]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/log
thirdparty/sqlalchemy/sqlalchemy.git
15 years agoDisabling tests for mssql+mxodbc where mxODBC cursor.execute chokes on invalid placem...
Brad Allen [Tue, 16 Mar 2010 22:57:54 +0000 (16:57 -0600)] 
Disabling tests for mssql+mxodbc where mxODBC cursor.execute chokes on invalid placement of bind parameter "?" within the SQL statement.

15 years agoFixes to pass numeric tests; now by default, the mxodbc connector natively returns...
Brad Allen [Tue, 16 Mar 2010 22:53:21 +0000 (16:53 -0600)] 
Fixes to pass numeric tests; now by default, the mxodbc connector natively returns Python Decimal data types from columns of type SQL.NUMERIC or SQL.DECIMAL

15 years agoAdded module docstring for mxodbc connector
Brad Allen [Tue, 16 Mar 2010 22:50:20 +0000 (16:50 -0600)] 
Added module docstring for mxodbc connector

15 years agoremoved junk pdb.set_trace()
Brad Allen [Tue, 16 Mar 2010 02:35:14 +0000 (20:35 -0600)] 
removed junk pdb.set_trace()

15 years agoRemoved custom do_execute which called cursor.executedirect, after discussion with...
Brad Allen [Tue, 16 Mar 2010 02:34:34 +0000 (20:34 -0600)] 
Removed custom do_execute which called cursor.executedirect, after discussion with eGenix and zzzeek. Even though more tests pass with executedirect, it is not appropriate for default usage because it does not issue prepare statements to the db, and it forces Python type binding. The executedirect method should only be called when the API user specifically requests it for special case performance reasons.

15 years agomerged mainline default branch
Brad Allen [Mon, 15 Mar 2010 17:42:35 +0000 (11:42 -0600)] 
merged mainline default branch

15 years agousing executemany instead of executedirect, which passes more tests than with plain...
Brad Allen [Mon, 15 Mar 2010 17:39:03 +0000 (11:39 -0600)] 
using executemany instead of executedirect, which passes more tests than with plain execute. Later we need to define logic to determine when to call the normal execute.

15 years ago- The visit_pool() method of Dialect is removed, and replaced with
Mike Bayer [Mon, 15 Mar 2010 17:08:31 +0000 (13:08 -0400)] 
- The visit_pool() method of Dialect is removed, and replaced with
on_connect().  This method returns a callable which receives
the raw DBAPI connection after each one is created.   The callable
is assembled into a first_connect/connect pool listener by the
connection strategy if non-None.   Provides a simpler interface
for dialects.

15 years agowhitespace fixes to improve readability (shortened excessively long lines, and one...
Brad Allen [Mon, 15 Mar 2010 17:01:50 +0000 (11:01 -0600)] 
whitespace fixes to improve readability (shortened excessively long lines, and one 5-space indentation block corrected to 4 spaces).

15 years agofix failing test due to sybase paramstyle
Mike Bayer [Sun, 14 Mar 2010 23:50:50 +0000 (19:50 -0400)] 
fix failing test due to sybase paramstyle

15 years ago- name all the "sub" dialect components <DB><component>_<dialectname>, [ticket:1738]
Mike Bayer [Sun, 14 Mar 2010 23:31:30 +0000 (19:31 -0400)] 
- name all the "sub" dialect components <DB><component>_<dialectname>, [ticket:1738]

15 years ago- many incantations to get the tests to run reasonably
Mike Bayer [Sun, 14 Mar 2010 23:03:24 +0000 (23:03 +0000)] 
- many incantations to get the tests to run reasonably
- executemany() for some reason uses some tiny buffer, overriding it
- we need to use the IDENTITY_INSERT thing

15 years ago- initial working version of sybase, with modifications to the transactional
Mike Bayer [Sun, 14 Mar 2010 22:04:20 +0000 (22:04 +0000)] 
- initial working version of sybase, with modifications to the transactional
model to accomodate Sybase's default mode of "no ddl in transactions".
- identity insert not working yet.  it seems the default here might be the
opposite of that of MSSQL.
- reflection will be a full rewrite
- default DBAPI is python-sybase, well documented and nicely DBAPI compliant
except for the bind parameter situation, where we have a straightforward workaround
- full Sybase docs at: http://infocenter.sybase.com/help/index.jsp?topic=/com.sybase.help.ase_15.0/title.htm

15 years agostarting to add notes for sybase
Mike Bayer [Sun, 14 Mar 2010 19:00:27 +0000 (15:00 -0400)] 
starting to add notes for sybase

15 years agomerge of the "database chart" diff
Mike Bayer [Sun, 14 Mar 2010 01:06:07 +0000 (20:06 -0500)] 
merge of the "database chart" diff

15 years ago- Fixed bug in 0.6-reworked "many-to-one" optimizations
Mike Bayer [Sun, 14 Mar 2010 01:05:17 +0000 (20:05 -0500)] 
- Fixed bug in 0.6-reworked "many-to-one" optimizations
such that a many-to-one that is against a non-primary key
column on the remote table (i.e. foreign key against a
UNIQUE column) will pull the "old" value in from the
database during a change, since if it's in the session
we will need it for proper history/backref accounting,
and we can't pull from the local identity map on a
non-primary key column. [ticket:1737] (test is pending)

15 years agoongoing typos
Mike Bayer [Sat, 13 Mar 2010 23:17:26 +0000 (18:17 -0500)] 
ongoing typos

15 years agogr and pysqlite
Mike Bayer [Sat, 13 Mar 2010 23:13:41 +0000 (18:13 -0500)] 
gr and pysqlite

15 years agoupdate link for sqlite3
Mike Bayer [Sat, 13 Mar 2010 23:10:31 +0000 (18:10 -0500)] 
update link for sqlite3

15 years agothe big table of all the DBAPIs
Mike Bayer [Sat, 13 Mar 2010 23:04:39 +0000 (18:04 -0500)] 
the big table of all the DBAPIs

15 years agoworking dialect support into a chart
Mike Bayer [Sat, 13 Mar 2010 22:20:37 +0000 (17:20 -0500)] 
working dialect support into a chart

15 years ago- further checks when running in WITH_UNICODE
Mike Bayer [Sat, 13 Mar 2010 21:12:29 +0000 (16:12 -0500)] 
- further checks when running in WITH_UNICODE
- When WITH_UNICODE
mode is used in Python 2.xx, a large and scary warning
is emitted asking that the user seriously consider
the usage of this difficult mode of operation.
[ticket:1670]

15 years agocallcount...
Mike Bayer [Sat, 13 Mar 2010 19:54:41 +0000 (14:54 -0500)] 
callcount...

15 years agoarg, oracle, bah
Mike Bayer [Sat, 13 Mar 2010 19:53:38 +0000 (14:53 -0500)] 
arg, oracle, bah

15 years agoensure with_unicode flag is false for cx_oracle < 5
Mike Bayer [Sat, 13 Mar 2010 18:56:05 +0000 (13:56 -0500)] 
ensure with_unicode flag is false for cx_oracle < 5

15 years agomerge tip
Mike Bayer [Sat, 13 Mar 2010 18:54:24 +0000 (13:54 -0500)] 
merge tip

15 years ago- Added "logging_name" argument to create_engine(), Pool() constructor
Mike Bayer [Sat, 13 Mar 2010 18:53:31 +0000 (13:53 -0500)] 
- Added "logging_name" argument to create_engine(), Pool() constructor
as well as "pool_logging_name" argument to create_engine() which
filters down to that of Pool.   Issues the given string name
within the "name" field of logging messages instead of the default
hex identifier string.  [ticket:1555]

15 years agoadjust per a couple zxjdbc failures
Philip Jenvey [Sat, 13 Mar 2010 18:29:09 +0000 (10:29 -0800)] 
adjust per a couple zxjdbc failures

15 years agoonly enable WITH_UNICODE when >= cx_Oracle v5
Philip Jenvey [Sat, 13 Mar 2010 18:20:57 +0000 (10:20 -0800)] 
only enable WITH_UNICODE when >= cx_Oracle v5

15 years ago- query.get() now returns None if queried for an identifier
Mike Bayer [Sat, 13 Mar 2010 17:28:50 +0000 (12:28 -0500)] 
- query.get() now returns None if queried for an identifier
that is present in the identity map with a different class
than the one requested, i.e. when using polymorphic loading.
[ticket:1727]

15 years agomerge tip
Mike Bayer [Sat, 13 Mar 2010 16:52:50 +0000 (11:52 -0500)] 
merge tip

15 years agofix the timeout test once and for all
Mike Bayer [Sat, 13 Mar 2010 16:52:17 +0000 (11:52 -0500)] 
fix the timeout test once and for all

15 years agotweak to take advantage of returning support in executemany
Philip Jenvey [Sat, 13 Mar 2010 03:01:59 +0000 (19:01 -0800)] 
tweak to take advantage of returning support in executemany

15 years agodisable interval tests on oracle+zxjdbc for now, and also test_fk_nonpassive
Philip Jenvey [Sat, 13 Mar 2010 01:50:33 +0000 (17:50 -0800)] 
disable interval tests on oracle+zxjdbc for now, and also test_fk_nonpassive

15 years agofix extending of a tuple
Philip Jenvey [Sat, 13 Mar 2010 01:19:07 +0000 (17:19 -0800)] 
fix extending of a tuple

15 years agomove the initial connect out of the profile zone so we don't get a method call hit...
Mike Bayer [Sat, 13 Mar 2010 00:57:06 +0000 (19:57 -0500)] 
move the initial connect out of the profile zone so we don't get a method call hit on that

15 years agofurther refinement
Mike Bayer [Fri, 12 Mar 2010 23:45:53 +0000 (23:45 +0000)] 
further refinement

15 years agomerging master...
Mike Bayer [Fri, 12 Mar 2010 23:23:44 +0000 (23:23 +0000)] 
merging master...

15 years agoconsolidate unicode/CLOB handling
Mike Bayer [Fri, 12 Mar 2010 23:22:20 +0000 (23:22 +0000)] 
consolidate unicode/CLOB handling

15 years agoadjust to allow mock DBAPIs
Mike Bayer [Fri, 12 Mar 2010 21:49:32 +0000 (16:49 -0500)] 
adjust to allow mock DBAPIs

15 years agooursql doesn't like the cursor being reused without fetch being finished
Mike Bayer [Fri, 12 Mar 2010 21:19:41 +0000 (16:19 -0500)] 
oursql doesn't like the cursor being reused without fetch being finished

15 years agomerge from oracle WITH_UNICODE branch
Mike Bayer [Fri, 12 Mar 2010 21:07:03 +0000 (21:07 +0000)] 
merge from oracle WITH_UNICODE branch

15 years ago- Added preliminary support for Oracle's WITH_UNICODE
Mike Bayer [Fri, 12 Mar 2010 21:05:53 +0000 (21:05 +0000)] 
- Added preliminary support for Oracle's WITH_UNICODE
mode.  At the very least this establishes initial
support for cx_Oracle with Python 3.
[ticket:1670]

15 years agoMerged latest changes from the mainline (default branch)
Brad Allen [Fri, 12 Mar 2010 19:09:31 +0000 (13:09 -0600)] 
Merged latest changes from the mainline (default branch)

15 years agothese pass on +zxjdbc, seemingly because it lacks a sane rowcount
Philip Jenvey [Fri, 12 Mar 2010 19:07:53 +0000 (11:07 -0800)] 
these pass on +zxjdbc, seemingly because it lacks a sane rowcount

15 years ago- introduce an optimizing type _NativeUnicodeMixin to oracle plus supporting
Mike Bayer [Fri, 12 Mar 2010 18:30:38 +0000 (18:30 +0000)] 
- introduce an optimizing type _NativeUnicodeMixin to oracle plus supporting
changes to Enum/SchemaType to re-support adaptation of string types.
This approach can be adapted by "conditional" unicode returning dialects
(i.e. pyodbc and possibly mxodbc) to remove the overhead
of isinstance(value, unicode) calls when the dialect returned type is
of dbapi.UNICODE, dbapi.NVARCHAR, etc.

15 years agothink I came across a syntactical weirdism there
Mike Bayer [Fri, 12 Mar 2010 03:38:15 +0000 (22:38 -0500)] 
think I came across a syntactical weirdism there

15 years agorefine that a bit
Mike Bayer [Fri, 12 Mar 2010 03:12:49 +0000 (22:12 -0500)] 
refine that a bit

15 years agoadding windows-appropriate levels of pain to unicode handling
Mike Bayer [Fri, 12 Mar 2010 03:10:25 +0000 (22:10 -0500)] 
adding windows-appropriate levels of pain to unicode handling

15 years agomerge tds changes
Mike Bayer [Fri, 12 Mar 2010 02:55:09 +0000 (21:55 -0500)] 
merge tds changes

15 years agoadd a third state to converts_unicode_strings - "conditional". at the moment
Mike Bayer [Fri, 12 Mar 2010 02:52:54 +0000 (21:52 -0500)] 
add a third state to converts_unicode_strings - "conditional".  at the moment
this will have us do a check.  i.e. for MSSQL where NVARCHAR is unicode and VARCHAR is not.

15 years agoAdded a non-freetds specific exclusion for pyodbc
Mike Bayer [Fri, 12 Mar 2010 02:23:25 +0000 (21:23 -0500)] 
Added a non-freetds specific exclusion for pyodbc

15 years agoformatting
Mike Bayer [Thu, 11 Mar 2010 23:46:21 +0000 (18:46 -0500)] 
formatting

15 years ago- query.scalar() now raises an exception if more than one
Mike Bayer [Thu, 11 Mar 2010 23:38:51 +0000 (18:38 -0500)] 
- query.scalar() now raises an exception if more than one
row is returned.  All other behavior remains the same.
[ticket:1735]

15 years ago- Calling query.order_by() or query.distinct() before calling
Mike Bayer [Thu, 11 Mar 2010 23:04:57 +0000 (18:04 -0500)] 
- Calling query.order_by() or query.distinct() before calling
query.select_from(), query.with_polymorphic(), or
query.from_statement() raises an exception now instead of
silently dropping those criterion. [ticket:1736]

15 years agoAdded decorators to suppress unwanted warning messages during testing (not working...
Brad Allen [Thu, 11 Mar 2010 22:33:05 +0000 (16:33 -0600)] 
Added decorators to suppress unwanted warning messages during testing (not working yet, but committed for review purposes).

15 years agoNow using pyodbc execution context once again (to pass the test involving insert...
Brad Allen [Thu, 11 Mar 2010 22:30:49 +0000 (16:30 -0600)] 
Now using pyodbc execution context once again (to pass the test involving insert to trigger table)

15 years agocomments explaining temporary hack for using executedirect
Brad Allen [Thu, 11 Mar 2010 22:26:32 +0000 (16:26 -0600)] 
comments explaining temporary hack for using executedirect

15 years agoConverted mxODBC's raised Warning exceptions into normal Python warnings.
Brad Allen [Thu, 11 Mar 2010 22:24:06 +0000 (16:24 -0600)] 
Converted mxODBC's raised Warning exceptions into normal Python warnings.

15 years agothis resolves the returning failures from #1733
Mike Bayer [Thu, 11 Mar 2010 21:36:03 +0000 (21:36 +0000)] 
this resolves the returning failures from #1733

15 years agore-enable these tests
Mike Bayer [Thu, 11 Mar 2010 21:34:13 +0000 (21:34 +0000)] 
re-enable these tests

15 years agoadded better typing for integer expressions, since integer is implementing _DateAffinity
Mike Bayer [Thu, 11 Mar 2010 21:18:37 +0000 (16:18 -0500)] 
added better typing for integer expressions, since integer is implementing _DateAffinity

15 years agoforce an oracle test to pass. this is as usual an extremely fringy test for a featur...
Mike Bayer [Thu, 11 Mar 2010 18:52:02 +0000 (18:52 +0000)] 
force an oracle test to pass.  this is as usual an extremely fringy test for a feature nobody should be using normally.

15 years agothese fail for unknown reasons. Seems to be potentially VARCHAR used with RETURNING...
Mike Bayer [Thu, 11 Mar 2010 18:39:15 +0000 (18:39 +0000)] 
these fail for unknown reasons.  Seems to be potentially VARCHAR used with RETURNING, or out params, not sure.

15 years agofix this test for oracle so all types return datetime.date
Mike Bayer [Thu, 11 Mar 2010 18:34:27 +0000 (18:34 +0000)] 
fix this test for oracle so all types return datetime.date

15 years agothis test requires a length for the varchar in mysql/oracle
Mike Bayer [Thu, 11 Mar 2010 18:21:44 +0000 (18:21 +0000)] 
this test requires a length for the varchar in mysql/oracle

15 years ago- fix the sqlalchemy.test.schema.Column function to work with copies
Mike Bayer [Thu, 11 Mar 2010 18:01:04 +0000 (18:01 +0000)] 
- fix the sqlalchemy.test.schema.Column function to work with copies
- add sequences to new associationproxy tests
- test/ext passes 100% on oracle here

15 years agomerged default tip
Mike Bayer [Thu, 11 Mar 2010 17:43:55 +0000 (17:43 +0000)] 
merged default tip

15 years agoadd autoincrement flags to decls
Mike Bayer [Thu, 11 Mar 2010 17:43:34 +0000 (17:43 +0000)] 
add autoincrement flags to decls

15 years ago- the copy() method of Column now copies over uninitialized
Mike Bayer [Thu, 11 Mar 2010 17:42:06 +0000 (12:42 -0500)] 
- the copy() method of Column now copies over uninitialized
"on table attach" events.  Helps with the new declarative
"mixin" capability.

15 years ago- mark tests unsupported by sqlite unsupported by oracle as well
Mike Bayer [Thu, 11 Mar 2010 17:22:49 +0000 (17:22 +0000)] 
- mark tests unsupported by sqlite unsupported by oracle as well

15 years ago- fixed internal error which would occur if calling has()
Mike Bayer [Thu, 11 Mar 2010 17:07:08 +0000 (12:07 -0500)] 
- fixed internal error which would occur if calling has()
or similar complex expression on a single-table inheritance
relation(). [ticket:1731]

15 years agomerged latest commits from default branch on server
Brad Allen [Thu, 11 Mar 2010 16:29:40 +0000 (10:29 -0600)] 
merged latest commits from default branch on server

15 years ago- Oracle 'DATE' now does not perform any result processing,
Mike Bayer [Thu, 11 Mar 2010 15:27:18 +0000 (10:27 -0500)] 
- Oracle 'DATE' now does not perform any result processing,
as the DATE type in Oracle stores full date+time objects,
that's what you'll get.  Note that the generic types.Date
type *will* still call value.date() on incoming values,
however.  When reflecting a table, the reflected type
will be 'DATE'.

15 years ago- emphasized query.join() in ORM tutorial as per [ticket:1708]
Mike Bayer [Wed, 10 Mar 2010 23:29:32 +0000 (18:29 -0500)] 
- emphasized query.join() in ORM tutorial as per [ticket:1708]
- cleaned up tutorial w.r.t. eagerload, added a section for contains_eager as this function
is equally important
- added better linkages in sqlalchemy.orm reference documentation, updated antiquated
docs for contains_eager(), got aliased()/AliasedClass documented as well as Sphinx will
allow us

15 years agofixed up docs for execution_options() across all three locations.
Mike Bayer [Tue, 9 Mar 2010 23:45:30 +0000 (18:45 -0500)] 
fixed up docs for execution_options() across all three locations.

15 years ago- moved the logic within _BindParamClause which decides about coercing the bind's...
Mike Bayer [Tue, 9 Mar 2010 22:46:00 +0000 (17:46 -0500)] 
- moved the logic within _BindParamClause which decides about coercing the bind's type
based on the compared type into AbstractType.   The new method is called _coerce_compared_value()
and receives the operator and the raw python value to be coerced.    TypeDecorator
overrides this to provude the "old" 0.5 behavior of coercing the other side of the expression
unconditonally.
- added docs to TypeDecorator attempting to explain this though they are a little verbose.
- added caveats to Interval, which in "non-native" mode can't really handle proper expression behavior.  the "typing" would have to move into the compiler, such that on a DB like SQLite we're actually wrapping left/right side into its epoch functions, etc., its a big job.  current infrastructure, as powerful as it is, still isn't at that level.

15 years agomeaningless change
Mike Bayer [Tue, 9 Mar 2010 21:47:45 +0000 (16:47 -0500)] 
meaningless change

15 years agobase tests pass
Mike Bayer [Tue, 9 Mar 2010 19:32:38 +0000 (14:32 -0500)] 
base tests pass

15 years agoalmost there, some tests failing still.
Mike Bayer [Tue, 9 Mar 2010 18:41:43 +0000 (13:41 -0500)] 
almost there, some tests failing still.

15 years agoImproved the docstring of test_fetchid_trigger
Brad Allen [Tue, 9 Mar 2010 18:30:55 +0000 (12:30 -0600)] 
Improved the docstring of test_fetchid_trigger

15 years agomerge latest tip. practicing on mercurial.
Mike Bayer [Tue, 9 Mar 2010 04:54:30 +0000 (23:54 -0500)] 
merge latest tip.  practicing on mercurial.

15 years agostill thinking about stuff here
Mike Bayer [Tue, 9 Mar 2010 02:15:16 +0000 (21:15 -0500)] 
still thinking about stuff here

15 years agoAdded docstring & comments explaining test_fetchid_trigger
Brad Allen [Tue, 9 Mar 2010 00:19:25 +0000 (18:19 -0600)] 
Added docstring & comments explaining test_fetchid_trigger

15 years agoworking on getting operators/left hand type awareness into the "bind" coercion. ...
Mike Bayer [Mon, 8 Mar 2010 23:27:35 +0000 (18:27 -0500)] 
working on getting operators/left hand type awareness into the "bind" coercion.  this system has to be figured out somehow

15 years agoTemporary kludge to fix InterfaceError, in places where cursor.executedirect is neede...
Brad Allen [Mon, 8 Mar 2010 21:57:18 +0000 (15:57 -0600)] 
Temporary kludge to fix InterfaceError, in places where cursor.executedirect is needed instead of cursor.execute.

15 years agofix last commit
Philip Jenvey [Mon, 8 Mar 2010 21:04:32 +0000 (13:04 -0800)] 
fix last commit

15 years agosmall cleanup
Philip Jenvey [Mon, 8 Mar 2010 20:22:33 +0000 (12:22 -0800)] 
small cleanup

15 years agoRemoved pyodbc execution context, because output inserted is now handled by the base...
Brad Allen [Mon, 8 Mar 2010 19:53:19 +0000 (13:53 -0600)] 
Removed pyodbc execution context, because output inserted is now handled by the base. No more need for SELECT SCOPE_IDENTITY().

15 years agoRemove isnew argument from row_processors. It wasn't used anyway in most cases.
Gaëtan de Menten [Mon, 8 Mar 2010 16:02:56 +0000 (17:02 +0100)] 
Remove isnew argument from row_processors. It wasn't used anyway in most cases.

15 years agodistutils choked on the trailing slash on the lib directory in the manifest.
Brad Allen [Mon, 8 Mar 2010 02:43:11 +0000 (20:43 -0600)] 
distutils choked on the trailing slash on the lib directory in the manifest.

15 years agothe mercurial extension seems to not work in py3k and also I don't
Mike Bayer [Mon, 8 Mar 2010 01:56:32 +0000 (20:56 -0500)] 
the mercurial extension seems to not work in py3k and also I don't
like its usage during build/install.   going with the manual approach
for the moment.

15 years agouse setuptools_hg to auto-discover the same way we did with SVN.
Mike Bayer [Mon, 8 Mar 2010 00:29:39 +0000 (19:29 -0500)] 
use setuptools_hg to auto-discover the same way we did with SVN.
Thanks *much* to Jannis Leidel for the suggestion/creation !

15 years ago- plug a minor ORM speed hit in Events (**kwargs).
Gaëtan de Menten [Sun, 7 Mar 2010 19:45:39 +0000 (20:45 +0100)] 
- plug a minor ORM speed hit in Events (**kwargs).
- added comment explaining some strange code

15 years agoCleaned up some long lines to test the commit hooks.
Michael Trier [Sun, 7 Mar 2010 15:56:11 +0000 (10:56 -0500)] 
Cleaned up some long lines to test the commit hooks.

15 years ago- cleanup some long lines. (test hg commit hook)
Mike Bayer [Sun, 7 Mar 2010 06:56:25 +0000 (01:56 -0500)] 
- cleanup some long lines.  (test hg commit hook)

15 years agoupdate tags
convert-repo [Sat, 6 Mar 2010 21:24:12 +0000 (21:24 +0000)] 
update tags

15 years agouse a flag to build cextensions (thanks again to Genshi for the tip)
Mike Bayer [Fri, 5 Mar 2010 15:33:21 +0000 (15:33 +0000)] 
use a flag to build cextensions (thanks again to Genshi for the tip)

15 years agotypo
Mike Bayer [Fri, 5 Mar 2010 14:52:57 +0000 (14:52 +0000)] 
typo

15 years agotimings should use getitem syntax and not the slow getattr one
Gaëtan de Menten [Wed, 3 Mar 2010 20:40:39 +0000 (20:40 +0000)] 
timings should use getitem syntax and not the slow getattr one