1
2Revision 0.4.3, released 23-05-2018
3-----------------------------------
4
5- Copyright notice extended to the year 2018
6- Fixed GeneralizedTime.asDateTime to perform milliseconds conversion
7  correctly
8
9Revision 0.4.2, released 23-11-2017
10-----------------------------------
11
12- Fixed explicit tag splitting in chunked encoding mode at
13  OctetString and BitString encoders
14
15Revision 0.4.1, released 23-11-2017
16-----------------------------------
17
18- ANY DEFINED BY clause support implemented
19- Encoders refactored to take either a value (as ASN.1 object)
20  or a Python value plus ASN.1 schema
21- BitString decoder optimised for better performance when running on
22  constructed encoding
23- Constructed types' .getComponentBy*() methods accept the `default`
24  parameter to return instead if schema object is to be returned
25- Constructed types' .getComponentBy*() methods accept the `instantiate`
26  parameter to disable automatic inner component instantiation
27- The ASN.1 types' `__repr__` implementation reworked for better readability
28  at the cost of not being `eval`-compliant
29- Most ASN.1 types' `__str__` magic methods (except for OctetString and
30  character types) reworked to call `.prettyPrint()` rather than
31  `.prettyPrint` calling `__str__` as it was before. The intention is
32  to eventually deprecate `.prettyPrint()` in favor of `str()`.
33  The other related change is that `str()` of enumerations and boolean
34  types will return string label instead of number.
35- Fixed Choice.clear() to fully reset internal state of the object
36- Sphinx documentation rearranged, simplified and reworded
37- The `isValue` singleton is now the only way to indicate ASN.1 schema
38  as opposed to ASN.1 schema instance. The legacy `None` initializer
39  support has been removed.
40- Changed `Null` object initialization behaviour: previous default
41  value (`''`) is not set anymore. Thus `Null()` call produces a
42  ASN.1 schema object, while `Null('')` - value object.
43- Migrated all docs and references from SourceForge
44- Imports PEP8'ed
45- Fixed ASN.1 encoder not to omit empty substrate produced for inner
46  component if the inner component belongs to the simple class (as
47  opposed to constructed class)
48- Fixed CER/DER encoders to respect tagged CHOICE when ordering
49  SET components
50- Fixed ASN.1 types not to interfere with the Pickle protocol
51- Fixed Sequence/SequenceOf types decoding heuristics in schema-less
52  decoding mode
53
54Revision 0.3.7, released 04-10-2017
55-----------------------------------
56
57- Fixed ASN.1 time types pickling/deepcopy'ing
58
59Revision 0.3.6, released 21-09-2017
60-----------------------------------
61
62- End-of-octets encoding optimized at ASN.1 encoders
63- The __getitem__/__setitem__ behavior of Set/Sequence and SetOf/SequenceOf
64  objects aligned with the canonical Mapping and Sequence protocols in part
65- Fixed crash in ASN.1 encoder when encoding an explicitly tagged
66  component of a Sequence
67
68Revision 0.3.5, released 16-09-2017
69-----------------------------------
70
71- Codecs signatures unified and pass the options kwargs through the
72  call chain
73- Explicit tag encoding optimized to avoid unnecessary copying
74- End-of-octets sentinel encoding optimized
75- Refactored ASN.1 codecs properties to silently enforce proper
76  length and chunk size encoding modes
77- Fixed DER encoder to always produce primitive encoding
78- Fixed crash at SequenceOf native decoder
79- Fixed Real.prettyPrint() to fail gracefully on overflow
80- Fixed a couple of crashes when debug mode is enabled
81
82Revision 0.3.4, released 07-09-2017
83-----------------------------------
84
85- Fixed Native encoder to handle SEQUENCE/SET objects without
86  the componentType property
87- Added missing component-less SEQUENCE/SET objects dict duck-typing support
88- Fixed unnecessary duplicate tags detection at NamesType.tagMap
89- Fixed crash at SEQUENCE and SEQUENCE OF CER encoder when running
90  in schemaless mode
91- Fixed Character types instantiation from OctetString type -- double
92  unicode decoding may have scrambled the data
93
94Revision 0.3.3, released 27-08-2017
95-----------------------------------
96
97- Improved ASN.1 types instantiation performance
98- Improved BER/CER/DER decoder performance by not unconditionally casting
99  substrate into str/bytes.
100- Fixed exponential index size growth bug when building ambiguous
101  NamedTypes tree
102- Fixed constructed types decoding failure at BER codec if running
103  in schema-less mode
104- Fixed crash on prettyPrint'ing a SEQUENCE with no defined components
105- Fixed SetOf ordering at CER/DER encoder
106- Fixed crash on conditional binascii module import
107- Fix to TagSet hash value build
108
109Revision 0.3.2, released 04-08-2017
110-----------------------------------
111
112- Fixed SequenceOf/SetOf types initialization syntax to remain
113  backward compatible with pyasn1 0.2.*
114- Rectified thread safety issues by moving lazy, run-time computation
115  into object initializer.
116- Fixed .isValue property to return True for empty SetOf/SequenceOf
117  objects
118- Fixed GeneralizedTime/UTCTime CER/DER codecs to actually get invoked
119- Fixed DER/CER encoders handling optional SEQUENCE/SET fields containing
120  nested SEQUENCE/SET with optional fields.
121- Fixed crash in SequenceOf/SetOf pretty printing and decoding (in some
122  cases)
123- Fixed documentation markup issues.
124
125Revision 0.3.1, released 26-07-2017
126-----------------------------------
127
128- ASN.1 types __init__(), .clone() and .subtype() signatures
129  refactored into keyword arguments to simplify their signatures.
130- ASN.1 types initialization refactored to minimize the use of
131  relatively expensive isNoValue() call
132- Lazily pre-populate list of values of Sequence/Set/Choice types
133- NamedTypes comparison made more efficient
134- More efficient constraints computation and code clean up
135- The __getitem__() implementation of some ASN.1 types & tag object
136  refactored for better performance
137- BER/CER/DER value encoders refactored to produce either tuple of
138  bytes or octet-stream depending on what is more optimal
139- Reduced the frequency of expensive isinstance() calls
140- Tag-related classes optimized, refactored into properties and
141  documented.
142- The NamedValues implementation refactored to mimic Python dict, its use
143  in ASN.1 types refactored into properties and better documented.
144  WARNING: this change introduces a deviation from original API.
145- NamedType family of classes overhauled, optimized and documented.
146- The `componentType` attribute of constructed ASN.1 types turned
147  read-only on instances.
148- Sequence/Set DER/CER/DER decoder optimized to skip the case of
149  reordered components handling when not necessary.
150- Tags and constraints-related getter methods refactored into read-only
151  instance attributes.
152- The .hasValue() method refactored into .isValue property. All ASN.1
153  objects now support them, not just scalars.
154- The Real.{isInfinity, isPlusInfinity, isMinusInfinity} methods
155  refactored into properties and renamed into IsInf, IsPlusInf and isMinusInf
156- The end-of-octets type refactored to ensure it is a singleton. Codecs
157  changed to rely on that for better performance.
158- Codecs lookup made more efficient at BER/CER/DER decoder main loop by
159  assigning `typeId` to every ASN.1 type, not just ambiguous ones.
160- The .getComponent*() methods of constructed ASN.1 types changed
161  to lazily instantiate underlying type rather than return `None`.
162  This should simplify its API as initialization like `X[0][1] = 2` becomes
163  possible.
164  WARNING: this change introduces a deviation from the original API.
165- The .setComponent*() methods of SetOf/SequenceOf types changed not
166  to allow uninitialized "holes" inside the sequences of their components.
167  They now behave similarly to Python lists.
168  WARNING: this change introduces a deviation from the original API.
169- Default and optional components en/decoding of Constructed type
170  refactored towards better efficiency and more control.
171- OctetsString and Any decoder optimized to avoid creating ASN.1
172  objects for chunks of substrate. Instead they now join substrate
173  chunks together and create ASN.1 object from it just once.
174- The GeneralizedTime and UTCTime types now support to/from Python
175  datetime object conversion.
176- Unit tests added for the `compat` sub-package.
177- Fixed BitString named bits initialization bug.
178- Fixed non-functional tag cache (when running Python 2) at DER decoder.
179- Fixed chunked encoding restriction on DER encoder.
180- Fixed SET components ordering at DER encoder.
181- Fixed BIT STRING & OCTET STRING encoding to be always non-chunked (e.g.
182  primitive) at DER encoder
183- Fixed `compat.integer.from_bytes()` behaviour on empty input.
184
185Revision 0.2.3, released 25-02-2017
186-----------------------------------
187
188- Improved SEQUENCE/SET/CHOICE decoding performance by maintaining a single shared
189  NamedType object for all instances of SEQUENCE/SET object.
190- Improved INTEGER encoding/decoding by switching to Python's built-in
191  integer serialisation functions.
192- Improved BitString performance by rebasing it onto Python int type and leveraging
193  fast Integer serialisation functions.
194- BitString type usability improved in many ways: for example bitshifting and
195  numeric operation on BitString is now possible.
196- Minor ObjectIdentifier type performance optimization.
197- ASN.1 character types refactored to keep unicode contents internally
198  (rather than serialised octet stream) and duck-type it directly.
199- ASN.1 OctetString initialized from a Python object performs bytes()
200  on it when running on Python 3 (used to do str() which is probably
201  less logical).
202- Missing support for NoValue.__sizeof__ added.
203- Added checks to make sure SEQUENCE/SET components being assigned
204  match the prototypes.
205- Setter methods for constructed types consistently accept matchTags
206  and matchConstraints flags to control the strictness of inner
207  components compatibility verification. Previously, these checks
208  were tied to verifyConstraints flag, now they are all independent.
209- General documentation improvements here and there.
210- Fix to __reversed__() magic to make it returning an iterator.
211- Test suite simplified and unified.
212- The __all__ variable added to most of the Python modules.
213- The "test" directory renamed into "tests" not to collide with
214  the "test" module.
215
216Revision 0.2.2, released 07-02-2017
217-----------------------------------
218
219- FIX TO A SECURITY WEAKNESS: define length only decoders could have successfully
220  processed indefinite length serialisation.
221- FIX TO A SECURITY WEAKNESS: canonical decoders (CER/DER) may have successfully
222  consumed non-canonical variations of (otherwise valid) serialisation.
223- Broken Enumerated subtyping fixed.
224
225Revision 0.2.1, released 05-02-2017
226-----------------------------------
227
228- FIX TO A SECURITY WEAKNESS: BER decoder improperly cached long tags.
229- New "native" codec implemented to transform pyasn1 types to Python built-in types and back.
230- Switched to new-style classes.
231- Sphinx documentation added.
232- BitString improvements:
233
234  * simple string of binary digits is now supported as initializer
235  * default str() yields string of binary digits (used to yield str(tuple())
236  * binValue and hexValue initializers added
237  * .asNumbers(), .asOctets() and asInteger() representation added
238
239- Components of constructed ASN.1 types can now be populated with
240  uninitialized ASN.1 objects by assigning either noValue sentinel or
241  setupComponent() function return in addition to now-legacy None sentinel.
242  This should improve code readability.
243- NoValue class improved to become a singleton and catch more kinds
244  of access to it.
245- Compatibility wrappers str2octs() and oct2strs() fixed to run over
246  iso-8859-1 encoding.
247- Integer changed to emit Real instance if division produces a float.
248- True division operation now supported by Integer type.
249- The __contains__(), __reverse__() methods implemented for container types
250- Iterator protocol support implemented for all container types.
251  Warning, warning, warning: this change may potentially affect backward
252  compatibility when:
253
254  * user class overrides __getitem__() without overriding __iter__()
255  * when user code iterates over SEQUENCE object to get its components (now keys will be yielded)
256
257- Almost complete Python list and dict protocols added to SequenceOf/SetOf and
258  Sequence/Set respectively
259- Fix to divmod operation implementation in Integer type.
260- Fix to IntegerDecoder's precomputed value map on Python 3.
261- Fix to base ASN.1 types to run in "unicode_literals" mode.
262- Fix to composite constraints "+" operands ordering (AbstractConstraintSet.__radd__)
263- Fix to constraints merge in .subtype() -- on merge existing constraints are
264  expanded to accommodate new constraints, not the other way round. When existing
265  constraints are wrapped in ConstraintsIntersection composite, additional
266  constraints being added on subtyping effectively further narrow the set of
267  allowed values, which aligns well with the notion of subtyping.
268- Fix to NamedTypes methods to handle .getTagMap() returning None
269- Fix to Set/Sequence.setDefaultComponents() to return self
270- Copyright notice added to non-trivial source code files.
271- Author's email changed, copyright extended to 2017
272
273Revision 0.1.9, released 28-09-2015
274-----------------------------------
275
276- Wheel distribution format now supported.
277- Extensions added to text files, CVS attic flushed.
278- Fix to make uninitialized pyasn1 objects failing properly on hash().
279- Fix to ObjectIdentifier initialization from unicode string.
280- Fix to CER/DER Boolean decoder - fail on non single-octet payload.
281
282Revision 0.1.8, released 22-06-2015
283-----------------------------------
284
285- ObjectIdentifier codec fixed to work properly with arc 0 and arc 2 values.
286- Explicit limit on ObjectIdentifier arc value size removed.
287- Unicode initializer support added to OctetString type and derivatives.
288- New prettyPrintType() abstract method implemented to base pyasn1 types
289  to facilitate encoding errors analysis.
290- The __str__() method implemented to Tag, TagSet and TagMap classes to
291  ease encoding errors troubleshooting.
292  easing encoding errors
293- Fix to SEQUENCE and SET types to give them their private componentTypes
294  collection (which is a NamedTypes object) so that they won't collide in
295  a MT execution environment.
296- Missing T61String,ISO646String character types and ObjectDescriptor useful
297  type added.
298- Distribute is gone, switched to setuptools completely.
299- Missing NamedValues.__repr__() added.
300- The base.NoValue() class, that indicates uninitialized ASN.1 object,
301  made public.
302- The base.NoValue() class instances now support __repr__() what makes
303  possible to perform repr() on uninitialized pyasn1 types objects.
304- When comparing ASN.1 types, by-tag and/or by-constraints matching
305  can now be performed with the isSuperTypeOf()/isSameTypeWith() optional
306  flags.
307- Constructed types now verify their consistency by invoking
308  isSameTypeWith(matchTags=True, matchConstraints=False) and
309  isSuperTypeOf(matchTags=False, matchConstraints=True) for each of their
310  components rather than isSuperTypeOf() as it used to be. Constriants check
311  could be enforced to isSameTypeWith() with the strictConstraints=True
312  constructed classes attribute.
313- Constructed types can now be initialized with new .setComponents() method
314  which accepts both var-args and keyword-args. Default repr() modified to
315  reflect this change.
316- NamedTypes() and NamedValues() made comparable.
317- Test coverage extended to cover pyasn1 types __repr__() function.
318- The abs(Integer()) & abs(Real()) operation now returns respective pyasn1
319  type, not a Python type.
320- More Python magic methods implementations added to Integer & Real classes
321  (e.g.  __pos__, __neg__, __round__, __floor__, __ceil__, __trunc__)
322- The Integer.__invert__ Python magic method implemented.
323- The OctetString.__int__() and .__float__() magic methods implemented.
324- Handle the case of null writer at Debug printer.
325- BitString encoder/decoder performance improved.
326- Built-in debugging is now based on Python logging module.
327- Fix to NamedType.__repr__() to work properly.
328- Fixes to __repr__() implementation of many built-in ASN.1 types to take into
329  account all of their initializers such as tagSet, subtypeSpec etc.
330- String typed float initializer to REAL type now supported.
331- Float typed mantissa initializer to REAL type for base 2 added.
332- Encoding bases 8 and 16 support for REAL type binary encoder added.
333- More strict CER/DER encoders added for GeneralizedTime and UTCTime types.
334- Asn1Item.hasValue() added to easily distinguish initalized ASN.1 objects
335  from uninitialized ones (e.g. pure types).
336- Fix to REAL type binary decoder to handle different bases and scale factor.
337- Fix to TagSet.repr() to include [obsolete] baseTag information.
338- Fix to broken REAL type decoding handling.
339- Fix to BitString and OctetString decoders dealing with constructed
340  encoding -- it used to be possible to embed other types in substrate.
341- DER codec hardened not to tolerate indefinite length encoding/decoding.
342- Fix to end-of-octest sentinel handling:
343
344  + require strict two-zeros sentinel encoding
345  + recognize EOO sentinel only when explicitly requested by caller
346    of the decoder via allowEoo=True parameter (warning: API change)
347
348Revision 0.1.7
349--------------
350
351- License updated to vanilla BSD 2-Clause to ease package use
352  (https://opensource.org/licenses/BSD-2-Clause).
353- Test suite made discoverable by unittest/unittest2 discovery feature.
354- Fix to decoder working on indefinite length substrate -- end-of-octets
355  marker is now detected by both tag and value. Otherwise zero values may
356  interfere with end-of-octets marker.
357- Fix to decoder to fail in cases where tagFormat indicates inappropriate
358  format for the type (e.g. BOOLEAN is always PRIMITIVE, SET is always
359  CONSTRUCTED and OCTET STRING is either of the two)
360- Fix to REAL type encoder to force primitive encoding form encoding.
361- Fix to CHOICE decoder to handle explicitly tagged, indefinite length
362  mode encoding
363- Fix to REAL type decoder to handle negative REAL values correctly. Test
364  case added.
365
366Revision 0.1.6
367--------------
368
369- The compact (valueless) way of encoding zero INTEGERs introduced in
370  0.1.5 seems to fail miserably as the world is filled with broken
371  BER decoders. So we had to back off the *encoder* for a while.
372  There's still the IntegerEncoder.supportCompactZero flag which
373  enables compact encoding form whenever it evaluates to True.
374- Report package version on debugging code initialization.
375
376Revision 0.1.5
377--------------
378
379- Documentation updated and split into chapters to better match
380  web-site contents.
381- Make prettyPrint() working for non-initialized pyasn1 data objects. It
382  used to throw an exception.
383- Fix to encoder to produce empty-payload INTEGER values for zeros
384- Fix to decoder to support empty-payload INTEGER and REAL values
385- Fix to unit test suites imports to be able to run each from
386  their current directory
387
388Revision 0.1.4
389--------------
390
391- Built-in codec debugging facility added
392- Added some more checks to ObjectIdentifier BER encoder catching
393  posible 2^8 overflow condition by two leading sub-OIDs
394- Implementations overriding the AbstractDecoder.valueDecoder method
395  changed to return the rest of substrate behind the item being processed
396  rather than the unprocessed substrate within the item (which is usually
397  empty).
398- Decoder's recursiveFlag feature generalized as a user callback function
399  which is passed an uninitialized object recovered from substrate and
400  its uninterpreted payload.
401- Catch inappropriate substrate type passed to decoder.
402- Expose tagMap/typeMap/Decoder objects at DER decoder to uniform API.
403- Obsolete __init__.MajorVersionId replaced with __init__.__version__
404  which is now in-sync with distutils.
405- Package classifiers updated.
406- The __init__.py's made non-empty (rumors are that they may be optimized
407  out by package managers).
408- Bail out gracefully whenever Python version is older than 2.4.
409- Fix to Real codec exponent encoding (should be in 2's complement form),
410  some more test cases added.
411- Fix in Boolean truth testing built-in methods
412- Fix to substrate underrun error handling at ObjectIdentifier BER decoder
413- Fix to BER Boolean decoder that allows other pre-computed
414  values besides 0 and 1
415- Fix to leading 0x80 octet handling in DER/CER/DER ObjectIdentifier decoder.
416  See https://www.esat.kuleuven.be/cosic/publications/article-1432.pdf
417
418Revision 0.1.3
419--------------
420
421- Include class name into asn1 value constraint violation exception.
422- Fix to OctetString.prettyOut() method that looses leading zero when
423  building hex string.
424
425Revision 0.1.2
426--------------
427
428- Fix to __long__() to actually return longs on py2k
429- Fix to OctetString.__str__() workings of a non-initialized object.
430- Fix to quote initializer of OctetString.__repr__()
431- Minor fix towards ObjectIdentifier.prettyIn() reliability
432- ObjectIdentifier.__str__() is aliased to prettyPrint()
433- Exlicit repr() calls replaced with '%r'
434
435Revision 0.1.1
436--------------
437
438- Hex/bin string initializer to OctetString object reworked
439  (in a backward-incompatible manner)
440- Fixed float() infinity compatibility issue (affects 2.5 and earlier)
441- Fixed a bug/typo at Boolean CER encoder.
442- Major overhawl for Python 2.4 -- 3.2 compatibility:
443  + get rid of old-style types
444  + drop string module usage
445  + switch to rich comparation
446  + drop explicit long integer type use
447  + map()/filter() replaced with list comprehension
448  + apply() replaced with \*/\*\*args
449  + switched to use 'key' sort() callback function
450  + support both __nonzero__() and __bool__() methods
451  + modified not to use py3k-incompatible exception syntax
452  + getslice() operator fully replaced with getitem()
453  + dictionary operations made 2K/3K compatible
454  + base type for encoding substrate and OctetString-based types
455  is now 'bytes' when running py3k and 'str' otherwise
456  + OctetString and derivatives now unicode compliant.
457  + OctetString now supports two python-neutral getters: asOcts() & asInts()
458  + print OctetString content in hex whenever it is not printable otherwise
459  + in test suite, implicit relative import replaced with the absolute one
460  + in test suite, string constants replaced with numerics
461
462Revision 0.0.13
463---------------
464
465- Fix to base10 normalization function that loops on univ.Real(0)
466
467Revision 0.0.13b
468----------------
469
470- ASN.1 Real type is now supported properly.
471- Objects of Constructed types now support __setitem__()
472- Set/Sequence objects can now be addressed by their field names (string index)
473  and position (integer index).
474- Typo fix to ber.SetDecoder code that prevented with schema decoding
475  operation.
476- Fix to explicitly tagged items decoding support.
477- Fix to OctetString.prettyPrint() to better handle non-printable content.
478- Fix to repr() workings of Choice objects.
479
480Revision 0.0.13a
481----------------
482
483- Major codec re-design.
484- Documentation significantly improved.
485- ASN.1 Any type is now supported.
486- All example ASN.1 modules moved to separate pyasn1-modules package.
487- Fix to initial sub-OID overflow condition detection an encoder.
488- BitString initialization value verification improved.
489- The Set/Sequence.getNameByPosition() method implemented.
490- Fix to proper behaviour of PermittedAlphabetConstraint object.
491- Fix to improper Boolean substrate handling at CER/DER decoders.
492- Changes towards performance improvement:
493
494  + all dict.has_key() & dict.get() invocations replaced with modern syntax
495    (this breaks compatibility with Python 2.1 and older).
496  + tag and tagset caches introduced to decoder
497  + decoder code improved to prevent unnecessary pyasn1 objects creation
498  + allow disabling components verification when setting components to
499    structured types, this is used by decoder whilst running with schema
500    mode.
501  + BER decoder for integer values now looks up a small set of pre-computed
502    substrate values to save on decoding.
503  + a few pre-computed values configured to ObjectIdentifier BER encoder.
504  + ChoiceDecoder split-off SequenceOf one to save on unnecessary checks.
505  + replace slow hasattr()/getattr() calls with isinstance() introspection.
506  + track the number of initialized components of Constructed types to save
507    on default/optional components initialization.
508  + added a shortcut ObjectIdentifier.asTuple() to be used instead of
509    __getitem__() in hotspots.
510  + use Tag.asTuple() and pure integers at tag encoder.
511  + introduce and use in decoder the baseTagSet attribute of the built-in
512    ASN.1 types.
513
514Revision 0.0.12a
515----------------
516
517- The individual tag/length/value processing methods of
518  encoder.AbstractItemEncoder renamed (leading underscore stripped)
519  to promote overloading in cases where partial substrate processing
520  is required.
521- The ocsp.py, ldap.py example scripts added.
522- Fix to univ.ObjectIdentifier input value handler to disallow negative
523  sub-IDs.
524
525Revision 0.0.11a
526----------------
527
528- Decoder can now treat values of unknown types as opaque OctetString.
529- Fix to Set/SetOf type decoder to handle uninitialized scalar SetOf
530  components correctly.
531
532Revision 0.0.10a
533----------------
534
535- API versioning mechanics retired (pyasn1.v1 -> pyasn1) what makes
536  it possible to zip-import pyasn1 sources (used by egg and py2exe).
537
538Revision 0.0.9a
539---------------
540
541- Allow any non-zero values in Boolean type BER decoder, as it's in
542  accordnance with the standard.
543
544Revision 0.0.8a
545---------------
546
547- Integer.__index__() now supported (for Python 2.5+).
548- Fix to empty value encoding in BitString encoder, test case added.
549- Fix to SequenceOf decoder that prevents it skipping possible Choice
550  typed inner component.
551- Choice.getName() method added for getting currently set component
552  name.
553- OctetsString.prettyPrint() does a single str() against its value
554  eliminating an extra quotes.
555
556Revision 0.0.7a
557---------------
558
559- Large tags (>31) now supported by codecs.
560- Fix to encoder to properly handle explicitly tagged untagged items.
561- All possible value lengths (up to 256^126) now supported by encoders.
562- Fix to Tag class constructor to prevent negative IDs.
563
564Revision 0.0.6a
565---------------
566
567- Make use of setuptools.
568- Constraints derivation verification (isSuperTypeOf()/isSubTypeOf()) fixed.
569- Fix to constraints comparation logic -- can't cmp() hash values as it
570  may cause false positives due to hash conflicts.
571
572Revision 0.0.5a
573---------------
574
575- Integer BER codec reworked fixing negative values encoding bug.
576- clone() and subtype() methods of Constructed ASN.1 classes now
577  accept optional cloneValueFlag flag which controls original value
578  inheritance. The default is *not* to inherit original value for
579  performance reasons (this may affect backward compatibility).
580  Performance penalty may be huge on deeply nested Constructed objects
581  re-creation.
582- Base ASN.1 types (pyasn1.type.univ.*) do not have default values
583  anymore. They remain uninitialized acting as ASN.1 types. In
584  this model, initialized ASN.1 types represent either types with
585  default value installed or a type instance.
586- Decoders' prototypes are now class instances rather than classes.
587  This is to simplify initial value installation to decoder's
588  prototype value.
589- Bugfix to BitString BER decoder (trailing bits not regarded).
590- Bugfix to Constraints use as mapping keys.
591- Bugfix to Integer & BitString clone() methods
592- Bugix to the way to distinguish Set from SetOf at CER/DER SetOfEncoder
593- Adjustments to make it running on Python 1.5.
594- In tests, substrate constants converted from hex escaped literals into
595  octals to overcome indefinite hex width issue occuring in young Python.
596- Minor performance optimization of TagSet.isSuperTagSetOf() method
597- examples/sshkey.py added
598
599Revision 0.0.4a
600---------------
601
602* Asn1ItemBase.prettyPrinter() -> \*.prettyPrint()
603
604Revision 0.0.3a
605---------------
606
607* Simple ASN1 objects now hash to their Python value and don't
608  depend upon tag/constraints/etc.
609* prettyIn & prettyOut methods of SimplleAsn1Object become public
610* many syntax fixes
611
612Revision 0.0.2a
613---------------
614
615* ConstraintsIntersection.isSuperTypeOf() and
616  ConstraintsIntersection.hasConstraint() implemented
617* Bugfix to NamedValues initialization code
618* +/- operators added to NamedValues objects
619* Integer.__abs__() & Integer.subtype() added
620* ObjectIdentifier.prettyOut() fixes
621* Allow subclass components at SequenceAndSetBase
622* AbstractConstraint.__cmp__() dropped
623* error.Asn1Error replaced with error.PyAsn1Error
624
625Revision 0.0.1a
626---------------
627
628* Initial public alpha release
629