c400de699af29ea3f1983bb26870ab_06c86654f9a77e82f9adaa0086fc391c_-u e43e3ac0afe1cc242549424755dbf612
 
@@ -88,7 +88,7 @@ repaginate/5f265db2736850782aeaba2571a3c749.pdf 2e3046813ce6e40a39bd759a3c8a3c8c
 repaginate/6a42c8c79b807bf164d31071749e07b0.pdf bf00d5e44869ae59eb859860d7d5373f
 repaginate/6f3a4de5c68ba3b5093e9b54b7c4e9f4.pdf 612cdd84eeac797a1c42fc91756b6d9e
 repaginate/7037a992b80b60f0294016037baa9292.pdf dd41b0104f185206b51e7ffe5b07d261
-repaginate/707e3e2d17cbe9ec2273414b3b63f333.pdf 6c65526ab372d72cb185933e3d2584ef
+repaginate/707e3e2d17cbe9ec2273414b3b63f333.pdf df4d756e2230c333f0c58ad354b5b51c
 repaginate/71a751ce2d93a6a5d6ff21735b701fb7.pdf a825f06c934319b93474902fcf300cd2
 repaginate/72eb207b8f882618899aa7a65d3cecda.pdf 0b64f19a8a39fadfa2a3eec3f1a01233
 repaginate/97ba0a239cefa0dc727c2f1be050ec6c.pdf a94fe7183ce8979174b2ac16dcd9b1ea
@@ -127,7 +127,7 @@ simple/5f265db2736850782aeaba2571a3c749.pdf d4d2e93ab22e866c86e32da84421f6f9
 simple/6a42c8c79b807bf164d31071749e07b0.pdf 221fec351c925a43f5f409fe03d90013
 simple/6f3a4de5c68ba3b5093e9b54b7c4e9f4.pdf fe8dd16dd7fef40338140e0610d0cbbf
 simple/7037a992b80b60f0294016037baa9292.pdf 6a2ef24e5f74dd74969ff8cefdfc6a05
-simple/707e3e2d17cbe9ec2273414b3b63f333.pdf 4bdf1e57a96ce42717110b4e55098c1a
+simple/707e3e2d17cbe9ec2273414b3b63f333.pdf fb6a8eb3cdc2fbef125babe8815f3b70
 simple/71a751ce2d93a6a5d6ff21735b701fb7.pdf a825f06c934319b93474902fcf300cd2
 simple/72eb207b8f882618899aa7a65d3cecda.pdf 4ce7ff29531cc417c26389af28dc1c5e
 simple/97ba0a239cefa0dc727c2f1be050ec6c.pdf c24873bab85b8ecc7c5433d8d802bceb
@@ -167,7 +167,7 @@ decompress/5f265db2736850782aeaba2571a3c749.pdf 836abcf6e6e1d39ad96481eb20e9b149
 decompress/6a42c8c79b807bf164d31071749e07b0.pdf 221fec351c925a43f5f409fe03d90013
 decompress/6f3a4de5c68ba3b5093e9b54b7c4e9f4.pdf 226773cac79e1a5fed1379a0501a5df0
 decompress/7037a992b80b60f0294016037baa9292.pdf c9a3602b26d82ae145d9f5822125a158
-decompress/707e3e2d17cbe9ec2273414b3b63f333.pdf f6d960e75480aa4f729059388dcedd71
+decompress/707e3e2d17cbe9ec2273414b3b63f333.pdf 3250a56e14a9855eccd67bb347808d24
 decompress/71a751ce2d93a6a5d6ff21735b701fb7.pdf a825f06c934319b93474902fcf300cd2
 decompress/72eb207b8f882618899aa7a65d3cecda.pdf a4366874fb6db1d9a0c998361ea32b8d
 decompress/97ba0a239cefa0dc727c2f1be050ec6c.pdf c24873bab85b8ecc7c5433d8d802bceb
@@ -208,7 +208,7 @@ compress/5f265db2736850782aeaba2571a3c749.pdf bb4898beac50171de7502f13925af80c
 compress/6a42c8c79b807bf164d31071749e07b0.pdf 221fec351c925a43f5f409fe03d90013
 compress/6f3a4de5c68ba3b5093e9b54b7c4e9f4.pdf 1c3fbae41e7cad7deca13fab93514bc7
 compress/7037a992b80b60f0294016037baa9292.pdf 9182a9765544e4a91404db65a6f951d7
-compress/707e3e2d17cbe9ec2273414b3b63f333.pdf bde552c97872c5a4eeafab3b8b38f703
+compress/707e3e2d17cbe9ec2273414b3b63f333.pdf 0e75dda73bf18d9968499277ab1a367e
 compress/71a751ce2d93a6a5d6ff21735b701fb7.pdf faa7eb31789a3789f65de30a4e58e594
 compress/72eb207b8f882618899aa7a65d3cecda.pdf 0155549fc04357220cc6be541dda7bc1
 compress/97ba0a239cefa0dc727c2f1be050ec6c.pdf 067bfee3b2bd9c250e7c4157ff543a81
diff --git a/tests/update_expected.py b/tests/update_expected.py
new file mode 100755
index 0000000..bed5331
--- /dev/null
+++ b/tests/update_expected.py
@@ -0,0 +1,84 @@
+#! /usr/bin/env python2
+"""
+Put old (good) results in ramdisk/reference,
+then generate new (unknown) test results in ramdisk/tmp_results,
+THEN SWITCH BACK TO KNOWN GOOD SYSTEM, and finally:
+
+run this to update any checksums in expected.txt where both versions
+parse to same PDFs.
+"""
+
+import os
+import hashlib
+from pdfrw import PdfReader, PdfWriter, PdfArray, PdfDict, PdfObject
+
+
+def make_canonical(trailer):
+    ''' Canonicalizes a PDF.  Assumes everything
+        is a Pdf object already.
+    '''
+    visited = set()
+    workitems = list(trailer.values())
+    while workitems:
+        obj = workitems.pop()
+        objid = id(obj)
+        if objid in visited:
+            continue
+        visited.add(objid)
+        obj.indirect = True
+        if isinstance(obj, (PdfArray, PdfDict)):
+            if isinstance(obj, PdfArray):
+                workitems += obj
+            else:
+                workitems += obj.values()
+    return trailer
+
+with open('expected.txt', 'rb') as f:
+    expected = f.read()
+
+def get_digest(fname):
+        with open(fname, 'rb') as f:
+            data = f.read()
+        if data:
+            return hashlib.md5(data).hexdigest()
+
+tmp = '_temp.pdf'
+count = 0
+goodcount = 0
+
+changes = []
+for (srcpath, _, filenames) in os.walk('ramdisk/reference'):
+    for name in filenames:
+        if not name.endswith('.pdf'):
+            continue
+        src = os.path.join(srcpath, name)
+        dst = src.replace('/reference/', '/tmp_results/')
+        if not os.path.exists(dst):
+            continue
+        src_digest = get_digest(src)
+        if not src_digest or src_digest not in expected:
+            continue
+        print src
+        count += 1
+        trailer = make_canonical(PdfReader(src))
+        out = PdfWriter(tmp)
+        out.write(trailer=trailer)
+        match_digest = get_digest(tmp)
+        if not match_digest:
+            continue
+        trailer = make_canonical(PdfReader(dst))
+        out = PdfWriter(tmp)
+        out.write(trailer=trailer)
+        if get_digest(tmp) != match_digest:
+            continue
+        goodcount += 1
+        print "OK"
+        changes.append((src_digest, get_digest(dst)))
+
+print count, goodcount
+
+for stuff in changes:
+    expected = expected.replace(*stuff)
+
+with open('expected.txt', 'wb') as f:
+    f.write(expected)