| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Various performance improvements to Row instanciation
- avoid passing processors if they are all None
- improve processor logic in cython
- improve tuplegetter using slices when contiguous indexes are used
Some timing follow.
In particular [base_]row_new_proc that tests using processors has
a 25% improvement compared to before in cython.
Looking at the [b]row_new_proc_none that test a list of processors
all None, this has 50% improvement in cython when passing the none list,
but in this patch it would usually be disabled by passing None, so the
performance gain is actually 90%, since it would run the case
[base_]row_new.
Tuplegetter is a bit faster in the single item get and when getting
sequential indexes (like indexes 1,2,3,4) at the cost of a bit
longer creation time in python, cython is mostly the same.
Current times
| python | cython | cy / py |
base_row_new | 0.639817400 | 0.118265500 | 0.184842582 |
row_new | 0.680355100 | 0.129714600 | 0.190657202 |
base_row_new_proc | 3.076538900 | 1.488428600 | 0.483799701 |
row_new_proc | 3.119700100 | 1.532197500 | 0.491136151 |
brow_new_proc_none | 1.917702300 | 0.475511500 | 0.247958977 |
row_new_proc_none | 1.956253300 | 0.497803100 | 0.254467609 |
tuplegetter_one | 0.152512600 | 0.148523900 | 0.973846751 |
tuplegetter_many | 0.184394100 | 0.184511500 | 1.000636680 |
tuplegetter_seq | 0.154832800 | 0.156270100 | 1.009282917 |
tuplegetter_new_one | 0.523730000 | 0.343402200 | 0.655685563 |
tuplegetter_new_many| 0.738924400 | 0.420961400 | 0.569694816 |
tuplegetter_new_seq | 1.062036900 | 0.495462000 | 0.466520514 |
Parent commit times
| python | cython | cy / py |
base_row_new | 0.643890800 | 0.113548300 | 0.176347138 |
row_new | 0.674885900 | 0.124391800 | 0.184315304 |
base_row_new_proc | 3.072020400 | 2.017367000 | 0.656690626 |
row_new_proc | 3.109943400 | 2.048359400 | 0.658648450 |
brow_new_proc_none | 1.967133700 | 1.006326000 | 0.511569702 |
row_new_proc_none | 1.960814900 | 1.025217800 | 0.522852922 |
tuplegetter_one | 0.197359900 | 0.205999000 | 1.043773330 |
tuplegetter_many | 0.196575900 | 0.194888500 | 0.991416038 |
tuplegetter_seq | 0.192723900 | 0.205635000 | 1.066992729 |
tuplegetter_new_one | 0.534644500 | 0.414311700 | 0.774929322 |
tuplegetter_new_many| 0.479376500 | 0.417448100 | 0.870814694 |
tuplegetter_new_seq | 0.481580200 | 0.412697900 | 0.856966088 |
Change-Id: I2ca1f49dca2beff625c283f1363c29c8ccc0c3f7
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Improved :class:`_engine.Row` implementation to optimize
``__getattr__`` performance.
The serialization of a :class:`_engine.Row` to pickle has changed with
this change. Pickle saved by older SQLAlchemy versions can still be loaded,
but new pickle saved by this version cannot be loaded by older ones.
Fixes: #9678
Closes: #9668
Pull-request: https://github.com/sqlalchemy/sqlalchemy/pull/9668
Pull-request-sha: 86b8ccd1959dbd91b1208f7a648a91f217e1f866
Change-Id: Ia85c26a59e1a57ba2bf0d65578c6168f82a559f2
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fixed regression involving pickling of Python rows between the cython and
pure Python implementations of :class:`.Row`, which occurred as part of
refactoring code for version 2.0 with typing. A particular constant were
turned into a string based ``Enum`` for the pure Python version of
:class:`.Row` whereas the cython version continued to use an integer
constant, leading to deserialization failures.
Regression occurred in a4bb502cf95ea3523e4d383c4377e50f402d7d52
Fixes: #9423
Change-Id: Icbd85cacb2d589cef7c246de7064249926146f2e
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
All modules in sqlalchemy.engine are strictly
typed with the exception of cursor, default, and
reflection. cursor and default pass with non-strict
typing, reflection is waiting on the multi-reflection
refactor.
Behavioral changes:
* create_connect_args() methods return a tuple of list,
dict, rather than a list of list, dict
* removed allow_chars parameter from
pyodbc connector ._get_server_version_info()
method
* the parameter list passed to do_executemany is now
a list in all cases. previously, this was being run
through dialect.execute_sequence_format, which
defaults to tuple and was only intended for individual
tuple params.
* broke up dialect.dbapi into dialect.import_dbapi
class method and dialect.dbapi module object. added
a deprecation path for legacy dialects. it's not
really feasible to type a single attr as a classmethod
vs. module type. The "type_compiler" attribute also
has this problem with greater ability to work around,
left that one for now.
* lots of constants changing to be Enum, so that we can
type them. for fixed tuple-position constants in
cursor.py / compiler.py (which are used to avoid the
speed overhead of namedtuple), using Literal[value]
which seems to work well
* some tightening up in Row regarding __getitem__, which
we can do since we are on full 2.0 style result use
* altered the set_connection_execution_options and
set_engine_execution_options event flows so that the
dictionary of options may be mutated within the event
hook, where it will then take effect as the actual
options used. Previously, changing the dict would
be silently ignored which seems counter-intuitive
and not very useful.
* A lot of DefaultDialect/DefaultExecutionContext
methods and attributes, including underscored ones, move
to interfaces. This is not fully ideal as it means
the Dialect/ExecutionContext interfaces aren't publicly
subclassable directly, but their current purpose
is more of documentation for dialect authors who should
(and certainly are) still be subclassing the DefaultXYZ
versions in all cases
Overall, Result was the most extremely difficult class
hierarchy to type here as this hierarchy passes through
largely amorphous "row" datatypes throughout, which
can in fact by all kinds of different things, like
raw DBAPI rows, or Row objects, or "scalar"/Any, but
at the same time these types have meaning so I tried still
maintaining some level of semantic markings for these,
it highlights how complex Result is now, as it's trying
to be extremely efficient and inlined while also being
very open-ended and extensible.
Change-Id: I98b75c0c09eab5355fc7a33ba41dd9874274f12a
|
| |
|
|
|
|
|
|
|
|
| |
__future__.annotations mode allows us to use non-string
annotations for argument and return types in most cases,
but more importantly it removes a large amount of runtime
overhead that would be spent in evaluating the annotations.
Change-Id: I2f5b6126fe0019713fc50001be3627b664019ede
References: #6810
|
|
|
Re-implement c version immutabledict / processors / resultproxy / utils with cython.
Performance is in general in par or better than the c version
Added a collection module that has cython version of OrderedSet and IdentitySet
Added a new test/perf file to compare the implementations.
Run ``python test/perf/compiled_extensions.py all`` to execute the comparison test.
See results here: https://docs.google.com/document/d/1nOcDGojHRtXEkuy4vNXcW_XOJd9gqKhSeALGG3kYr6A/edit?usp=sharing
Fixes: #7256
Change-Id: I2930ef1894b5048210384728118e586e813f6a76
Signed-off-by: Federico Caselli <cfederico87@gmail.com>
|
