Failing to do so was causing the issue reported in ticket #103. The issue
as reported was fixed when SET ISOLATION LEVEL was dropped, but the real
problem wasn't fixed.
This basically removes the READ UNCOMMITED level (that internally
PostgreSQL maps to READ COMMITED anyway) to keep the numeric values
compattible with old psycopg versions. For full details and discussion
see this thread:
http://archives.postgresql.org/psycopg/2011-12/msg00008.php
On windows it returns -1 instead of sometihing portable. So just ditch
the static buffer and just use a dynamic one to compose the command.
Also squashed a couple of buglets in copy_to: copyfile was decremented
before being set to null, size_t was used instead of Py_ssize_t.
In fact it doesn't change "the transaction", as there has to be no
transaction when invoked. The effect instead is to execute SET SESSION
CHARACTERISTICS.
The MSVC compiler does not have the strcasecmp(x, y) function, which is a
case insensitve string compare function. Instead, MSVC has a similar function,
lstrcmpi(x, y). Modified config.h to use this function when building with
MSVC.
Functions conn_setup(), conn_get_isolation_level(), conn_set_transaction(),
conn_switch_isolation_level(), conn_set_client_encoding() reimplemented
using the pqpath funtitons.
Dropped analogous function in the connection, as it had to take the lock,
thus it was hard to build consistent pieces of functionality with it.
The aim of these function is to allow the connection to make a better use
of the pqpath functions instead of using PQexec for these small things.
Also, the functions are to be called with the connection lock: this makes
composing higher level functions using them easier.
The function is always called in the context of functions grabbing the
connection lock, so just use the same critical section instead of releasing
and re-acquiring it. It is not a problem as serious as the notifies process
(ticket #55) as the notices are a psycopg structure, not libpq. However the
change allows again processing notices/notifies in the same place,
which makes sense conceptually, plus we save some lock dance.
Notifies process access the connection, is not limited to the result, so
There is the possibility of loss of protocol sync in multithread programs.
Closes ticket #55.
Don't issue a SET TRANSACTION ISOLATION LEVEL at every begin: use PG's
GUC default, eventually set by set_transaction.
Dropped the last query at connection, yay!
Method set_isolation_level() and property isolation_level refactored using
the new structures, keeping the previous semantic.
The encoding can be set by PGCLIENTENCODING, which may be an alternative
spelling. Bug reported by Peter Eisentraut.
At this point the idea of considering one of the random spellings such as
EUC_CN as somewhat "blessed" is debunked. So just store the cleaned-up
version of the encoding in the mapping table. Note that the cleaned-up
version was needed by the unicode adapter: this requirement has been
surpassed as the connection now contains a copy of the Python codec name
set whenever the client encoding is set.
PG 9.0 uses the hex format by default, and clients < 9.0 can't parse that
format, requiring client update and great care in what is linked at runtime,
and generally giving headache to users and transitively us.
- Check return value of PyErr_Malloc and set an exception in case of error
- Avoid exposing variables with refcount 0 as connection attributes.
- PyErr_Free guards itself for NULL input
It has long been used in wrong ways, with the function receiving a
connection or lobject instead of a cursor. It has always been unnoticed
(nobody has noticed the wrong object attached to the exception in the
wrong cases) but it started crashing the interpreter with Python 3.2 on
Windows.
Thanks to Jason Erickson for finding the problem and helping fixing it.
Empty array can be returned untyped by postgres. To handle
this case, a special handler is added for the type UNKNOWNOID.
If the value return by the database is strictly equal to "{}",
the value is converted. Otherwise, the conversion fallback on
the default handler.
Python 3.2 hash() function will now return a 64bit value when run on a 64bit
architecture, where as previously, it would always return a 32bit value.
Modified the code to use the now Py_hash_t typedef and for Python versions
less than 3.2, hard code Py_hash_t to long and Py_uhash_t to unsigned long.
- Raise an exception on incomplete placeholders.
- Minor speedups.
- Don't change the string in place (??!!) if the placeholder is not s
and the value is null.
The latter point can be done because downstream we don't accept anything
different from s anyway (in the Bytes_Format function).
Notice that now the format string is constant whatever the arguments.
This means that executemany is still more inefficient than it should be
as mogrify may work only on the parameters. However this is an
implementation only worthwhile if we start supporting real parameters.
Let's talk about that for the next release.
The value is used to control the number of records to fetch per network
roundtrip in named cursors iteration. Used to avoid the inefficient
arraysize default of 1 without giving this value the magic meaning of
2000.
Supporting this interface is required to adapt memoryview on Python 2.7 as they
don't support the old style. But because the old style is long deprecated it
makes sense to start supporting the new one.
The feature in itself is not extremely useful and instead PostgreSQL is
not always able to cast away from text[], which is a regression see
(ticket #42).
