forked from CNXTEoEorg/sqlalchemy
-
Notifications
You must be signed in to change notification settings - Fork 0
/
CHANGES
4890 lines (4060 loc) · 192 KB
/
CHANGES
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
-*- coding: utf-8; fill-column: 68 -*-
=======
CHANGES
=======
0.7.4
=====
- orm
- [bug] Fixed backref behavior when "popping" the
value off of a many-to-one in response to
a removal from a stale one-to-many - the operation
is skipped, since the many-to-one has since
been updated. [ticket:2315]
- [bug] After some years of not doing this, added
more granularity to the "is X a parent of Y"
functionality, which is used when determining
if the FK on "Y" needs to be "nulled out" as well
as if "Y" should be deleted with delete-orphan
cascade. The test now takes into account the
Python identity of the parent as well its identity
key, to see if the last known parent of Y is
definitely X. If a decision
can't be made, a StaleDataError is raised. The
conditions where this error is raised are fairly
rare, requiring that the previous parent was
garbage collected, and previously
could very well inappropriately update/delete
a record that's since moved onto a new parent,
though there may be some cases where
"silent success" occurred previously that will now
raise in the face of ambiguity.
Expiring "Y" resets the "parent" tracker, meaning
X.remove(Y) could then end up deleting Y even
if X is stale, but this is the same behavior
as before; it's advised to expire X also in that
case. [ticket:2264]
- [bug] fixed inappropriate evaluation of user-mapped
object in a boolean context within query.get()
[ticket:2310]. Also in 0.6.9.
- [bug] Added missing comma to PASSIVE_RETURN_NEVER_SET
symbol [ticket:2304]
- [bug] Cls.column.collate("some collation") now
works. [ticket:1776] Also in 0.6.9
- [bug] the value of a composite attribute is now
expired after an insert or update operation, instead
of regenerated in place. This ensures that a
column value which is expired within a flush
will be loaded first, before the composite
is regenerated using that value. [ticket:2309]
- [bug] The fix in [ticket:2309] also emits the
"refresh" event when the composite value is
loaded on access, even if all column
values were already present, as is appropriate.
This fixes the "mutable" extension which relies
upon the "load" event to ensure the _parents
dictionary is up to date, fixes [ticket:2308].
Thanks to Scott Torborg for the test case here.
- [bug] Fixed bug whereby a subclass of a subclass
using concrete inheritance in conjunction with
the new ConcreteBase or AbstractConcreteBase
would fail to apply the subclasses deeper than
one level to the "polymorphic loader" of each
base [ticket:2312]
- [bug] Fixed bug whereby a subclass of a subclass
using the new AbstractConcreteBase would fail
to acquire the correct "base_mapper" attribute
when the "base" mapper was generated, thereby
causing failures later on. [ticket:2312]
- [bug] Fixed bug whereby column_property() created
against ORM-level column could be treated as
a distinct entity when producing certain
kinds of joined-inh joins. [ticket:2316]
- [bug] Fixed the error formatting raised when
a tuple is inadvertently passed to session.query()
[ticket:2297]. Also in 0.6.9.
- [bug] Calls to query.join() to a single-table
inheritance subclass are now tracked, and
are used to eliminate the additional WHERE..
IN criterion normally tacked on with single
table inheritance, since the join should
accommodate it. This allows OUTER JOIN
to a single table subclass to produce
the correct results, and overall will produce
fewer WHERE criterion when dealing with
single table inheritance joins.
[ticket:2328]
- [bug] __table_args__ can now be passed as
an empty tuple as well as an empty dict.
[ticket:2339]. Thanks to Fayaz Yusuf Khan
for the patch.
- [bug] Updated warning message when setting
delete-orphan without delete to no longer
refer to 0.6, as we never got around to
upgrading this to an exception. Ideally
this might be better as an exception but
it's not critical either way. [ticket:2325]
- [feature] polymorphic_on now accepts many
new kinds of values:
- standalone expressions that aren't
otherwise mapped
- column_property() objects
- string names of any column_property()
or attribute name of a mapped Column
The docs include an example using
the case() construct, which is likely to be
a common constructed used here.
[ticket:2345] and part of [ticket:2238]
Standalone expressions in polymorphic_on
propagate to single-table inheritance
subclasses so that they are used in the
WHERE /JOIN clause to limit rows to that
subclass as is the usual behavior.
- [feature] IdentitySet supports the - operator
as the same as difference(), handy when dealing
with Session.dirty etc. [ticket:2301]
- [feature] Added new value for Column autoincrement
called "ignore_fk", can be used to force autoincrement
on a column that's still part of a ForeignKeyConstraint.
New example in the relationship docs illustrates
its use.
- [bug] Fixed bug in get_history() when referring
to a composite attribute that has no value;
added coverage for get_history() regarding
composites which is otherwise just a userland
function.
- sql
- [bug] related to [ticket:2316], made some
adjustments to the change from [ticket:2261]
regarding the "from" list on a select(). The
_froms collection is no longer memoized, as this
simplifies various use cases and removes the
need for a "warning" if a column is attached
to a table after it was already used in an
expression - the select() construct will now
always produce the correct expression.
There's probably no real-world
performance hit here; select() objects are
almost always made ad-hoc, and systems that
wish to optimize the re-use of a select()
would be using the "compiled_cache" feature.
A hit which would occur when calling select.bind
has been reduced, but the vast majority
of users shouldn't be using "bound metadata"
anyway :).
- [feature] The update() construct can now accommodate
multiple tables in the WHERE clause, which will
render an "UPDATE..FROM" construct, recognized by
Postgresql and MSSQL. When compiled on MySQL,
will instead generate "UPDATE t1, t2, ..". MySQL
additionally can render against multiple tables in the
SET clause, if Column objects are used as keys
in the "values" parameter or generative method.
[ticket:2166] [ticket:1944]
- [feature] Added accessor to types called "python_type",
returns the rudimentary Python type object
for a particular TypeEngine instance, if known,
else raises NotImplementedError. [ticket:77]
- [bug] further tweak to the fix from [ticket:2261],
so that generative methods work a bit better
off of cloned (this is almost a non-use case though).
In particular this allows with_only_columns()
to behave more consistently. Added additional
documentation to with_only_columns() to clarify
expected behavior, which changed as a result
of [ticket:2261]. [ticket:2319]
- engine
- [bug] Fixed bug whereby transaction.rollback()
would throw an error on an invalidated
connection if the transaction were a
two-phase or savepoint transaction.
For plain transactions, rollback() is a no-op
if the connection is invalidated, so while
it wasn't 100% clear if it should be a no-op,
at least now the interface is consistent.
[ticket:2317]
- schema
- [feature] Added new support for remote "schemas":
- MetaData() accepts "schema" and "quote_schema"
arguments, which will be applied to the same-named
arguments of a Table
or Sequence which leaves these at their default
of ``None``.
- Sequence accepts "quote_schema" argument
- tometadata() for Table will use the "schema"
of the incoming MetaData for the new Table
if the schema argument is explicitly "None"
- Added CreateSchema and DropSchema DDL
constructs - these accept just the string
name of a schema and a "quote" flag.
- When using default "schema" with MetaData,
ForeignKey will also assume the "default" schema
when locating remote table. This allows the "schema"
argument on MetaData to be applied to any
set of Table objects that otherwise don't have
a "schema".
- a "has_schema" method has been implemented
on dialect, but only works on Postgresql so far.
Courtesy Manlio Perillo, [ticket:1679]
- [feature] The "extend_existing" flag on Table
now allows for the reflection process to take
effect for a Table object that's already been
defined; when autoload=True and extend_existing=True
are both set, the full set of columns will be
reflected from the Table which will then
*overwrite* those columns already present,
rather than no activity occurring. Columns that
are present directly in the autoload run
will be used as always, however.
[ticket:1410]
- [bug] Fixed bug whereby TypeDecorator would
return a stale value for _type_affinity, when
using a TypeDecorator that "switches" types,
like the CHAR/UUID type.
- [bug] Fixed bug whereby "order_by='foreign_key'"
option to Inspector.get_table_names
wasn't implementing the sort properly, replaced
with the existing sort algorithm
- [bug] the "name" of a column-level CHECK constraint,
if present, is now rendered in the CREATE TABLE
statement using "CONSTRAINT <name> CHECK <expression>".
[ticket:2305]
- pyodbc
- [bug] pyodbc-based dialects now parse the
pyodbc accurately as far as observed
pyodbc strings, including such gems
as "py3-3.0.1-beta4" [ticket:2318]
- postgresql
- [bug] Postgresql dialect memoizes that an ENUM of a
particular name was processed
during a create/drop sequence. This allows
a create/drop sequence to work without any
calls to "checkfirst", and also means with
"checkfirst" turned on it only needs to
check for the ENUM once. [ticket:2311]
- [feature] Added create_type constructor argument
to pg.ENUM. When False, no CREATE/DROP or
checking for the type will be performed as part
of a table create/drop event; only the
create()/drop)() methods called directly
will do this. Helps with Alembic "offline"
scripts.
- mssql
- [feature] lifted the restriction on SAVEPOINT
for SQL Server. All tests pass using it,
it's not known if there are deeper issues
however. [ticket:822]
- [bug] repaired the with_hint() feature which
wasn't implemented correctly on MSSQL -
usually used for the "WITH (NOLOCK)" hint
(which you shouldn't be using anyway !
use snapshot isolation instead :) )
[ticket:2336]
- [bug] use new pyodbc version detection for
_need_decimal_fix option, [ticket:2318]
- [bug] don't cast "table name" as NVARCHAR
on SQL Server 2000. Still mostly in the dark
what incantations are needed to make PyODBC
work fully with FreeTDS 0.91 here, however.
[ticket:2343]
- [bug] Decode incoming values when retrieving
list of index names and the names of columns
within those indexes. [ticket:2269]
- mysql
- [bug] Unicode adjustments allow latest pymysql
(post 0.4) to pass 100% on Python 2.
- ext
- [feature] Added an example to the hybrid docs
of a "transformer" - a hybrid that returns a
query-transforming callable in combination
with a custom comparator. Uses a new method
on Query called with_transformation(). The use
case here is fairly experimental, but only
adds one line of code to Query.
- [bug] the @compiles decorator raises an
informative error message when no "default"
compilation handler is present, rather
than KeyError.
- examples
- [bug] Fixed bug in history_meta.py example where
the "unique" flag was not removed from a
single-table-inheritance subclass which
generates columns to put up onto the base.
0.7.3
=====
- general
- Adjusted the "importlater" mechanism, which is
used internally to resolve import cycles,
such that the usage of __import__ is completed
when the import of sqlalchemy or sqlalchemy.orm
is done, thereby avoiding any usage of __import__
after the application starts new threads,
fixes [ticket:2279]. Also in 0.6.9.
- orm
- Improved query.join() such that the "left" side
can more flexibly be a non-ORM selectable,
such as a subquery. A selectable placed
in select_from() will now be used as the left
side, favored over implicit usage
of a mapped entity.
If the join still fails based on lack of
foreign keys, the error message includes
this detail. Thanks to brianrhude
on IRC for the test case. [ticket:2298]
- Added after_soft_rollback() Session event. This
event fires unconditionally whenever rollback()
is called, regardless of if an actual DBAPI
level rollback occurred. This event
is specifically designed to allow operations
with the Session to proceed after a rollback
when the Session.is_active is True.
[ticket:2241]
- added "adapt_on_names" boolean flag to orm.aliased()
construct. Allows an aliased() construct
to link the ORM entity to a selectable that contains
aggregates or other derived forms of a particular
attribute, provided the name is the same as that
of the entity mapped column.
- Added new flag expire_on_flush=False to column_property(),
marks those properties that would otherwise be considered
to be "readonly", i.e. derived from SQL expressions,
to retain their value after a flush has occurred, including
if the parent object itself was involved in an update.
- Enhanced the instrumentation in the ORM to support
Py3K's new argument style of "required kw arguments",
i.e. fn(a, b, *, c, d), fn(a, b, *args, c, d).
Argument signatures of mapped object's __init__
method will be preserved, including required kw rules.
[ticket:2237]
- Fixed bug in unit of work whereby detection of
"cycles" among classes in highly interlinked patterns
would not produce a deterministic
result; thereby sometimes missing some nodes that
should be considered cycles and causing further
issues down the road. Note this bug is in 0.6
also; not backported at the moment.
[ticket:2282]
- Fixed a variety of synonym()-related regressions
from 0.6:
- making a synonym against a synonym now works.
- synonyms made against a relationship() can
be passed to query.join(), options sent
to query.options(), passed by name
to query.with_parent().
- Fixed bug whereby mapper.order_by attribute would
be ignored in the "inner" query within a
subquery eager load. [ticket:2287].
Also in 0.6.9.
- Identity map .discard() uses dict.pop(,None)
internally instead of "del" to avoid KeyError/warning
during a non-determinate gc teardown [ticket:2267]
- Fixed regression in new composite rewrite where
deferred=True option failed due to missing
import [ticket:2253]
- Reinstated "comparator_factory" argument to
composite(), removed when 0.7 was released.
[ticket:2248]
- Fixed bug in query.join() which would occur
in a complex multiple-overlapping path scenario,
where the same table could be joined to
twice. Thanks *much* to Dave Vitek
for the excellent fix here. [ticket:2247]
- Query will convert an OFFSET of zero when
slicing into None, so that needless OFFSET
clauses are not invoked.
- Repaired edge case where mapper would fail
to fully update internal state when a relationship
on a new mapper would establish a backref on the
first mapper.
- Fixed bug whereby if __eq__() was
redefined, a relationship many-to-one lazyload
would hit the __eq__() and fail. [ticket:2260]
Does not apply to 0.6.9.
- Calling class_mapper() and passing in an object
that is not a "type" (i.e. a class that could
potentially be mapped) now raises an informative
ArgumentError, rather than UnmappedClassError.
[ticket:2196]
- New event hook, MapperEvents.after_configured().
Called after a configure() step has completed and
mappers were in fact affected. Theoretically this
event is called once per application, unless new mappings
are constructed after existing ones have been used
already.
- When an open Session is garbage collected, the objects
within it which remain are considered detached again
when they are add()-ed to a new Session.
This is accomplished by an extra check that the previous
"session_key" doesn't actually exist among the pool
of Sessions. [ticket:2281]
- New declarative features:
- __declare_last__() method, establishes an event
listener for the class method that will be called
when mappers are completed with the final "configure"
step.
- __abstract__ flag. The class will not be mapped
at all when this flag is present on the class.
- New helper classes ConcreteBase, AbstractConcreteBase.
Allow concrete mappings using declarative which automatically
set up the "polymorphic_union" when the "configure"
mapper step is invoked.
- The mapper itself has semi-private methods that allow
the "with_polymorphic" selectable to be assigned
to the mapper after it has already been configured.
[ticket:2239]
- Declarative will warn when a subclass' base uses
@declared_attr for a regular column - this attribute
does not propagate to subclasses. [ticket:2283]
- The integer "id" used to link a mapped instance with
its owning Session is now generated by a sequence
generation function rather than id(Session), to
eliminate the possibility of recycled id() values
causing an incorrect result, no need to check that
object actually in the session. [ticket:2280]
-sql
- Behavioral improvement: empty
conjunctions such as and_() and or_() will be
flattened in the context of an enclosing conjunction,
i.e. and_(x, or_()) will produce 'X' and not 'X AND
()'. [ticket:2257].
- Fixed bug regarding calculation of "from" list
for a select() element. The "from" calc is now
delayed, so that if the construct uses a Column
object that is not yet attached to a Table,
but is later associated with a Table, it generates
SQL using the table as a FROM. This change
impacted fairly deeply the mechanics of how
the FROM list as well as the "correlates" collection
is calculated, as some "clause adaption" schemes
(these are used very heavily in the ORM)
were relying upon the fact that the "froms"
collection would typically be cached before the
adaption completed. The rework allows it
such that the "froms" collection can be cleared
and re-generated at any time. [ticket:2261]
- Fixed bug whereby with_only_columns() method of
Select would fail if a selectable were passed.
[ticket:2270]. Also in 0.6.9.
- schema
- Modified Column.copy() to use _constructor(),
which defaults to self.__class__, in order to
create the new object. This allows easier support
of subclassing Column. [ticket:2284]
- Added a slightly nicer __repr__() to SchemaItem
classes. Note the repr here can't fully support
the "repr is the constructor" idea since schema
items can be very deeply nested/cyclical, have
late initialization of some things, etc.
[ticket:2223]
- engine
- The recreate() method in all pool classes uses
self.__class__ to get at the type of pool
to produce, in the case of subclassing. Note
there's no usual need to subclass pools.
[ticket:2254]
- Improvement to multi-param statement logging,
long lists of bound parameter sets will be
compressed with an informative indicator
of the compression taking place. Exception
messages use the same improved formatting.
[ticket:2243]
- Added optional "sa_pool_key" argument to
pool.manage(dbapi).connect() so that serialization
of args is not necessary.
- The entry point resolution supported by
create_engine() now supports resolution of
individual DBAPI drivers on top of a built-in
or entry point-resolved dialect, using the
standard '+' notation - it's converted to
a '.' before being resolved as an entry
point. [ticket:2286]
- Added an exception catch + warning for the
"return unicode detection" step within connect,
allows databases that crash on NVARCHAR to
continue initializing, assuming no NVARCHAR
type implemented. [ticket:2299]
- types
- Extra keyword arguments to the base Float
type beyond "precision" and "asdecimal" are ignored;
added a deprecation warning here and additional
docs, related to [ticket:2258]
- sqlite
- Ensured that the same ValueError is raised for
illegal date/time/datetime string parsed from
the database regardless of whether C
extensions are in use or not.
- postgresql
- Added "postgresql_using" argument to Index(), produces
USING clause to specify index implementation for
PG. [ticket:2290]. Thanks to Ryan P. Kelly for
the patch.
- Added client_encoding parameter to create_engine()
when the postgresql+psycopg2 dialect is used;
calls the psycopg2 set_client_encoding() method
with the value upon connect. [ticket:1839]
- Fixed bug related to [ticket:2141] whereby the
same modified index behavior in PG 9 affected
primary key reflection on a renamed column.
[ticket:2291]. Also in 0.6.9.
- Reflection functions for Table, Sequence no longer
case insensitive. Names can be differ only in case
and will be correctly distinguished. [ticket:2256]
- Use an atomic counter as the "random number"
source for server side cursor names;
conflicts have been reported in rare cases.
- Narrowed the assumption made when reflecting
a foreign-key referenced table with schema in
the current search path; an explicit schema will
be applied to the referenced table only if
it actually matches that of the referencing table,
which also has an explicit schema. Previously
it was assumed that "current" schema was synonymous
with the full search_path. [ticket:2249]
- mysql
- a CREATE TABLE will put the COLLATE option
after CHARSET, which appears to be part of
MySQL's arbitrary rules regarding if it will actually
work or not. [ticket:2225] Also in 0.6.9.
- Added mysql_length parameter to Index construct,
specifies "length" for indexes. [ticket:2293]
- mssql
- Changes to attempt support of FreeTDS 0.91 with
Pyodbc. This includes that string binds are sent as
Python unicode objects when FreeTDS 0.91 is detected,
and a CAST(? AS NVARCHAR) is used when we detect
for a table. However, I'd continue
to characterize Pyodbc + FreeTDS 0.91 behavior as
pretty crappy, there are still many queries such
as used in reflection which cause a core dump on
Linux, and it is not really usable at all
on OSX, MemoryErrors abound and just plain broken
unicode support. [ticket:2273]
- The behavior of =/!= when comparing a scalar select
to a value will no longer produce IN/NOT IN as of 0.8;
this behavior is a little too heavy handed (use in_() if
you want to emit IN) and now emits a deprecation warning.
To get the 0.8 behavior immediately and remove the warning,
a compiler recipe is given at
http://www.sqlalchemy.org/docs/07/dialects/mssql.html#scalar-select-comparisons
to override the behavior of visit_binary().
[ticket:2277]
- "0" is accepted as an argument for limit() which
will produce "TOP 0". [ticket:2222]
- oracle
- Fixed ReturningResultProxy for zxjdbc dialect.
[ticket:2272]. Regression from 0.6.
- The String type now generates VARCHAR2 on Oracle
which is recommended as the default VARCHAR.
Added an explicit VARCHAR2 and NVARCHAR2 to the Oracle
dialect as well. Using NVARCHAR still generates
"NVARCHAR2" - there is no "NVARCHAR" on Oracle -
this remains a slight breakage of the "uppercase types
always give exactly that" policy. VARCHAR still
generates "VARCHAR", keeping with the policy. If
Oracle were to ever define "VARCHAR" as something
different as they claim (IMHO this will never happen),
the type would be available. [ticket:2252]
- ext
- SQLSoup will not be included in version 0.8
of SQLAlchemy; while useful, we would like to
keep SQLAlchemy itself focused on one ORM
usage paradigm. SQLSoup will hopefully
soon be superseded by a third party
project. [ticket:2262]
- Added local_attr, remote_attr, attr accessors
to AssociationProxy, providing quick access
to the proxied attributes at the class
level. [ticket:2236]
- Changed the update() method on association proxy
dictionary to use a duck typing approach, i.e.
checks for "keys", to discern between update({})
and update((a, b)). Previously, passing a
dictionary that had tuples as keys would be misinterpreted
as a sequence. [ticket:2275]
- examples
- Adjusted dictlike-polymorphic.py example
to apply the CAST such that it works on
PG, other databases. [ticket:2266]
Also in 0.6.9.
0.7.2
=====
- orm
- Feature enhancement: joined and subquery
loading will now traverse already-present related
objects and collections in search of unpopulated
attributes throughout the scope of the eager load
being defined, so that the eager loading that is
specified via mappings or query options
unconditionally takes place for the full depth,
populating whatever is not already populated.
Previously, this traversal would stop if a related
object or collection were already present leading
to inconsistent behavior (though would save on
loads/cycles for an already-loaded graph). For a
subqueryload, this means that the additional
SELECT statements emitted by subqueryload will
invoke unconditionally, no matter how much of the
existing graph is already present (hence the
controversy). The previous behavior of "stopping"
is still in effect when a query is the result of
an attribute-initiated lazyload, as otherwise an
"N+1" style of collection iteration can become
needlessly expensive when the same related object
is encountered repeatedly. There's also an
as-yet-not-public generative Query method
_with_invoke_all_eagers()
which selects old/new behavior [ticket:2213]
- A rework of "replacement traversal" within
the ORM as it alters selectables to be against
aliases of things (i.e. clause adaption) includes
a fix for multiply-nested any()/has() constructs
against a joined table structure. [ticket:2195]
- Fixed bug where query.join() + aliased=True
from a joined-inh structure to itself on
relationship() with join condition on the child
table would convert the lead entity into the
joined one inappropriately. [ticket:2234]
Also in 0.6.9.
- Fixed regression from 0.6 where Session.add()
against an object which contained None in a
collection would raise an internal exception.
Reverted this to 0.6's behavior which is to
accept the None but obviously nothing is
persisted. Ideally, collections with None
present or on append() should at least emit a
warning, which is being considered for 0.8.
[ticket:2205]
- Load of a deferred() attribute on an object
where row can't be located raises
ObjectDeletedError instead of failing later
on; improved the message in ObjectDeletedError
to include other conditions besides a simple
"delete". [ticket:2191]
- Fixed regression from 0.6 where a get history
operation on some relationship() based attributes
would fail when a lazyload would emit; this could
trigger within a flush() under certain conditions.
[ticket:2224] Thanks to the user who submitted
the great test for this.
- Fixed bug apparent only in Python 3 whereby
sorting of persistent + pending objects during
flush would produce an illegal comparison,
if the persistent object primary key
is not a single integer. [ticket:2228]
Also in 0.6.9
- Fixed bug whereby the source clause
used by query.join() would be inconsistent
if against a column expression that combined
multiple entities together. [ticket:2197]
Also in 0.6.9
- Fixed bug whereby if a mapped class
redefined __hash__() or __eq__() to something
non-standard, which is a supported use case
as SQLA should never consult these,
the methods would be consulted if the class
was part of a "composite" (i.e. non-single-entity)
result set. [ticket:2215]
Also in 0.6.9.
- Added public attribute ".validators" to
Mapper, an immutable dictionary view of
all attributes that have been decorated
with the @validates decorator.
[ticket:2240] courtesy Stefano Fontanelli
- Fixed subtle bug that caused SQL to blow
up if: column_property() against subquery +
joinedload + LIMIT + order by the column
property() occurred. [ticket:2188].
Also in 0.6.9
- The join condition produced by with_parent
as well as when using a "dynamic" relationship
against a parent will generate unique
bindparams, rather than incorrectly repeating
the same bindparam. [ticket:2207].
Also in 0.6.9.
- Added the same "columns-only" check to
mapper.polymorphic_on as used when
receiving user arguments to
relationship.order_by, foreign_keys,
remote_side, etc.
- Fixed bug whereby comparison of column
expression to a Query() would not call
as_scalar() on the underlying SELECT
statement to produce a scalar subquery,
in the way that occurs if you called
it on Query().subquery(). [ticket:2190]
- Fixed declarative bug where a class inheriting
from a superclass of the same name would fail
due to an unnecessary lookup of the name
in the _decl_class_registry. [ticket:2194]
- Repaired the "no statement condition"
assertion in Query which would attempt
to raise if a generative method were called
after from_statement() were called.
[ticket:2199]. Also in 0.6.9.
- sql
- Fixed two subtle bugs involving column
correspondence in a selectable,
one with the same labeled subquery repeated, the other
when the label has been "grouped" and
loses itself. Affects [ticket:2188].
- schema
- New feature: with_variant() method on
all types. Produces an instance of Variant(),
a special TypeDecorator which will select
the usage of a different type based on the
dialect in use. [ticket:2187]
- Added an informative error message when
ForeignKeyConstraint refers to a column name in
the parent that is not found. Also in 0.6.9.
- Fixed bug whereby adaptation of old append_ddl_listener()
function was passing unexpected **kw through
to the Table event. Table gets no kws, the MetaData
event in 0.6 would get "tables=somecollection",
this behavior is preserved. [ticket:2206]
- Fixed bug where "autoincrement" detection on
Table would fail if the type had no "affinity"
value, in particular this would occur when using
the UUID example on the site that uses TypeEngine
as the "impl".
- Added an improved repr() to TypeEngine objects
that will only display constructor args which
are positional or kwargs that deviate
from the default. [ticket:2209]
- engine
- Context manager provided by Connection.begin()
will issue rollback() if the commit() fails,
not just if an exception occurs.
- Use urllib.parse_qsl() in Python 2.6 and above,
no deprecation warning about cgi.parse_qsl()
[ticket:1682]
- Added mixin class sqlalchemy.ext.DontWrapMixin.
User-defined exceptions of this type are never
wrapped in StatementException when they
occur in the context of a statement
execution.
- StatementException wrapping will display the
original exception class in the message.
- Failures on connect which raise dbapi.Error
will forward the error to dialect.is_disconnect()
and set the "connection_invalidated" flag if
the dialect knows this to be a potentially
"retryable" condition. Only Oracle ORA-01033
implemented for now. [ticket:2201]
- sqlite
- SQLite dialect no longer strips quotes
off of reflected default value, allowing
a round trip CREATE TABLE to work.
This is consistent with other dialects
that also maintain the exact form of
the default. [ticket:2189]
- postgresql
- Added new "postgresql_ops" argument to
Index, allows specification of PostgreSQL
operator classes for indexed columns.
[ticket:2198] Courtesy Filip Zyzniewski.
- mysql
- Fixed OurSQL dialect to use ansi-neutral
quote symbol "'" for XA commands instead
of '"'. [ticket:2186]. Also in 0.6.9.
- mssql
- Adjusted the pyodbc dialect such that bound
values are passed as bytes and not unicode
if the "Easysoft" unix drivers are detected.
This is the same behavior as occurs with
FreeTDS. Easysoft appears to segfault
if Python unicodes are passed under
certain circumstances.
- oracle
- Added ORA-00028 to disconnect codes, use
cx_oracle _Error.code to get at the code,
[ticket:2200]. Also in 0.6.9.
- Added ORA-01033 to disconnect codes, which
can be caught during a connection
event. [ticket:2201]
- repaired the oracle.RAW type which did not
generate the correct DDL. [ticket:2220]
Also in 0.6.9.
- added CURRENT to reserved word list.
[ticket:2212] Also in 0.6.9.
-ext
- Fixed bug in the mutable extension whereby
if the same type were used twice in one
mapping, the attributes beyond the first
would not get instrumented.
- Fixed bug in the mutable extension whereby
if None or a non-corresponding type were set,
an error would be raised. None is now accepted
which assigns None to all attributes,
illegal values raise ValueError.
- examples
- Repaired the examples/versioning test runner
to not rely upon SQLAlchemy test libs,
nosetests must be run from within
examples/versioning to get around setup.cfg
breaking it.
- Tweak to examples/versioning to pick the
correct foreign key in a multi-level
inheritance situation.
- Fixed the attribute shard example to check
for bind param callable correctly in 0.7
style.
0.7.1
=====
- general
- Added a workaround for Python bug 7511 where
failure of C extension build does not
raise an appropriate exception on Windows 64
bit + VC express [ticket:2184]
- orm
- "delete-orphan" cascade is now allowed on
self-referential relationships - this since
SQLA 0.7 no longer enforces "parent with no
child" at the ORM level; this check is left
up to foreign key nullability.
Related to [ticket:1912]
- Repaired new "mutable" extension to propagate
events to subclasses correctly; don't
create multiple event listeners for
subclasses either. [ticket:2180]
- Modify the text of the message which occurs
when the "identity" key isn't detected on
flush, to include the common cause that
the Column isn't set up to detect
auto-increment correctly; [ticket:2170].
Also in 0.6.8.
- Fixed bug where transaction-level "deleted"
collection wouldn't be cleared of expunged
states, raising an error if they later
became transient [ticket:2182].
Also in 0.6.8.
- sql
- Fixed bug whereby metadata.reflect(bind)
would close a Connection passed as a
bind argument. Regression from 0.6.
- Streamlined the process by which a Select
determines what's in it's '.c' collection.
Behaves identically, except that a
raw ClauseList() passed to select([])
(which is not a documented case anyway) will
now be expanded into its individual column
elements instead of being ignored.
- engine
- Deprecate schema/SQL-oriented methods on
Connection/Engine that were never well known
and are redundant: reflecttable(), create(),
drop(), text(), engine.func
- Adjusted the __contains__() method of
a RowProxy result row such that no exception
throw is generated internally;
NoSuchColumnError() also will generate its
message regardless of whether or not the column
construct can be coerced to a string.
[ticket:2178]. Also in 0.6.8.
- sqlite
- Accept None from cursor.fetchone() when
"PRAGMA read_uncommitted" is called to determine
current isolation mode at connect time and
default to SERIALIZABLE; this to support SQLite