Qué hay de nuevo en Python 2.7

Autor

A.M. Kuchling (amk en amk.ca)

Este articulo explica las nuevas características en Python 2.7. Python 2.7 fue publicado el 3 de Julio de 2010.

El manejo numérico ha sido mejorado en muchas formas, tanto para los números de punto flotante como para la clase Decimal. Hay algunas adiciones útiles a la biblioteca estándar, como una gran mejora al modulo unittest, el modulo argparse para analizar las opciones de la línea de comandos, adecuando las clases OrderedDict y Counter en el modulo collections, y muchas otras mejoras.

Python 2.7 esta previsto que sea el ultimo lanzamiento de la serie 2.x, así que hemos trabajado para que sea una versión a largo plazo. Para ayudar con la migración a Python 3, se han incluido varias características nuevas de la serie Python 3.x en 2.7.

Este articulo no intenta proporcionar una especificación completa de las nuevas características, sino que proporciona una visión general conveniente. Para obtener mas información, debería consultar la documentación de Python 2.7 en https://docs.python.org. Si desea comprender la justificación para el diseño y la implementación, consultar el PEP para una nueva característica particular o el error en el que se discutió un cambio en https://bugs.python.org. Siempre que sea posible, «Que hay de nuevo en Python» enlaza el elemento error/parche para cada cambio.

El futuro de Python 2.x

Python 2.7 is the last major release in the 2.x series, as the Python maintainers have shifted the focus of their new feature development efforts to the Python 3.x series. This means that while Python 2 continues to receive bug fixes, and to be updated to build correctly on new hardware and versions of supported operated systems, there will be no new full feature releases for the language or standard library.

Sin embargo, si bien existe un gran subconjunto común entre Python 2.7 y Python 3, y muchos de los cambios involucrados en la migración a ese subconjunto común, o directamente a Python 3, se pueden automatizar de manera segura, algunos otros cambios (en particular los asociados con el manejo de Unicode) puede requerir una consideración cuidadosa, y preferiblemente conjuntos de pruebas de regresión automatizada robustos, para migrar de manera efectiva.

Esto significa que Python 2.7 permanecerá en su lugar durante un largo tiempo, proporcionando una plataforma base estable y compatible para sistemas de producción que aún no se han adaptado a Python 3. La expectativa completa del ciclo de vida de la versión de Python 2.7 esta detallada en PEP 373.

Algunas consecuencias clave de la importancia a largo plazo de 2.7 son:

  • Como se señalo anteriormente, la versión 2.7 tiene un período de mantenimiento mucho más largo en comparación con las versiones anteriores 2.x. Actualmente se espera que Python 2.7 siga siendo compatible con el equipo de desarrollo central (recibiendo actualizaciones de seguridad y otras correcciones de errores) al menos hasta 2020 (10 años después de su lanzamiento inicial, comparado con el típico periodo de mantenimiento de 18–24 meses).

  • A medida que la biblioteca estándar de Python envejece, hacer un uso eficaz del índice de paquetes de Python (ya sea directamente o a través de un distribuidor) se vuelve mas importante para los usuarios de Python 2. Además de una amplia variedad de paquetes de terceros para diversas tareas, los paquetes disponibles incluyen backports de nuevos módulos y características de la biblioteca estándar de Python 3 que son compatibles con Python 2, así como varias herramientas y librerías que pueden hacer mas fácil la migración a Python 3. La Guía de usuario de paquetes de Python proporciona un información sobre como descargar e instalar software desde el índice de paquetes de Python.

  • Aunque el enfoque preferido para mejorar Python 2 es ahora la publicación de nuevos paquetes en el índice de paquetes de Python, este enfoque no necesariamente funciona en todos los casos, especialmente aquellos relacionados a la seguridad de la red. En casos excepcionales que no puedan ser manejados adecuadamente mediante la publicación de nuevos paquetes o actualizaciones en PyPI, el proceso de propuestas de mejoras de Python podría ser usado para crear el caso para añadir nuevas funciones a la librería estándar de Python 2. Dichas adiciones, y las versiones de mantenimiento donde fueron agregadas, se observaran en la sección New Features Added to Python 2.7 Maintenance Releases mas abajo.

Para los proyectos que deseen migrar de Python 2 a Python 3, o para los desarrolladores de frameworks y libreras que deseen dar soporte a usuarios en Python 2 y Python 3, hay una variedad de herramientas y guías disponibles para ayudarles a decidir sobre un enfoque adecuado y manejar algunos de los detalles técnicos involucrados. El punto de partida recomendado es la guía HOWTO Portando código de Python 2 a Python 3.

Cambios en el manejo de las advertencias de desuso

Para Python 2.7, se tomo la decisión de silenciar las advertencias solo de interés para los desarrolladores de manera predeterminada. DeprecationWarning`y sus descendientes son ahora ignoradas a menos que se solicite lo contrario, impidiendo a los usuarios estar viendo las advertencias disparadas por una aplicación. Este cambio se realizo en la rama que se convirtió en Python 3.2. (Discutido en stdlib-sig y llevado a cabo en :issue:`7319).

En versiones previas, los mensajes DeprecationWarning estaban habilitados de manera predeterminada, proporcionando a los desarrolladores de Python una indicación clara de donde podría romperse su código en una futura versión principal de Python.

Sin embargo, cada vez hay más usuarios de aplicaciones basadas en Python que no están directamente involucrados en el desarrollo de esas aplicaciones. Los mensajes DeprecationWarning son irrelevantes para estos usuarios, hacen que se preocupen por una aplicación que realmente está funcionando correctamente y sobrecargando a los desarrolladores de aplicaciones con responder a estas preocupaciones.

Puede volver a habilitar la visualización de los mensajes DeprecationWarning ejecutando Python con el modificador -Wdefault (forma corta: -Wd), o estableciendo la variable de entorno PYTHONWARNINGS en "default" (o "d") antes de ejecutar Python. El código Python también puede volver a habilitarlos llamando a “”warnings.simplefilter(“default”)``.

El módulo unittest también vuelve a instalar automáticamente las advertencias de desuso al ejecutar pruebas.

Características de Python 3.1

Al igual que Python 2.6 incorporó características de Python 3.0, la versión 2.7 incorpora algunas de las nuevas características de Python 3.1. La serie 2.x continúa proporcionando herramientas para migrar a la serie 3.x.

Una lista parcial de características 3.1 que se retro importaron a 2.7:

  • La sintaxis para establecer literales ({1,2,3} es un conjunto mutable).

  • Diccionario y establecimiento de compresiones ({i: i*2 for i in range(3)}).

  • Múltiples administradores de contexto en una sola declaración with.

  • Una nueva versión para la librería io, rescrito en C para un mejor desempeño.

  • El tipo de ordenador de diccionario descrito en PEP 372: Adición de un diccionario ordenado a las colecciones.

  • El nuevo especificador de formato "," descrito en PEP 378: Especificador de formato para separador de miles.

  • El objeto memoryview.

  • Un pequeño subconjunto del modulo importlib, descrito abajo.

  • La repr() de un flotante x es más corto en muchos casos: ahora se basa en la cadena decimal más corta que se garantiza que redondea a x. Como en versiones anteriores de Python, se garantiza que float(repr(x)) devuelva x.

  • Las conversiones flotante a cadena y cadena a flotante se redondean correctamente. La función round() también ahora se redondea correctamente.

  • El tipo PyCapsule, utilizado para proporcionar una API de C para módulos de extensión.

  • La función de API de C PyLong_AsLongAndOverflow().

Otras nuevas advertencias en modo Python3 incluyen:

  • operator.isCallable() y operator.sequenceIncludes(), que no se admiten en 3.x, ahora activan advertencias.

  • El modificador -3 ahora habilita automáticamente el modificador -Qwarn que provoca advertencias sobre el uso de la división clásica con enteros y enteros largos.

PEP 372: Adición de un diccionario ordenado a las colecciones

Los diccionarios de python normales recorren en iteración los pares clave/valor en un orden arbitrario. A lo largo de los años, varios autores han escrito implementaciones alternativas que recuerdan el orden en que se insertaron originalmente las claves. Basado en las experiencias de esas implementaciones, 2.7 presenta una nueva clase OrderedDict en el módulo collections.

La API OrderedDict proporciona la misma interfaz que los diccionarios normales, pero recorre en iteración las claves y los valores en un orden garantizado en función de cuándo se insertó una clave por primera vez:

>>> from collections import OrderedDict
>>> d = OrderedDict([('first', 1),
...                  ('second', 2),
...                  ('third', 3)])
>>> d.items()
[('first', 1), ('second', 2), ('third', 3)]

Si una nueva entrada sobrescribe una entrada existente, la posición de inserción original no se modifica:

>>> d['second'] = 4
>>> d.items()
[('first', 1), ('second', 4), ('third', 3)]

Borrando una entrada y luego reinsertándola, moverá la misma al final:

>>> del d['second']
>>> d['second'] = 5
>>> d.items()
[('first', 1), ('third', 3), ('second', 5)]

El método popitem() tiene un argumento last opcional que por defecto esta en True (verdadero). Si last es verdadero, la ultima llave ingresada es regresada y eliminada; si es falso, se selecciona la llave más antigua:

>>> od = OrderedDict([(x,0) for x in range(20)])
>>> od.popitem()
(19, 0)
>>> od.popitem()
(18, 0)
>>> od.popitem(last=False)
(0, 0)
>>> od.popitem(last=False)
(1, 0)

La comparación de dos diccionarios ordenados comprueba tanto las claves como los valores, y requiere que el orden de inserción sea el mismo:

>>> od1 = OrderedDict([('first', 1),
...                    ('second', 2),
...                    ('third', 3)])
>>> od2 = OrderedDict([('third', 3),
...                    ('first', 1),
...                    ('second', 2)])
>>> od1 == od2
False
>>> # Move 'third' key to the end
>>> del od2['third']; od2['third'] = 3
>>> od1 == od2
True

Comparando a un OrderedDict con un diccionario normal ignora el orden de inserción y simplemente compara las llaves y los valores.

¿Cómo funciona OrderedDict? Mantiene una lista doblemente vinculada de claves, añadiendo nuevas claves a la lista a medida que se insertan. Un diccionario secundario asigna las claves a su nodo de lista correspondiente, por lo que la eliminación no tiene que recorrer toda la lista vinculada y, por lo tanto, sigue siendo O(1).

La biblioteca estándar ahora admite el uso de diccionarios ordenados en varios módulos.

  • El módulo ConfigParser los utiliza de forma predeterminada, lo que significa que los archivos de configuración ahora se pueden leer, modificar y volver a escribir en su orden original.

  • El método _asdict() para collections.namedtuple() ahora devuelve un diccionario ordenado con los valores que aparecen en el mismo orden que los índices de tupla subyacentes.

  • El módulo json de la clase constructor JSONDecoder se amplió con un parámetro object_pairs_hook para permitir que el descodificador compilara instancias OrderedDict. También se agregó compatibilidad con herramientas de terceros como PyYAML.

Ver también

PEP 372 - Adición de un diccionario ordenado a las collections

PEP escrito por Armin Ronacher y Raymond Hettinger; implementado por Raymond Hettinger.

PEP 378: Especificador de formato para separador de miles

Para que la salida del programa sea más legible, puede ser útil agregar separadores a números grandes, representándolos como 18,446,744,073,709,551,616 en lugar de 18446744073709551616.

La solución totalmente general para hacer esto es el módulo locale, que puede utilizar diferentes separadores («,» en América del Norte, «.» en Europa) y diferentes tamaños de agrupación, pero locale es complicado de usar y no es adecuado para aplicaciones multiproceso donde diferentes hilos están produciendo resultados para diferentes configuraciones regionales.

Por lo tanto, se ha añadido un mecanismo simple de agrupación de comas al mini-lenguaje utilizado por el método str.format(). Al dar formato a un número de punto flotante, simplemente incluya una coma entre el ancho y la precisión:

>>> '{:20,.2f}'.format(18446744073709551616.0)
'18,446,744,073,709,551,616.00'

Al dar formato a un entero, incluya la coma después del ancho:

>>> '{:20,d}'.format(18446744073709551616)
'18,446,744,073,709,551,616'

Este mecanismo no es adaptable en absoluto; Las comas siempre se utilizan como separador y la agrupación siempre está en grupos de tres dígitos. El mecanismo de formato de coma no es tan general como el módulo locale, pero es más fácil de usar.

Ver también

PEP 378 - Especificador de formato para separador de miles

PEP escrito por Raymond Hettinger; implementado por Eric Smith.

PEP 389: El módulo argparse para el análisis de líneas de comando

El módulo argparse para analizar argumentos de línea de comandos se agregó como un reemplazo más potente para el módulo optparse.

Esto significa que Python ahora admite tres módulos diferentes para analizar argumentos de línea de comandos: getopt, optparse, y argparse. El módulo getopt se parece mucho a la función getopt() de la biblioteca C, por lo que sigue siendo útil si estás escribiendo un prototipo de Python que finalmente se reescribe en C. optparse se vuelve redundante, pero no hay planes para eliminarlo porque hay muchos scripts que todavía lo usan, y no hay una manera automatizada de actualizar estos scripts. (Hacer que la API argparse sea coherente con la interfaz de optparse fue discutido pero rechazado como demasiado desordenado y difícil.)

En resumen, si estás escribiendo un nuevo script y no necesitas preocuparte por la compatibilidad con versiones anteriores de Python, usa argparse en lugar de optparse.

Este es un ejemplo:

import argparse

parser = argparse.ArgumentParser(description='Command-line example.')

# Add optional switches
parser.add_argument('-v', action='store_true', dest='is_verbose',
                    help='produce verbose output')
parser.add_argument('-o', action='store', dest='output',
                    metavar='FILE',
                    help='direct output to FILE instead of stdout')
parser.add_argument('-C', action='store', type=int, dest='context',
                    metavar='NUM', default=0,
                    help='display NUM lines of added context')

# Allow any number of additional arguments.
parser.add_argument(nargs='*', action='store', dest='inputs',
                    help='input filenames (default is stdin)')

args = parser.parse_args()
print args.__dict__

A menos que lo reemplace, los modificadores -h y --help se agregan automáticamente, y producen una salida con formato ordenado:

-> ./python.exe argparse-example.py --help
usage: argparse-example.py [-h] [-v] [-o FILE] [-C NUM] [inputs [inputs ...]]

Command-line example.

positional arguments:
  inputs      input filenames (default is stdin)

optional arguments:
  -h, --help  show this help message and exit
  -v          produce verbose output
  -o FILE     direct output to FILE instead of stdout
  -C NUM      display NUM lines of added context

Al igual que con optparse, los modificadores y argumentos de línea de comandos se devuelven como un objeto con atributos denominados por los parámetros dest:

-> ./python.exe argparse-example.py -v
{'output': None,
 'is_verbose': True,
 'context': 0,
 'inputs': []}

-> ./python.exe argparse-example.py -v -o /tmp/output -C 4 file1 file2
{'output': '/tmp/output',
 'is_verbose': True,
 'context': 4,
 'inputs': ['file1', 'file2']}

argparse tiene una validación mucho más fantasiosa que optparse; puede especificar un número exacto de argumentos como un entero, 0 o más argumentos pasando '*', 1 o más pasando '+', o un argumento opcional con '?'. Un analizador de nivel superior puede contener sub analizadores para definir subcomandos que tienen diferentes conjuntos de modificadores, como en svn commit, svn checkout, etc. Puede especificar el tipo de un argumento como FileType, que abrirá automáticamente los archivos y entiende que '-' significa entrada o salida estándar.

Ver también

argparse documentación

La página de documentación del módulo argparse.

Actualizar el código de optparse

Parte de la documentación de Python, que describe cómo convertir código que usa optparse.

PEP 389 - argparse - Nuevo módulo de análisis de línea de comandos

PEP escrito e implementado por Steven Bethard.

PEP 391: Configuración basada en diccionarios para el registro

El módulo logging es muy flexible; las aplicaciones pueden definir un árbol de subsistemas de registro, y cada registrador de este árbol puede filtrar ciertos mensajes, formatearlos de forma diferente y dirigir mensajes a un número variable de controladores.

Toda esta flexibilidad puede requerir mucha configuración. Puede escribir instrucciones Python para crear objetos y establecer sus propiedades, pero una configuración compleja requiere código detallado pero aburrido. logging también es compatible con una función fileConfig() que analiza un archivo, pero el formato de archivo no admite la configuración de filtros, y es más complicado generar mediante programación.

Python 2.7 agrega una función dictConfig() que utiliza un diccionario para configurar el registro. Hay muchas maneras de producir un diccionario de diferentes fuentes: construir uno con código; analizar un archivo que contenga JSON; o utilice una biblioteca de análisis YAML si hay una instalada. Para obtener más información, consulte Funciones de configuración.

En el ejemplo siguiente se configuran dos registradores, el registrador raíz y un registrador denominado «network». Los mensajes enviados al registrador raíz se enviarán al registro del sistema mediante el protocolo syslog, y los mensajes al registrador de la «network» se escribirán en un archivo network.log que se rotará una vez que el registro alcance 1MB.

import logging
import logging.config

configdict = {
 'version': 1,    # Configuration schema in use; must be 1 for now
 'formatters': {
     'standard': {
         'format': ('%(asctime)s %(name)-15s '
                    '%(levelname)-8s %(message)s')}},

 'handlers': {'netlog': {'backupCount': 10,
                     'class': 'logging.handlers.RotatingFileHandler',
                     'filename': '/logs/network.log',
                     'formatter': 'standard',
                     'level': 'INFO',
                     'maxBytes': 1000000},
              'syslog': {'class': 'logging.handlers.SysLogHandler',
                         'formatter': 'standard',
                         'level': 'ERROR'}},

 # Specify all the subordinate loggers
 'loggers': {
             'network': {
                         'handlers': ['netlog']
             }
 },
 # Specify properties of the root logger
 'root': {
          'handlers': ['syslog']
 },
}

# Set up configuration
logging.config.dictConfig(configdict)

# As an example, log two error messages
logger = logging.getLogger('/')
logger.error('Database not found')

netlogger = logging.getLogger('network')
netlogger.error('Connection failed')

Tres mejoras más pequeñas en el módulo logging, todas implementadas por Vinay Sajip, son:

  • La clase SysLogHandler ahora admite syslogging a través de TCP. El constructor tiene un parámetro socktype que proporciona el tipo de socket que se va a usar, ya sea socket.SOCK_DGRAM para UDP o socket.SOCK_STREAM para TCP. El protocolo predeterminado sigue siendo UDP.

  • Las instancias Logger ganaron un método getChild() que recupera un registrador descendiente mediante una ruta de acceso relativa. Por ejemplo, una vez que se recupera un registrador haciendo log = getLogger('app'), llamando log.getChild('network.listen') es equivalente a getLogger('app.network.listen').

  • La clase LoggerAdapter gano un método isEnabledFor() que toma un nivel y devuelve si el registrador subyacente procesará un mensaje de ese nivel de importancia.

Ver también

PEP 391 - Configuración basada en diccionarios para el registro

PEP escrito e implementado por Vinay Sajip.

PEP 3106: Vistas de diccionario

Los métodos de diccionarios keys(), values() y items() son diferentes en Python 3.x. Ellos regresan un objeto llamado view en lugar de una lista completamente materializada.

No es posible cambiar el retorno de valores de los métodos keys(), values() y items() en Python 2.7 porque se rompería demasiado código. En su lugar en las versiones 3.x se fueron agregando bajo los nuevos nombres viewkeys(), viewvalues() y viewitems().

>>> d = dict((i*10, chr(65+i)) for i in range(26))
>>> d
{0: 'A', 130: 'N', 10: 'B', 140: 'O', 20: ..., 250: 'Z'}
>>> d.viewkeys()
dict_keys([0, 130, 10, 140, 20, 150, 30, ..., 250])

Las vistas se pueden iterar, pero las vistas de elementos y claves también se comportan como conjuntos. El operador & realiza la intersección y | realiza una unión:

>>> d1 = dict((i*10, chr(65+i)) for i in range(26))
>>> d2 = dict((i**.5, i) for i in range(1000))
>>> d1.viewkeys() & d2.viewkeys()
set([0.0, 10.0, 20.0, 30.0])
>>> d1.viewkeys() | range(0, 30)
set([0, 1, 130, 3, 4, 5, 6, ..., 120, 250])

La vista realiza un seguimiento del diccionario y su contenido cambia a medida que se modifica el diccionario:

>>> vk = d.viewkeys()
>>> vk
dict_keys([0, 130, 10, ..., 250])
>>> d[260] = '&'
>>> vk
dict_keys([0, 130, 260, 10, ..., 250])

Sin embargo, tenga en cuenta que no puede agregar o eliminar claves mientras está iterando sobre la vista:

>>> for k in vk:
...     d[k*2] = k
...
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
RuntimeError: dictionary changed size during iteration

Puede usar los métodos de vista en el código en Python 2.x, y el convertidor 2to3 los cambiará a los métodos estándar keys(), values() y items().

Ver también

PEP 3106 - Renovación dict.keys(), .values() y .items()

PEP escrito por Guido van Rossum. Portado a 2.7 por Alexandre Vassalotti; bpo-1967.

PEP 3137: El objeto memoryview

El objeto memoryview proporciona una vista del contenido de la memoria de otro objeto que coincide con la interfaz de tipo bytes.

>>> import string
>>> m = memoryview(string.letters)
>>> m
<memory at 0x37f850>
>>> len(m)           # Returns length of underlying object
52
>>> m[0], m[25], m[26]   # Indexing returns one byte
('a', 'z', 'A')
>>> m2 = m[0:26]         # Slicing returns another memoryview
>>> m2
<memory at 0x37f080>

El contenido de la vista se puede convertir en una cadena de bytes o una lista de números enteros:

>>> m2.tobytes()
'abcdefghijklmnopqrstuvwxyz'
>>> m2.tolist()
[97, 98, 99, 100, 101, 102, 103, ... 121, 122]
>>>

Los objetos memoryview permiten modificar el objeto subyacente si es un objeto mutable.

>>> m2[0] = 75
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: cannot modify read-only memory
>>> b = bytearray(string.letters)  # Creating a mutable object
>>> b
bytearray(b'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ')
>>> mb = memoryview(b)
>>> mb[0] = '*'         # Assign to view, changing the bytearray.
>>> b[0:5]              # The bytearray has been changed.
bytearray(b'*bcde')
>>>

Ver también

PEP 3137 - Bytes inmutables y búfer mutable

PEP escrito por Guido van Rossum. Implementado por Travis Oliphant, Antoine Pitrou y otros. Portado a 2.7 por Antoine Pitrou; bpo-2396.

Otros cambios de lenguaje

Algunos de los cambios mas pequeños hechos al núcleo del lenguaje de Python son:

  • La sintaxis para literales de conjuntos se ha actualizado desde Python 3.x. Los corchetes se utilizan para rodear el contenido del conjunto mutable resultante; los literales de conjunto se distinguen de los diccionarios por no contener dos puntos ni valores. {} sigue representando un diccionario vacío; use set() para un conjunto vacío.

    >>> {1, 2, 3, 4, 5}
    set([1, 2, 3, 4, 5])
    >>> set() # empty set
    set([])
    >>> {}    # empty dict
    {}
    

    Portado por Alexandre Vassalotti; bpo-2335.

  • El diccionario y las comprensiones de conjuntos son otra característica portada desde 3.x, que generaliza las comprensiones de listas / generadores para usar la sintaxis literal para conjuntos y diccionarios.

    >>> {x: x*x for x in range(6)}
    {0: 0, 1: 1, 2: 4, 3: 9, 4: 16, 5: 25}
    >>> {('a'*x) for x in range(6)}
    set(['', 'a', 'aa', 'aaa', 'aaaa', 'aaaaa'])
    

    Portado por Alexandre Vassalotti; bpo-2333.

  • La declaración with ahora puede usar múltiples administradores de contexto en una declaración. Los administradores de contexto se procesan de izquierda a derecha y cada uno se trata como si comenzara una nueva declaración with. Esto significa que:

    with A() as a, B() as b:
        ... suite of statements ...
    

    es equivalente a:

    with A() as a:
        with B() as b:
            ... suite of statements ...
    

    La función contextlib.nested() provee una función muy similar, por lo que ya no es necesario y ha quedado obsoleto.

    (Propuesta en https://codereview.appspot.com/53094; implementado por Georg Brandl.)

  • Las conversiones entre números de punto flotante y cadenas ahora se redondean correctamente en la mayoría de las plataformas. Estas conversiones ocurren en muchos lugares diferentes: str() en flotantes y números complejos; los constructores float y complex; formato numérico; serializar y deserializar flotantes y números complejos usando los módulos marshal, pickle y json; análisis de literales flotantes e imaginarios en código Python; y conversión Decimal a flotante.

    Relacionado con esto, el repr() de un número de punto flotante x ahora devuelve un resultado basado en la cadena decimal más corta que se garantiza que se redondeará a x con el redondeo correcto (con el modo de redondeo round-half-to-even). Anteriormente, daba una cadena basada en redondear x a 17 dígitos decimales.

    The rounding library responsible for this improvement works on Windows and on Unix platforms using the gcc, icc, or suncc compilers. There may be a small number of platforms where correct operation of this code cannot be guaranteed, so the code is not used on such systems. You can find out which code is being used by checking sys.float_repr_style, which will be short if the new code is in use and legacy if it isn’t.

    Implemented by Eric Smith and Mark Dickinson, using David Gay’s dtoa.c library; bpo-7117.

  • Conversions from long integers and regular integers to floating point now round differently, returning the floating-point number closest to the number. This doesn’t matter for small integers that can be converted exactly, but for large numbers that will unavoidably lose precision, Python 2.7 now approximates more closely. For example, Python 2.6 computed the following:

    >>> n = 295147905179352891391
    >>> float(n)
    2.9514790517935283e+20
    >>> n - long(float(n))
    65535L
    

    Python 2.7’s floating-point result is larger, but much closer to the true value:

    >>> n = 295147905179352891391
    >>> float(n)
    2.9514790517935289e+20
    >>> n - long(float(n))
    -1L
    

    (Implemented by Mark Dickinson; bpo-3166.)

    Integer division is also more accurate in its rounding behaviours. (Also implemented by Mark Dickinson; bpo-1811.)

  • Implicit coercion for complex numbers has been removed; the interpreter will no longer ever attempt to call a __coerce__() method on complex objects. (Removed by Meador Inge and Mark Dickinson; bpo-5211.)

  • The str.format() method now supports automatic numbering of the replacement fields. This makes using str.format() more closely resemble using %s formatting:

    >>> '{}:{}:{}'.format(2009, 04, 'Sunday')
    '2009:4:Sunday'
    >>> '{}:{}:{day}'.format(2009, 4, day='Sunday')
    '2009:4:Sunday'
    

    The auto-numbering takes the fields from left to right, so the first {...} specifier will use the first argument to str.format(), the next specifier will use the next argument, and so on. You can’t mix auto-numbering and explicit numbering – either number all of your specifier fields or none of them – but you can mix auto-numbering and named fields, as in the second example above. (Contributed by Eric Smith; bpo-5237.)

    Complex numbers now correctly support usage with format(), and default to being right-aligned. Specifying a precision or comma-separation applies to both the real and imaginary parts of the number, but a specified field width and alignment is applied to the whole of the resulting 1.5+3j output. (Contributed by Eric Smith; bpo-1588 and bpo-7988.)

    The “F” format code now always formats its output using uppercase characters, so it will now produce “INF” and “NAN”. (Contributed by Eric Smith; bpo-3382.)

    A low-level change: the object.__format__() method now triggers a PendingDeprecationWarning if it’s passed a format string, because the __format__() method for object converts the object to a string representation and formats that. Previously the method silently applied the format string to the string representation, but that could hide mistakes in Python code. If you’re supplying formatting information such as an alignment or precision, presumably you’re expecting the formatting to be applied in some object-specific way. (Fixed by Eric Smith; bpo-7994.)

  • The int() and long() types gained a bit_length method that returns the number of bits necessary to represent its argument in binary:

    >>> n = 37
    >>> bin(n)
    '0b100101'
    >>> n.bit_length()
    6
    >>> n = 2**123-1
    >>> n.bit_length()
    123
    >>> (n+1).bit_length()
    124
    

    (Contributed by Fredrik Johansson and Victor Stinner; bpo-3439.)

  • The import statement will no longer try an absolute import if a relative import (e.g. from .os import sep) fails. This fixes a bug, but could possibly break certain import statements that were only working by accident. (Fixed by Meador Inge; bpo-7902.)

  • It’s now possible for a subclass of the built-in unicode type to override the __unicode__() method. (Implemented by Victor Stinner; bpo-1583863.)

  • The bytearray type’s translate() method now accepts None as its first argument. (Fixed by Georg Brandl; bpo-4759.)

  • When using @classmethod and @staticmethod to wrap methods as class or static methods, the wrapper object now exposes the wrapped function as their __func__ attribute. (Contributed by Amaury Forgeot d’Arc, after a suggestion by George Sakkis; bpo-5982.)

  • When a restricted set of attributes were set using __slots__, deleting an unset attribute would not raise AttributeError as you would expect. Fixed by Benjamin Peterson; bpo-7604.)

  • Two new encodings are now supported: «cp720», used primarily for Arabic text; and «cp858», a variant of CP 850 that adds the euro symbol. (CP720 contributed by Alexander Belchenko and Amaury Forgeot d’Arc in bpo-1616979; CP858 contributed by Tim Hatch in bpo-8016.)

  • The file object will now set the filename attribute on the IOError exception when trying to open a directory on POSIX platforms (noted by Jan Kaliszewski; bpo-4764), and now explicitly checks for and forbids writing to read-only file objects instead of trusting the C library to catch and report the error (fixed by Stefan Krah; bpo-5677).

  • The Python tokenizer now translates line endings itself, so the compile() built-in function now accepts code using any line-ending convention. Additionally, it no longer requires that the code end in a newline.

  • Extra parentheses in function definitions are illegal in Python 3.x, meaning that you get a syntax error from def f((x)): pass. In Python3-warning mode, Python 2.7 will now warn about this odd usage. (Noted by James Lingard; bpo-7362.)

  • It’s now possible to create weak references to old-style class objects. New-style classes were always weak-referenceable. (Fixed by Antoine Pitrou; bpo-8268.)

  • When a module object is garbage-collected, the module’s dictionary is now only cleared if no one else is holding a reference to the dictionary (bpo-7140).

Interpreter Changes

A new environment variable, PYTHONWARNINGS, allows controlling warnings. It should be set to a string containing warning settings, equivalent to those used with the -W switch, separated by commas. (Contributed by Brian Curtin; bpo-7301.)

For example, the following setting will print warnings every time they occur, but turn warnings from the Cookie module into an error. (The exact syntax for setting an environment variable varies across operating systems and shells.)

export PYTHONWARNINGS=all,error:::Cookie:0

Optimizations

Several performance enhancements have been added:

  • A new opcode was added to perform the initial setup for with statements, looking up the __enter__() and __exit__() methods. (Contributed by Benjamin Peterson.)

  • The garbage collector now performs better for one common usage pattern: when many objects are being allocated without deallocating any of them. This would previously take quadratic time for garbage collection, but now the number of full garbage collections is reduced as the number of objects on the heap grows. The new logic only performs a full garbage collection pass when the middle generation has been collected 10 times and when the number of survivor objects from the middle generation exceeds 10% of the number of objects in the oldest generation. (Suggested by Martin von Löwis and implemented by Antoine Pitrou; bpo-4074.)

  • The garbage collector tries to avoid tracking simple containers which can’t be part of a cycle. In Python 2.7, this is now true for tuples and dicts containing atomic types (such as ints, strings, etc.). Transitively, a dict containing tuples of atomic types won’t be tracked either. This helps reduce the cost of each garbage collection by decreasing the number of objects to be considered and traversed by the collector. (Contributed by Antoine Pitrou; bpo-4688.)

  • Long integers are now stored internally either in base 2**15 or in base 2**30, the base being determined at build time. Previously, they were always stored in base 2**15. Using base 2**30 gives significant performance improvements on 64-bit machines, but benchmark results on 32-bit machines have been mixed. Therefore, the default is to use base 2**30 on 64-bit machines and base 2**15 on 32-bit machines; on Unix, there’s a new configure option --enable-big-digits that can be used to override this default.

    Apart from the performance improvements this change should be invisible to end users, with one exception: for testing and debugging purposes there’s a new structseq sys.long_info that provides information about the internal format, giving the number of bits per digit and the size in bytes of the C type used to store each digit:

    >>> import sys
    >>> sys.long_info
    sys.long_info(bits_per_digit=30, sizeof_digit=4)
    

    (Contributed by Mark Dickinson; bpo-4258.)

    Another set of changes made long objects a few bytes smaller: 2 bytes smaller on 32-bit systems and 6 bytes on 64-bit. (Contributed by Mark Dickinson; bpo-5260.)

  • The division algorithm for long integers has been made faster by tightening the inner loop, doing shifts instead of multiplications, and fixing an unnecessary extra iteration. Various benchmarks show speedups of between 50% and 150% for long integer divisions and modulo operations. (Contributed by Mark Dickinson; bpo-5512.) Bitwise operations are also significantly faster (initial patch by Gregory Smith; bpo-1087418).

  • The implementation of % checks for the left-side operand being a Python string and special-cases it; this results in a 1–3% performance increase for applications that frequently use % with strings, such as templating libraries. (Implemented by Collin Winter; bpo-5176.)

  • List comprehensions with an if condition are compiled into faster bytecode. (Patch by Antoine Pitrou, back-ported to 2.7 by Jeffrey Yasskin; bpo-4715.)

  • Converting an integer or long integer to a decimal string was made faster by special-casing base 10 instead of using a generalized conversion function that supports arbitrary bases. (Patch by Gawain Bolton; bpo-6713.)

  • The split(), replace(), rindex(), rpartition(), and rsplit() methods of string-like types (strings, Unicode strings, and bytearray objects) now use a fast reverse-search algorithm instead of a character-by-character scan. This is sometimes faster by a factor of 10. (Added by Florent Xicluna; bpo-7462 and bpo-7622.)

  • The pickle and cPickle modules now automatically intern the strings used for attribute names, reducing memory usage of the objects resulting from unpickling. (Contributed by Jake McGuire; bpo-5084.)

  • The cPickle module now special-cases dictionaries, nearly halving the time required to pickle them. (Contributed by Collin Winter; bpo-5670.)

New and Improved Modules

As in every release, Python’s standard library received a number of enhancements and bug fixes. Here’s a partial list of the most notable changes, sorted alphabetically by module name. Consult the Misc/NEWS file in the source tree for a more complete list of changes, or look through the Subversion logs for all the details.

  • The bdb module’s base debugging class Bdb gained a feature for skipping modules. The constructor now takes an iterable containing glob-style patterns such as django.*; the debugger will not step into stack frames from a module that matches one of these patterns. (Contributed by Maru Newby after a suggestion by Senthil Kumaran; bpo-5142.)

  • The binascii module now supports the buffer API, so it can be used with memoryview instances and other similar buffer objects. (Backported from 3.x by Florent Xicluna; bpo-7703.)

  • Updated module: the bsddb module has been updated from 4.7.2devel9 to version 4.8.4 of the pybsddb package. The new version features better Python 3.x compatibility, various bug fixes, and adds several new BerkeleyDB flags and methods. (Updated by Jesús Cea Avión; bpo-8156. The pybsddb changelog can be read at http://hg.jcea.es/pybsddb/file/tip/ChangeLog.)

  • The bz2 module’s BZ2File now supports the context management protocol, so you can write with bz2.BZ2File(...) as f:. (Contributed by Hagen Fürstenau; bpo-3860.)

  • New class: the Counter class in the collections module is useful for tallying data. Counter instances behave mostly like dictionaries but return zero for missing keys instead of raising a KeyError:

    >>> from collections import Counter
    >>> c = Counter()
    >>> for letter in 'here is a sample of english text':
    ...   c[letter] += 1
    ...
    >>> c 
    Counter({' ': 6, 'e': 5, 's': 3, 'a': 2, 'i': 2, 'h': 2,
    'l': 2, 't': 2, 'g': 1, 'f': 1, 'm': 1, 'o': 1, 'n': 1,
    'p': 1, 'r': 1, 'x': 1})
    >>> c['e']
    5
    >>> c['z']
    0
    

    There are three additional Counter methods. most_common() returns the N most common elements and their counts. elements() returns an iterator over the contained elements, repeating each element as many times as its count. subtract() takes an iterable and subtracts one for each element instead of adding; if the argument is a dictionary or another Counter, the counts are subtracted.

    >>> c.most_common(5)
    [(' ', 6), ('e', 5), ('s', 3), ('a', 2), ('i', 2)]
    >>> c.elements() ->
       'a', 'a', ' ', ' ', ' ', ' ', ' ', ' ',
       'e', 'e', 'e', 'e', 'e', 'g', 'f', 'i', 'i',
       'h', 'h', 'm', 'l', 'l', 'o', 'n', 'p', 's',
       's', 's', 'r', 't', 't', 'x'
    >>> c['e']
    5
    >>> c.subtract('very heavy on the letter e')
    >>> c['e']    # Count is now lower
    -1
    

    Contributed by Raymond Hettinger; bpo-1696199.

    New class: OrderedDict is described in the earlier section PEP 372: Adición de un diccionario ordenado a las colecciones.

    New method: The deque data type now has a count() method that returns the number of contained elements equal to the supplied argument x, and a reverse() method that reverses the elements of the deque in-place. deque also exposes its maximum length as the read-only maxlen attribute. (Both features added by Raymond Hettinger.)

    The namedtuple class now has an optional rename parameter. If rename is true, field names that are invalid because they’ve been repeated or aren’t legal Python identifiers will be renamed to legal names that are derived from the field’s position within the list of fields:

    >>> from collections import namedtuple
    >>> T = namedtuple('T', ['field1', '$illegal', 'for', 'field2'], rename=True)
    >>> T._fields
    ('field1', '_1', '_2', 'field2')
    

    (Added by Raymond Hettinger; bpo-1818.)

    Finally, the Mapping abstract base class now returns NotImplemented if a mapping is compared to another type that isn’t a Mapping. (Fixed by Daniel Stutzbach; bpo-8729.)

  • Constructors for the parsing classes in the ConfigParser module now take an allow_no_value parameter, defaulting to false; if true, options without values will be allowed. For example:

    >>> import ConfigParser, StringIO
    >>> sample_config = """
    ... [mysqld]
    ... user = mysql
    ... pid-file = /var/run/mysqld/mysqld.pid
    ... skip-bdb
    ... """
    >>> config = ConfigParser.RawConfigParser(allow_no_value=True)
    >>> config.readfp(StringIO.StringIO(sample_config))
    >>> config.get('mysqld', 'user')
    'mysql'
    >>> print config.get('mysqld', 'skip-bdb')
    None
    >>> print config.get('mysqld', 'unknown')
    Traceback (most recent call last):
      ...
    NoOptionError: No option 'unknown' in section: 'mysqld'
    

    (Contributed by Mats Kindahl; bpo-7005.)

  • Deprecated function: contextlib.nested(), which allows handling more than one context manager with a single with statement, has been deprecated, because the with statement now supports multiple context managers.

  • The cookielib module now ignores cookies that have an invalid version field, one that doesn’t contain an integer value. (Fixed by John J. Lee; bpo-3924.)

  • The copy module’s deepcopy() function will now correctly copy bound instance methods. (Implemented by Robert Collins; bpo-1515.)

  • The ctypes module now always converts None to a C NULL pointer for arguments declared as pointers. (Changed by Thomas Heller; bpo-4606.) The underlying libffi library has been updated to version 3.0.9, containing various fixes for different platforms. (Updated by Matthias Klose; bpo-8142.)

  • New method: the datetime module’s timedelta class gained a total_seconds() method that returns the number of seconds in the duration. (Contributed by Brian Quinlan; bpo-5788.)

  • New method: the Decimal class gained a from_float() class method that performs an exact conversion of a floating-point number to a Decimal. This exact conversion strives for the closest decimal approximation to the floating-point representation’s value; the resulting decimal value will therefore still include the inaccuracy, if any. For example, Decimal.from_float(0.1) returns Decimal('0.1000000000000000055511151231257827021181583404541015625'). (Implemented by Raymond Hettinger; bpo-4796.)

    Comparing instances of Decimal with floating-point numbers now produces sensible results based on the numeric values of the operands. Previously such comparisons would fall back to Python’s default rules for comparing objects, which produced arbitrary results based on their type. Note that you still cannot combine Decimal and floating-point in other operations such as addition, since you should be explicitly choosing how to convert between float and Decimal. (Fixed by Mark Dickinson; bpo-2531.)

    The constructor for Decimal now accepts floating-point numbers (added by Raymond Hettinger; bpo-8257) and non-European Unicode characters such as Arabic-Indic digits (contributed by Mark Dickinson; bpo-6595).

    Most of the methods of the Context class now accept integers as well as Decimal instances; the only exceptions are the canonical() and is_canonical() methods. (Patch by Juan José Conti; bpo-7633.)

    When using Decimal instances with a string’s format() method, the default alignment was previously left-alignment. This has been changed to right-alignment, which is more sensible for numeric types. (Changed by Mark Dickinson; bpo-6857.)

    Comparisons involving a signaling NaN value (or sNAN) now signal InvalidOperation instead of silently returning a true or false value depending on the comparison operator. Quiet NaN values (or NaN) are now hashable. (Fixed by Mark Dickinson; bpo-7279.)

  • The difflib module now produces output that is more compatible with modern diff/patch tools through one small change, using a tab character instead of spaces as a separator in the header giving the filename. (Fixed by Anatoly Techtonik; bpo-7585.)

  • The Distutils sdist command now always regenerates the MANIFEST file, since even if the MANIFEST.in or setup.py files haven’t been modified, the user might have created some new files that should be included. (Fixed by Tarek Ziadé; bpo-8688.)

  • The doctest module’s IGNORE_EXCEPTION_DETAIL flag will now ignore the name of the module containing the exception being tested. (Patch by Lennart Regebro; bpo-7490.)

  • The email module’s Message class will now accept a Unicode-valued payload, automatically converting the payload to the encoding specified by output_charset. (Added by R. David Murray; bpo-1368247.)

  • The Fraction class now accepts a single float or Decimal instance, or two rational numbers, as arguments to its constructor. (Implemented by Mark Dickinson; rationals added in bpo-5812, and float/decimal in bpo-8294.)

    Ordering comparisons (<, <=, >, >=) between fractions and complex numbers now raise a TypeError. This fixes an oversight, making the Fraction match the other numeric types.

  • New class: FTP_TLS in the ftplib module provides secure FTP connections using TLS encapsulation of authentication as well as subsequent control and data transfers. (Contributed by Giampaolo Rodola; bpo-2054.)

    The storbinary() method for binary uploads can now restart uploads thanks to an added rest parameter (patch by Pablo Mouzo; bpo-6845.)

  • New class decorator: total_ordering() in the functools module takes a class that defines an __eq__() method and one of __lt__(), __le__(), __gt__(), or __ge__(), and generates the missing comparison methods. Since the __cmp__() method is being deprecated in Python 3.x, this decorator makes it easier to define ordered classes. (Added by Raymond Hettinger; bpo-5479.)

    New function: cmp_to_key() will take an old-style comparison function that expects two arguments and return a new callable that can be used as the key parameter to functions such as sorted(), min() and max(), etc. The primary intended use is to help with making code compatible with Python 3.x. (Added by Raymond Hettinger.)

  • New function: the gc module’s is_tracked() returns true if a given instance is tracked by the garbage collector, false otherwise. (Contributed by Antoine Pitrou; bpo-4688.)

  • The gzip module’s GzipFile now supports the context management protocol, so you can write with gzip.GzipFile(...) as f: (contributed by Hagen Fürstenau; bpo-3860), and it now implements the io.BufferedIOBase ABC, so you can wrap it with io.BufferedReader for faster processing (contributed by Nir Aides; bpo-7471). It’s also now possible to override the modification time recorded in a gzipped file by providing an optional timestamp to the constructor. (Contributed by Jacques Frechet; bpo-4272.)

    Files in gzip format can be padded with trailing zero bytes; the gzip module will now consume these trailing bytes. (Fixed by Tadek Pietraszek and Brian Curtin; bpo-2846.)

  • New attribute: the hashlib module now has an algorithms attribute containing a tuple naming the supported algorithms. In Python 2.7, hashlib.algorithms contains ('md5', 'sha1', 'sha224', 'sha256', 'sha384', 'sha512'). (Contributed by Carl Chenet; bpo-7418.)

  • The default HTTPResponse class used by the httplib module now supports buffering, resulting in much faster reading of HTTP responses. (Contributed by Kristján Valur Jónsson; bpo-4879.)

    The HTTPConnection and HTTPSConnection classes now support a source_address parameter, a (host, port) 2-tuple giving the source address that will be used for the connection. (Contributed by Eldon Ziegler; bpo-3972.)

  • The ihooks module now supports relative imports. Note that ihooks is an older module for customizing imports, superseded by the imputil module added in Python 2.0. (Relative import support added by Neil Schemenauer.)

  • The imaplib module now supports IPv6 addresses. (Contributed by Derek Morr; bpo-1655.)

  • New function: the inspect module’s getcallargs() takes a callable and its positional and keyword arguments, and figures out which of the callable’s parameters will receive each argument, returning a dictionary mapping argument names to their values. For example:

    >>> from inspect import getcallargs
    >>> def f(a, b=1, *pos, **named):
    ...     pass
    >>> getcallargs(f, 1, 2, 3)
    {'a': 1, 'b': 2, 'pos': (3,), 'named': {}}
    >>> getcallargs(f, a=2, x=4)
    {'a': 2, 'b': 1, 'pos': (), 'named': {'x': 4}}
    >>> getcallargs(f)
    Traceback (most recent call last):
    ...
    TypeError: f() takes at least 1 argument (0 given)
    

    Contributed by George Sakkis; bpo-3135.

  • Updated module: The io library has been upgraded to the version shipped with Python 3.1. For 3.1, the I/O library was entirely rewritten in C and is 2 to 20 times faster depending on the task being performed. The original Python version was renamed to the _pyio module.

    One minor resulting change: the io.TextIOBase class now has an errors attribute giving the error setting used for encoding and decoding errors (one of 'strict', 'replace', 'ignore').

    The io.FileIO class now raises an OSError when passed an invalid file descriptor. (Implemented by Benjamin Peterson; bpo-4991.) The truncate() method now preserves the file position; previously it would change the file position to the end of the new file. (Fixed by Pascal Chambon; bpo-6939.)

  • New function: itertools.compress(data, selectors) takes two iterators. Elements of data are returned if the corresponding value in selectors is true:

    itertools.compress('ABCDEF', [1,0,1,0,1,1]) =>
      A, C, E, F
    

    New function: itertools.combinations_with_replacement(iter, r) returns all the possible r-length combinations of elements from the iterable iter. Unlike combinations(), individual elements can be repeated in the generated combinations:

    itertools.combinations_with_replacement('abc', 2) =>
      ('a', 'a'), ('a', 'b'), ('a', 'c'),
      ('b', 'b'), ('b', 'c'), ('c', 'c')
    

    Note that elements are treated as unique depending on their position in the input, not their actual values.

    The itertools.count() function now has a step argument that allows incrementing by values other than 1. count() also now allows keyword arguments, and using non-integer values such as floats or Decimal instances. (Implemented by Raymond Hettinger; bpo-5032.)

    itertools.combinations() and itertools.product() previously raised ValueError for values of r larger than the input iterable. This was deemed a specification error, so they now return an empty iterator. (Fixed by Raymond Hettinger; bpo-4816.)

  • Updated module: The json module was upgraded to version 2.0.9 of the simplejson package, which includes a C extension that makes encoding and decoding faster. (Contributed by Bob Ippolito; bpo-4136.)

    To support the new collections.OrderedDict type, json.load() now has an optional object_pairs_hook parameter that will be called with any object literal that decodes to a list of pairs. (Contributed by Raymond Hettinger; bpo-5381.)

  • The mailbox module’s Maildir class now records the timestamp on the directories it reads, and only re-reads them if the modification time has subsequently changed. This improves performance by avoiding unneeded directory scans. (Fixed by A.M. Kuchling and Antoine Pitrou; bpo-1607951, bpo-6896.)

  • New functions: the math module gained erf() and erfc() for the error function and the complementary error function, expm1() which computes e**x - 1 with more precision than using exp() and subtracting 1, gamma() for the Gamma function, and lgamma() for the natural log of the Gamma function. (Contributed by Mark Dickinson and nirinA raseliarison; bpo-3366.)

  • The multiprocessing module’s Manager* classes can now be passed a callable that will be called whenever a subprocess is started, along with a set of arguments that will be passed to the callable. (Contributed by lekma; bpo-5585.)

    The Pool class, which controls a pool of worker processes, now has an optional maxtasksperchild parameter. Worker processes will perform the specified number of tasks and then exit, causing the Pool to start a new worker. This is useful if tasks may leak memory or other resources, or if some tasks will cause the worker to become very large. (Contributed by Charles Cazabon; bpo-6963.)

  • The nntplib module now supports IPv6 addresses. (Contributed by Derek Morr; bpo-1664.)

  • New functions: the os module wraps the following POSIX system calls: getresgid() and getresuid(), which return the real, effective, and saved GIDs and UIDs; setresgid() and setresuid(), which set real, effective, and saved GIDs and UIDs to new values; initgroups(), which initialize the group access list for the current process. (GID/UID functions contributed by Travis H.; bpo-6508. Support for initgroups added by Jean-Paul Calderone; bpo-7333.)

    The os.fork() function now re-initializes the import lock in the child process; this fixes problems on Solaris when fork() is called from a thread. (Fixed by Zsolt Cserna; bpo-7242.)

  • In the os.path module, the normpath() and abspath() functions now preserve Unicode; if their input path is a Unicode string, the return value is also a Unicode string. (normpath() fixed by Matt Giuca in bpo-5827; abspath() fixed by Ezio Melotti in bpo-3426.)

  • The pydoc module now has help for the various symbols that Python uses. You can now do help('<<') or help('@'), for example. (Contributed by David Laban; bpo-4739.)

  • The re module’s split(), sub(), and subn() now accept an optional flags argument, for consistency with the other functions in the module. (Added by Gregory P. Smith.)

  • New function: run_path() in the runpy module will execute the code at a provided path argument. path can be the path of a Python source file (example.py), a compiled bytecode file (example.pyc), a directory (./package/), or a zip archive (example.zip). If a directory or zip path is provided, it will be added to the front of sys.path and the module __main__ will be imported. It’s expected that the directory or zip contains a __main__.py; if it doesn’t, some other __main__.py might be imported from a location later in sys.path. This makes more of the machinery of runpy available to scripts that want to mimic the way Python’s command line processes an explicit path name. (Added by Nick Coghlan; bpo-6816.)

  • New function: in the shutil module, make_archive() takes a filename, archive type (zip or tar-format), and a directory path, and creates an archive containing the directory’s contents. (Added by Tarek Ziadé.)

    shutil’s copyfile() and copytree() functions now raise a SpecialFileError exception when asked to copy a named pipe. Previously the code would treat named pipes like a regular file by opening them for reading, and this would block indefinitely. (Fixed by Antoine Pitrou; bpo-3002.)

  • The signal module no longer re-installs the signal handler unless this is truly necessary, which fixes a bug that could make it impossible to catch the EINTR signal robustly. (Fixed by Charles-Francois Natali; bpo-8354.)

  • New functions: in the site module, three new functions return various site- and user-specific paths. getsitepackages() returns a list containing all global site-packages directories, getusersitepackages() returns the path of the user’s site-packages directory, and getuserbase() returns the value of the USER_BASE environment variable, giving the path to a directory that can be used to store data. (Contributed by Tarek Ziadé; bpo-6693.)

    The site module now reports exceptions occurring when the sitecustomize module is imported, and will no longer catch and swallow the KeyboardInterrupt exception. (Fixed by Victor Stinner; bpo-3137.)

  • The create_connection() function gained a source_address parameter, a (host, port) 2-tuple giving the source address that will be used for the connection. (Contributed by Eldon Ziegler; bpo-3972.)

    The recv_into() and recvfrom_into() methods will now write into objects that support the buffer API, most usefully the bytearray and memoryview objects. (Implemented by Antoine Pitrou; bpo-8104.)

  • The SocketServer module’s TCPServer class now supports socket timeouts and disabling the Nagle algorithm. The disable_nagle_algorithm class attribute defaults to False; if overridden to be true, new request connections will have the TCP_NODELAY option set to prevent buffering many small sends into a single TCP packet. The timeout class attribute can hold a timeout in seconds that will be applied to the request socket; if no request is received within that time, handle_timeout() will be called and handle_request() will return. (Contributed by Kristján Valur Jónsson; bpo-6192 and bpo-6267.)

  • Updated module: the sqlite3 module has been updated to version 2.6.0 of the pysqlite package. Version 2.6.0 includes a number of bugfixes, and adds the ability to load SQLite extensions from shared libraries. Call the enable_load_extension(True) method to enable extensions, and then call load_extension() to load a particular shared library. (Updated by Gerhard Häring.)

  • The ssl module’s SSLSocket objects now support the buffer API, which fixed a test suite failure (fix by Antoine Pitrou; bpo-7133) and automatically set OpenSSL’s SSL_MODE_AUTO_RETRY, which will prevent an error code being returned from recv() operations that trigger an SSL renegotiation (fix by Antoine Pitrou; bpo-8222).

    The ssl.wrap_socket() constructor function now takes a ciphers argument that’s a string listing the encryption algorithms to be allowed; the format of the string is described in the OpenSSL documentation. (Added by Antoine Pitrou; bpo-8322.)

    Another change makes the extension load all of OpenSSL’s ciphers and digest algorithms so that they’re all available. Some SSL certificates couldn’t be verified, reporting an «unknown algorithm» error. (Reported by Beda Kosata, and fixed by Antoine Pitrou; bpo-8484.)

    The version of OpenSSL being used is now available as the module attributes ssl.OPENSSL_VERSION (a string), ssl.OPENSSL_VERSION_INFO (a 5-tuple), and ssl.OPENSSL_VERSION_NUMBER (an integer). (Added by Antoine Pitrou; bpo-8321.)

  • The struct module will no longer silently ignore overflow errors when a value is too large for a particular integer format code (one of bBhHiIlLqQ); it now always raises a struct.error exception. (Changed by Mark Dickinson; bpo-1523.) The pack() function will also attempt to use __index__() to convert and pack non-integers before trying the __int__() method or reporting an error. (Changed by Mark Dickinson; bpo-8300.)

  • New function: the subprocess module’s check_output() runs a command with a specified set of arguments and returns the command’s output as a string when the command runs without error, or raises a CalledProcessError exception otherwise.

    >>> subprocess.check_output(['df', '-h', '.'])
    'Filesystem     Size   Used  Avail Capacity  Mounted on\n
    /dev/disk0s2    52G    49G   3.0G    94%    /\n'
    
    >>> subprocess.check_output(['df', '-h', '/bogus'])
      ...
    subprocess.CalledProcessError: Command '['df', '-h', '/bogus']' returned non-zero exit status 1
    

    (Contributed by Gregory P. Smith.)

    The subprocess module will now retry its internal system calls on receiving an EINTR signal. (Reported by several people; final patch by Gregory P. Smith in bpo-1068268.)

  • New function: is_declared_global() in the symtable module returns true for variables that are explicitly declared to be global, false for ones that are implicitly global. (Contributed by Jeremy Hylton.)

  • The syslog module will now use the value of sys.argv[0] as the identifier instead of the previous default value of 'python'. (Changed by Sean Reifschneider; bpo-8451.)

  • The sys.version_info value is now a named tuple, with attributes named major, minor, micro, releaselevel, and serial. (Contributed by Ross Light; bpo-4285.)

    sys.getwindowsversion() also returns a named tuple, with attributes named major, minor, build, platform, service_pack, service_pack_major, service_pack_minor, suite_mask, and product_type. (Contributed by Brian Curtin; bpo-7766.)

  • The tarfile module’s default error handling has changed, to no longer suppress fatal errors. The default error level was previously 0, which meant that errors would only result in a message being written to the debug log, but because the debug log is not activated by default, these errors go unnoticed. The default error level is now 1, which raises an exception if there’s an error. (Changed by Lars Gustäbel; bpo-7357.)

    tarfile now supports filtering the TarInfo objects being added to a tar file. When you call add(), you may supply an optional filter argument that’s a callable. The filter callable will be passed the TarInfo for every file being added, and can modify and return it. If the callable returns None, the file will be excluded from the resulting archive. This is more powerful than the existing exclude argument, which has therefore been deprecated. (Added by Lars Gustäbel; bpo-6856.) The TarFile class also now supports the context management protocol. (Added by Lars Gustäbel; bpo-7232.)

  • The wait() method of the threading.Event class now returns the internal flag on exit. This means the method will usually return true because wait() is supposed to block until the internal flag becomes true. The return value will only be false if a timeout was provided and the operation timed out. (Contributed by Tim Lesher; bpo-1674032.)

  • The Unicode database provided by the unicodedata module is now used internally to determine which characters are numeric, whitespace, or represent line breaks. The database also includes information from the Unihan.txt data file (patch by Anders Chrigström and Amaury Forgeot d’Arc; bpo-1571184) and has been updated to version 5.2.0 (updated by Florent Xicluna; bpo-8024).

  • The urlparse module’s urlsplit() now handles unknown URL schemes in a fashion compliant with RFC 3986: if the URL is of the form "<something>://...", the text before the :// is treated as the scheme, even if it’s a made-up scheme that the module doesn’t know about. This change may break code that worked around the old behaviour. For example, Python 2.6.4 or 2.5 will return the following:

    >>> import urlparse
    >>> urlparse.urlsplit('invented://host/filename?query')
    ('invented', '', '//host/filename?query', '', '')
    

    Python 2.7 (and Python 2.6.5) will return:

    >>> import urlparse
    >>> urlparse.urlsplit('invented://host/filename?query')
    ('invented', 'host', '/filename?query', '', '')
    

    (Python 2.7 actually produces slightly different output, since it returns a named tuple instead of a standard tuple.)

    The urlparse module also supports IPv6 literal addresses as defined by RFC 2732 (contributed by Senthil Kumaran; bpo-2987).

    >>> urlparse.urlparse('http://[1080::8:800:200C:417A]/foo')
    ParseResult(scheme='http', netloc='[1080::8:800:200C:417A]',
                path='/foo', params='', query='', fragment='')
    
  • New class: the WeakSet class in the weakref module is a set that only holds weak references to its elements; elements will be removed once there are no references pointing to them. (Originally implemented in Python 3.x by Raymond Hettinger, and backported to 2.7 by Michael Foord.)

  • The ElementTree library, xml.etree, no longer escapes ampersands and angle brackets when outputting an XML processing instruction (which looks like <?xml-stylesheet href="#style1"?>) or comment (which looks like <!-- comment -->). (Patch by Neil Muller; bpo-2746.)

  • The XML-RPC client and server, provided by the xmlrpclib and SimpleXMLRPCServer modules, have improved performance by supporting HTTP/1.1 keep-alive and by optionally using gzip encoding to compress the XML being exchanged. The gzip compression is controlled by the encode_threshold attribute of SimpleXMLRPCRequestHandler, which contains a size in bytes; responses larger than this will be compressed. (Contributed by Kristján Valur Jónsson; bpo-6267.)

  • The zipfile module’s ZipFile now supports the context management protocol, so you can write with zipfile.ZipFile(...) as f:. (Contributed by Brian Curtin; bpo-5511.)

    zipfile now also supports archiving empty directories and extracts them correctly. (Fixed by Kuba Wieczorek; bpo-4710.) Reading files out of an archive is faster, and interleaving read() and readline() now works correctly. (Contributed by Nir Aides; bpo-7610.)

    The is_zipfile() function now accepts a file object, in addition to the path names accepted in earlier versions. (Contributed by Gabriel Genellina; bpo-4756.)

    The writestr() method now has an optional compress_type parameter that lets you override the default compression method specified in the ZipFile constructor. (Contributed by Ronald Oussoren; bpo-6003.)

New module: importlib

Python 3.1 includes the importlib package, a re-implementation of the logic underlying Python’s import statement. importlib is useful for implementors of Python interpreters and to users who wish to write new importers that can participate in the import process. Python 2.7 doesn’t contain the complete importlib package, but instead has a tiny subset that contains a single function, import_module().

import_module(name, package=None) imports a module. name is a string containing the module or package’s name. It’s possible to do relative imports by providing a string that begins with a . character, such as ..utils.errors. For relative imports, the package argument must be provided and is the name of the package that will be used as the anchor for the relative import. import_module() both inserts the imported module into sys.modules and returns the module object.

Here are some examples:

>>> from importlib import import_module
>>> anydbm = import_module('anydbm')  # Standard absolute import
>>> anydbm
<module 'anydbm' from '/p/python/Lib/anydbm.py'>
>>> # Relative import
>>> file_util = import_module('..file_util', 'distutils.command')
>>> file_util
<module 'distutils.file_util' from '/python/Lib/distutils/file_util.pyc'>

importlib was implemented by Brett Cannon and introduced in Python 3.1.

New module: sysconfig

The sysconfig module has been pulled out of the Distutils package, becoming a new top-level module in its own right. sysconfig provides functions for getting information about Python’s build process: compiler switches, installation paths, the platform name, and whether Python is running from its source directory.

Some of the functions in the module are:

  • get_config_var() returns variables from Python’s Makefile and the pyconfig.h file.

  • get_config_vars() returns a dictionary containing all of the configuration variables.

  • get_path() returns the configured path for a particular type of module: the standard library, site-specific modules, platform-specific modules, etc.

  • is_python_build() returns true if you’re running a binary from a Python source tree, and false otherwise.

Consult the sysconfig documentation for more details and for a complete list of functions.

The Distutils package and sysconfig are now maintained by Tarek Ziadé, who has also started a Distutils2 package (source repository at https://hg.python.org/distutils2/) for developing a next-generation version of Distutils.

ttk: Themed Widgets for Tk

Tcl/Tk 8.5 includes a set of themed widgets that re-implement basic Tk widgets but have a more customizable appearance and can therefore more closely resemble the native platform’s widgets. This widget set was originally called Tile, but was renamed to Ttk (for «themed Tk») on being added to Tcl/Tck release 8.5.

To learn more, read the ttk module documentation. You may also wish to read the Tcl/Tk manual page describing the Ttk theme engine, available at https://www.tcl.tk/man/tcl8.5/TkCmd/ttk_intro.htm. Some screenshots of the Python/Ttk code in use are at https://code.google.com/archive/p/python-ttk/wikis/Screenshots.wiki.

The ttk module was written by Guilherme Polo and added in bpo-2983. An alternate version called Tile.py, written by Martin Franklin and maintained by Kevin Walzer, was proposed for inclusion in bpo-2618, but the authors argued that Guilherme Polo’s work was more comprehensive.

Updated module: unittest

The unittest module was greatly enhanced; many new features were added. Most of these features were implemented by Michael Foord, unless otherwise noted. The enhanced version of the module is downloadable separately for use with Python versions 2.4 to 2.6, packaged as the unittest2 package, from https://pypi.org/project/unittest2.

When used from the command line, the module can automatically discover tests. It’s not as fancy as py.test or nose, but provides a simple way to run tests kept within a set of package directories. For example, the following command will search the test/ subdirectory for any importable test files named test*.py:

python -m unittest discover -s test

Consult the unittest module documentation for more details. (Developed in bpo-6001.)

The main() function supports some other new options:

  • -b or --buffer will buffer the standard output and standard error streams during each test. If the test passes, any resulting output will be discarded; on failure, the buffered output will be displayed.

  • -c or --catch will cause the control-C interrupt to be handled more gracefully. Instead of interrupting the test process immediately, the currently running test will be completed and then the partial results up to the interruption will be reported. If you’re impatient, a second press of control-C will cause an immediate interruption.

    This control-C handler tries to avoid causing problems when the code being tested or the tests being run have defined a signal handler of their own, by noticing that a signal handler was already set and calling it. If this doesn’t work for you, there’s a removeHandler() decorator that can be used to mark tests that should have the control-C handling disabled.

  • -f or --failfast makes test execution stop immediately when a test fails instead of continuing to execute further tests. (Suggested by Cliff Dyer and implemented by Michael Foord; bpo-8074.)

The progress messages now show “x” for expected failures and “u” for unexpected successes when run in verbose mode. (Contributed by Benjamin Peterson.)

Test cases can raise the SkipTest exception to skip a test (bpo-1034053).

The error messages for assertEqual(), assertTrue(), and assertFalse() failures now provide more information. If you set the longMessage attribute of your TestCase classes to true, both the standard error message and any additional message you provide will be printed for failures. (Added by Michael Foord; bpo-5663.)

The assertRaises() method now returns a context handler when called without providing a callable object to run. For example, you can write this:

with self.assertRaises(KeyError):
    {}['foo']

(Implemented by Antoine Pitrou; bpo-4444.)

Module- and class-level setup and teardown fixtures are now supported. Modules can contain setUpModule() and tearDownModule() functions. Classes can have setUpClass() and tearDownClass() methods that must be defined as class methods (using @classmethod or equivalent). These functions and methods are invoked when the test runner switches to a test case in a different module or class.

The methods addCleanup() and doCleanups() were added. addCleanup() lets you add cleanup functions that will be called unconditionally (after setUp() if setUp() fails, otherwise after tearDown()). This allows for much simpler resource allocation and deallocation during tests (bpo-5679).

A number of new methods were added that provide more specialized tests. Many of these methods were written by Google engineers for use in their test suites; Gregory P. Smith, Michael Foord, and GvR worked on merging them into Python’s version of unittest.

  • assertIsNone() and assertIsNotNone() take one expression and verify that the result is or is not None.

  • assertIs() and assertIsNot() take two values and check whether the two values evaluate to the same object or not. (Added by Michael Foord; bpo-2578.)

  • assertIsInstance() and assertNotIsInstance() check whether the resulting object is an instance of a particular class, or of one of a tuple of classes. (Added by Georg Brandl; bpo-7031.)

  • assertGreater(), assertGreaterEqual(), assertLess(), and assertLessEqual() compare two quantities.

  • assertMultiLineEqual() compares two strings, and if they’re not equal, displays a helpful comparison that highlights the differences in the two strings. This comparison is now used by default when Unicode strings are compared with assertEqual().

  • assertRegexpMatches() and assertNotRegexpMatches() checks whether the first argument is a string matching or not matching the regular expression provided as the second argument (bpo-8038).

  • assertRaisesRegexp() checks whether a particular exception is raised, and then also checks that the string representation of the exception matches the provided regular expression.

  • assertIn() and assertNotIn() tests whether first is or is not in second.

  • assertItemsEqual() tests whether two provided sequences contain the same elements.

  • assertSetEqual() compares whether two sets are equal, and only reports the differences between the sets in case of error.

  • Similarly, assertListEqual() and assertTupleEqual() compare the specified types and explain any differences without necessarily printing their full values; these methods are now used by default when comparing lists and tuples using assertEqual(). More generally, assertSequenceEqual() compares two sequences and can optionally check whether both sequences are of a particular type.

  • assertDictEqual() compares two dictionaries and reports the differences; it’s now used by default when you compare two dictionaries using assertEqual(). assertDictContainsSubset() checks whether all of the key/value pairs in first are found in second.

  • assertAlmostEqual() and assertNotAlmostEqual() test whether first and second are approximately equal. This method can either round their difference to an optionally-specified number of places (the default is 7) and compare it to zero, or require the difference to be smaller than a supplied delta value.

  • loadTestsFromName() properly honors the suiteClass attribute of the TestLoader. (Fixed by Mark Roddy; bpo-6866.)

  • A new hook lets you extend the assertEqual() method to handle new data types. The addTypeEqualityFunc() method takes a type object and a function. The function will be used when both of the objects being compared are of the specified type. This function should compare the two objects and raise an exception if they don’t match; it’s a good idea for the function to provide additional information about why the two objects aren’t matching, much as the new sequence comparison methods do.

unittest.main() now takes an optional exit argument. If false, main() doesn’t call sys.exit(), allowing main() to be used from the interactive interpreter. (Contributed by J. Pablo Fernández; bpo-3379.)

TestResult has new startTestRun() and stopTestRun() methods that are called immediately before and after a test run. (Contributed by Robert Collins; bpo-5728.)

With all these changes, the unittest.py was becoming awkwardly large, so the module was turned into a package and the code split into several files (by Benjamin Peterson). This doesn’t affect how the module is imported or used.

Ver también

http://www.voidspace.org.uk/python/articles/unittest2.shtml

Describes the new features, how to use them, and the rationale for various design decisions. (By Michael Foord.)

Updated module: ElementTree 1.3

The version of the ElementTree library included with Python was updated to version 1.3. Some of the new features are:

  • The various parsing functions now take a parser keyword argument giving an XMLParser instance that will be used. This makes it possible to override the file’s internal encoding:

    p = ET.XMLParser(encoding='utf-8')
    t = ET.XML("""<root/>""", parser=p)
    

    Errors in parsing XML now raise a ParseError exception, whose instances have a position attribute containing a (line, column) tuple giving the location of the problem.

  • ElementTree’s code for converting trees to a string has been significantly reworked, making it roughly twice as fast in many cases. The ElementTree.write() and Element.write() methods now have a method parameter that can be «xml» (the default), «html», or «text». HTML mode will output empty elements as <empty></empty> instead of <empty/>, and text mode will skip over elements and only output the text chunks. If you set the tag attribute of an element to None but leave its children in place, the element will be omitted when the tree is written out, so you don’t need to do more extensive rearrangement to remove a single element.

    Namespace handling has also been improved. All xmlns:<whatever> declarations are now output on the root element, not scattered throughout the resulting XML. You can set the default namespace for a tree by setting the default_namespace attribute and can register new prefixes with register_namespace(). In XML mode, you can use the true/false xml_declaration parameter to suppress the XML declaration.

  • New Element method: extend() appends the items from a sequence to the element’s children. Elements themselves behave like sequences, so it’s easy to move children from one element to another:

    from xml.etree import ElementTree as ET
    
    t = ET.XML("""<list>
      <item>1</item> <item>2</item>  <item>3</item>
    </list>""")
    new = ET.XML('<root/>')
    new.extend(t)
    
    # Outputs <root><item>1</item>...</root>
    print ET.tostring(new)
    
  • New Element method: iter() yields the children of the element as a generator. It’s also possible to write for child in elem: to loop over an element’s children. The existing method getiterator() is now deprecated, as is getchildren() which constructs and returns a list of children.

  • New Element method: itertext() yields all chunks of text that are descendants of the element. For example:

    t = ET.XML("""<list>
      <item>1</item> <item>2</item>  <item>3</item>
    </list>""")
    
    # Outputs ['\n  ', '1', ' ', '2', '  ', '3', '\n']
    print list(t.itertext())
    
  • Deprecated: using an element as a Boolean (i.e., if elem:) would return true if the element had any children, or false if there were no children. This behaviour is confusing – None is false, but so is a childless element? – so it will now trigger a FutureWarning. In your code, you should be explicit: write len(elem) != 0 if you’re interested in the number of children, or elem is not None.

Fredrik Lundh develops ElementTree and produced the 1.3 version; you can read his article describing 1.3 at http://effbot.org/zone/elementtree-13-intro.htm. Florent Xicluna updated the version included with Python, after discussions on python-dev and in bpo-6472.)

Build and C API Changes

Changes to Python’s build process and to the C API include:

  • The latest release of the GNU Debugger, GDB 7, can be scripted using Python. When you begin debugging an executable program P, GDB will look for a file named P-gdb.py and automatically read it. Dave Malcolm contributed a python-gdb.py that adds a number of commands useful when debugging Python itself. For example, py-up and py-down go up or down one Python stack frame, which usually corresponds to several C stack frames. py-print prints the value of a Python variable, and py-bt prints the Python stack trace. (Added as a result of bpo-8032.)

  • If you use the .gdbinit file provided with Python, the «pyo» macro in the 2.7 version now works correctly when the thread being debugged doesn’t hold the GIL; the macro now acquires it before printing. (Contributed by Victor Stinner; bpo-3632.)

  • Py_AddPendingCall() is now thread-safe, letting any worker thread submit notifications to the main Python thread. This is particularly useful for asynchronous IO operations. (Contributed by Kristján Valur Jónsson; bpo-4293.)

  • New function: PyCode_NewEmpty() creates an empty code object; only the filename, function name, and first line number are required. This is useful for extension modules that are attempting to construct a more useful traceback stack. Previously such extensions needed to call PyCode_New(), which had many more arguments. (Added by Jeffrey Yasskin.)

  • New function: PyErr_NewExceptionWithDoc() creates a new exception class, just as the existing PyErr_NewException() does, but takes an extra char * argument containing the docstring for the new exception class. (Added by “lekma” on the Python bug tracker; bpo-7033.)

  • New function: PyFrame_GetLineNumber() takes a frame object and returns the line number that the frame is currently executing. Previously code would need to get the index of the bytecode instruction currently executing, and then look up the line number corresponding to that address. (Added by Jeffrey Yasskin.)

  • New functions: PyLong_AsLongAndOverflow() and PyLong_AsLongLongAndOverflow() approximates a Python long integer as a C long or long long. If the number is too large to fit into the output type, an overflow flag is set and returned to the caller. (Contributed by Case Van Horsen; bpo-7528 and bpo-7767.)

  • New function: stemming from the rewrite of string-to-float conversion, a new PyOS_string_to_double() function was added. The old PyOS_ascii_strtod() and PyOS_ascii_atof() functions are now deprecated.

  • New function: PySys_SetArgvEx() sets the value of sys.argv and can optionally update sys.path to include the directory containing the script named by sys.argv[0] depending on the value of an updatepath parameter.

    This function was added to close a security hole for applications that embed Python. The old function, PySys_SetArgv(), would always update sys.path, and sometimes it would add the current directory. This meant that, if you ran an application embedding Python in a directory controlled by someone else, attackers could put a Trojan-horse module in the directory (say, a file named os.py) that your application would then import and run.

    If you maintain a C/C++ application that embeds Python, check whether you’re calling PySys_SetArgv() and carefully consider whether the application should be using PySys_SetArgvEx() with updatepath set to false.

    Security issue reported as CVE-2008-5983; discussed in bpo-5753, and fixed by Antoine Pitrou.

  • New macros: the Python header files now define the following macros: Py_ISALNUM, Py_ISALPHA, Py_ISDIGIT, Py_ISLOWER, Py_ISSPACE, Py_ISUPPER, Py_ISXDIGIT, Py_TOLOWER, and Py_TOUPPER. All of these functions are analogous to the C standard macros for classifying characters, but ignore the current locale setting, because in several places Python needs to analyze characters in a locale-independent way. (Added by Eric Smith; bpo-5793.)

  • Removed function: PyEval_CallObject is now only available as a macro. A function version was being kept around to preserve ABI linking compatibility, but that was in 1997; it can certainly be deleted by now. (Removed by Antoine Pitrou; bpo-8276.)

  • New format codes: the PyFormat_FromString(), PyFormat_FromStringV(), and PyErr_Format() functions now accept %lld and %llu format codes for displaying C’s long long types. (Contributed by Mark Dickinson; bpo-7228.)

  • The complicated interaction between threads and process forking has been changed. Previously, the child process created by os.fork() might fail because the child is created with only a single thread running, the thread performing the os.fork(). If other threads were holding a lock, such as Python’s import lock, when the fork was performed, the lock would still be marked as «held» in the new process. But in the child process nothing would ever release the lock, since the other threads weren’t replicated, and the child process would no longer be able to perform imports.

    Python 2.7 acquires the import lock before performing an os.fork(), and will also clean up any locks created using the threading module. C extension modules that have internal locks, or that call fork() themselves, will not benefit from this clean-up.

    (Fixed by Thomas Wouters; bpo-1590864.)

  • The Py_Finalize() function now calls the internal threading._shutdown() function; this prevents some exceptions from being raised when an interpreter shuts down. (Patch by Adam Olsen; bpo-1722344.)

  • When using the PyMemberDef structure to define attributes of a type, Python will no longer let you try to delete or set a T_STRING_INPLACE attribute.

  • Global symbols defined by the ctypes module are now prefixed with Py, or with _ctypes. (Implemented by Thomas Heller; bpo-3102.)

  • New configure option: the --with-system-expat switch allows building the pyexpat module to use the system Expat library. (Contributed by Arfrever Frehtes Taifersar Arahesis; bpo-7609.)

  • New configure option: the --with-valgrind option will now disable the pymalloc allocator, which is difficult for the Valgrind memory-error detector to analyze correctly. Valgrind will therefore be better at detecting memory leaks and overruns. (Contributed by James Henstridge; bpo-2422.)

  • New configure option: you can now supply an empty string to --with-dbmliborder= in order to disable all of the various DBM modules. (Added by Arfrever Frehtes Taifersar Arahesis; bpo-6491.)

  • The configure script now checks for floating-point rounding bugs on certain 32-bit Intel chips and defines a X87_DOUBLE_ROUNDING preprocessor definition. No code currently uses this definition, but it’s available if anyone wishes to use it. (Added by Mark Dickinson; bpo-2937.)

    configure also now sets a LDCXXSHARED Makefile variable for supporting C++ linking. (Contributed by Arfrever Frehtes Taifersar Arahesis; bpo-1222585.)

  • The build process now creates the necessary files for pkg-config support. (Contributed by Clinton Roy; bpo-3585.)

  • The build process now supports Subversion 1.7. (Contributed by Arfrever Frehtes Taifersar Arahesis; bpo-6094.)

Capsules

Python 3.1 adds a new C datatype, PyCapsule, for providing a C API to an extension module. A capsule is essentially the holder of a C void * pointer, and is made available as a module attribute; for example, the socket module’s API is exposed as socket.CAPI, and unicodedata exposes ucnhash_CAPI. Other extensions can import the module, access its dictionary to get the capsule object, and then get the void * pointer, which will usually point to an array of pointers to the module’s various API functions.

There is an existing data type already used for this, PyCObject, but it doesn’t provide type safety. Evil code written in pure Python could cause a segmentation fault by taking a PyCObject from module A and somehow substituting it for the PyCObject in module B. Capsules know their own name, and getting the pointer requires providing the name:

void *vtable;

if (!PyCapsule_IsValid(capsule, "mymodule.CAPI") {
        PyErr_SetString(PyExc_ValueError, "argument type invalid");
        return NULL;
}

vtable = PyCapsule_GetPointer(capsule, "mymodule.CAPI");

You are assured that vtable points to whatever you’re expecting. If a different capsule was passed in, PyCapsule_IsValid() would detect the mismatched name and return false. Refer to Proporcionar una API C para un módulo de extensión for more information on using these objects.

Python 2.7 now uses capsules internally to provide various extension-module APIs, but the PyCObject_AsVoidPtr() was modified to handle capsules, preserving compile-time compatibility with the CObject interface. Use of PyCObject_AsVoidPtr() will signal a PendingDeprecationWarning, which is silent by default.

Implemented in Python 3.1 and backported to 2.7 by Larry Hastings; discussed in bpo-5630.

Port-Specific Changes: Windows

  • The msvcrt module now contains some constants from the crtassem.h header file: CRT_ASSEMBLY_VERSION, VC_ASSEMBLY_PUBLICKEYTOKEN, and LIBRARIES_ASSEMBLY_NAME_PREFIX. (Contributed by David Cournapeau; bpo-4365.)

  • The _winreg module for accessing the registry now implements the CreateKeyEx() and DeleteKeyEx() functions, extended versions of previously-supported functions that take several extra arguments. The DisableReflectionKey(), EnableReflectionKey(), and QueryReflectionKey() were also tested and documented. (Implemented by Brian Curtin: bpo-7347.)

  • The new _beginthreadex() API is used to start threads, and the native thread-local storage functions are now used. (Contributed by Kristján Valur Jónsson; bpo-3582.)

  • The os.kill() function now works on Windows. The signal value can be the constants CTRL_C_EVENT, CTRL_BREAK_EVENT, or any integer. The first two constants will send Control-C and Control-Break keystroke events to subprocesses; any other value will use the TerminateProcess() API. (Contributed by Miki Tebeka; bpo-1220212.)

  • The os.listdir() function now correctly fails for an empty path. (Fixed by Hirokazu Yamamoto; bpo-5913.)

  • The mimelib module will now read the MIME database from the Windows registry when initializing. (Patch by Gabriel Genellina; bpo-4969.)

Port-Specific Changes: Mac OS X

  • The path /Library/Python/2.7/site-packages is now appended to sys.path, in order to share added packages between the system installation and a user-installed copy of the same version. (Changed by Ronald Oussoren; bpo-4865.)

    Distinto en la versión 2.7.13: As of 2.7.13, this change was removed. /Library/Python/2.7/site-packages, the site-packages directory used by the Apple-supplied system Python 2.7 is no longer appended to sys.path for user-installed Pythons such as from the python.org installers. As of macOS 10.12, Apple changed how the system site-packages directory is configured, which could cause installation of pip components, like setuptools, to fail. Packages installed for the system Python will no longer be shared with user-installed Pythons. (bpo-28440)

Port-Specific Changes: FreeBSD

  • FreeBSD 7.1’s SO_SETFIB constant, used with getsockopt()/setsockopt() to select an alternate routing table, is now available in the socket module. (Added by Kyle VanderBeek; bpo-8235.)

Other Changes and Fixes

  • Two benchmark scripts, iobench and ccbench, were added to the Tools directory. iobench measures the speed of the built-in file I/O objects returned by open() while performing various operations, and ccbench is a concurrency benchmark that tries to measure computing throughput, thread switching latency, and IO processing bandwidth when performing several tasks using a varying number of threads.

  • The Tools/i18n/msgfmt.py script now understands plural forms in .po files. (Fixed by Martin von Löwis; bpo-5464.)

  • When importing a module from a .pyc or .pyo file with an existing .py counterpart, the co_filename attributes of the resulting code objects are overwritten when the original filename is obsolete. This can happen if the file has been renamed, moved, or is accessed through different paths. (Patch by Ziga Seilnacht and Jean-Paul Calderone; bpo-1180193.)

  • The regrtest.py script now takes a --randseed= switch that takes an integer that will be used as the random seed for the -r option that executes tests in random order. The -r option also reports the seed that was used (Added by Collin Winter.)

  • Another regrtest.py switch is -j, which takes an integer specifying how many tests run in parallel. This allows reducing the total runtime on multi-core machines. This option is compatible with several other options, including the -R switch which is known to produce long runtimes. (Added by Antoine Pitrou, bpo-6152.) This can also be used with a new -F switch that runs selected tests in a loop until they fail. (Added by Antoine Pitrou; bpo-7312.)

  • When executed as a script, the py_compile.py module now accepts '-' as an argument, which will read standard input for the list of filenames to be compiled. (Contributed by Piotr Ożarowski; bpo-8233.)

Porting to Python 2.7

This section lists previously described changes and other bugfixes that may require changes to your code:

  • The range() function processes its arguments more consistently; it will now call __int__() on non-float, non-integer arguments that are supplied to it. (Fixed by Alexander Belopolsky; bpo-1533.)

  • The string format() method changed the default precision used for floating-point and complex numbers from 6 decimal places to 12, which matches the precision used by str(). (Changed by Eric Smith; bpo-5920.)

  • Because of an optimization for the with statement, the special methods __enter__() and __exit__() must belong to the object’s type, and cannot be directly attached to the object’s instance. This affects new-style classes (derived from object) and C extension types. (bpo-6101.)

  • Due to a bug in Python 2.6, the exc_value parameter to __exit__() methods was often the string representation of the exception, not an instance. This was fixed in 2.7, so exc_value will be an instance as expected. (Fixed by Florent Xicluna; bpo-7853.)

  • When a restricted set of attributes were set using __slots__, deleting an unset attribute would not raise AttributeError as you would expect. Fixed by Benjamin Peterson; bpo-7604.)

In the standard library:

  • Operations with datetime instances that resulted in a year falling outside the supported range didn’t always raise OverflowError. Such errors are now checked more carefully and will now raise the exception. (Reported by Mark Leander, patch by Anand B. Pillai and Alexander Belopolsky; bpo-7150.)

  • When using Decimal instances with a string’s format() method, the default alignment was previously left-alignment. This has been changed to right-alignment, which might change the output of your programs. (Changed by Mark Dickinson; bpo-6857.)

    Comparisons involving a signaling NaN value (or sNAN) now signal InvalidOperation instead of silently returning a true or false value depending on the comparison operator. Quiet NaN values (or NaN) are now hashable. (Fixed by Mark Dickinson; bpo-7279.)

  • The ElementTree library, xml.etree, no longer escapes ampersands and angle brackets when outputting an XML processing instruction (which looks like <?xml-stylesheet href=»#style1»?>) or comment (which looks like <!– comment –>). (Patch by Neil Muller; bpo-2746.)

  • The readline() method of StringIO objects now does nothing when a negative length is requested, as other file-like objects do. (bpo-7348).

  • The syslog module will now use the value of sys.argv[0] as the identifier instead of the previous default value of 'python'. (Changed by Sean Reifschneider; bpo-8451.)

  • The tarfile module’s default error handling has changed, to no longer suppress fatal errors. The default error level was previously 0, which meant that errors would only result in a message being written to the debug log, but because the debug log is not activated by default, these errors go unnoticed. The default error level is now 1, which raises an exception if there’s an error. (Changed by Lars Gustäbel; bpo-7357.)

  • The urlparse module’s urlsplit() now handles unknown URL schemes in a fashion compliant with RFC 3986: if the URL is of the form "<something>://...", the text before the :// is treated as the scheme, even if it’s a made-up scheme that the module doesn’t know about. This change may break code that worked around the old behaviour. For example, Python 2.6.4 or 2.5 will return the following:

    >>> import urlparse
    >>> urlparse.urlsplit('invented://host/filename?query')
    ('invented', '', '//host/filename?query', '', '')
    

    Python 2.7 (and Python 2.6.5) will return:

    >>> import urlparse
    >>> urlparse.urlsplit('invented://host/filename?query')
    ('invented', 'host', '/filename?query', '', '')
    

    (Python 2.7 actually produces slightly different output, since it returns a named tuple instead of a standard tuple.)

For C extensions:

  • C extensions that use integer format codes with the PyArg_Parse* family of functions will now raise a TypeError exception instead of triggering a DeprecationWarning (bpo-5080).

  • Use the new PyOS_string_to_double() function instead of the old PyOS_ascii_strtod() and PyOS_ascii_atof() functions, which are now deprecated.

For applications that embed Python:

New Features Added to Python 2.7 Maintenance Releases

New features may be added to Python 2.7 maintenance releases when the situation genuinely calls for it. Any such additions must go through the Python Enhancement Proposal process, and make a compelling case for why they can’t be adequately addressed by either adding the new feature solely to Python 3, or else by publishing it on the Python Package Index.

In addition to the specific proposals listed below, there is a general exemption allowing new -3 warnings to be added in any Python 2.7 maintenance release.

Two new environment variables for debug mode

In debug mode, the [xxx refs] statistic is not written by default, the PYTHONSHOWREFCOUNT environment variable now must also be set. (Contributed by Victor Stinner; bpo-31733.)

When Python is compiled with COUNT_ALLOC defined, allocation counts are no longer dumped by default anymore: the PYTHONSHOWALLOCCOUNT environment variable must now also be set. Moreover, allocation counts are now dumped into stderr, rather than stdout. (Contributed by Victor Stinner; bpo-31692.)

Nuevo en la versión 2.7.15.

PEP 434: IDLE Enhancement Exception for All Branches

PEP 434 describes a general exemption for changes made to the IDLE development environment shipped along with Python. This exemption makes it possible for the IDLE developers to provide a more consistent user experience across all supported versions of Python 2 and 3.

For details of any IDLE changes, refer to the NEWS file for the specific release.

PEP 466: Network Security Enhancements for Python 2.7

PEP 466 describes a number of network security enhancement proposals that have been approved for inclusion in Python 2.7 maintenance releases, with the first of those changes appearing in the Python 2.7.7 release.

PEP 466 related features added in Python 2.7.7:

  • hmac.compare_digest() was backported from Python 3 to make a timing attack resistant comparison operation available to Python 2 applications. (Contributed by Alex Gaynor; bpo-21306.)

  • OpenSSL 1.0.1g was upgraded in the official Windows installers published on python.org. (Contributed by Zachary Ware; bpo-21462.)

PEP 466 related features added in Python 2.7.8:

  • hashlib.pbkdf2_hmac() was backported from Python 3 to make a hashing algorithm suitable for secure password storage broadly available to Python 2 applications. (Contributed by Alex Gaynor; bpo-21304.)

  • OpenSSL 1.0.1h was upgraded for the official Windows installers published on python.org. (contributed by Zachary Ware in bpo-21671 for CVE-2014-0224)

PEP 466 related features added in Python 2.7.9:

  • Most of Python 3.4’s ssl module was backported. This means ssl now supports Server Name Indication, TLS1.x settings, access to the platform certificate store, the SSLContext class, and other features. (Contributed by Alex Gaynor and David Reid; bpo-21308.)

    Refer to the «Version added: 2.7.9» notes in the module documentation for specific details.

  • os.urandom() was changed to cache a file descriptor to /dev/urandom instead of reopening /dev/urandom on every call. (Contributed by Alex Gaynor; bpo-21305.)

  • hashlib.algorithms_guaranteed and hashlib.algorithms_available were backported from Python 3 to make it easier for Python 2 applications to select the strongest available hash algorithm. (Contributed by Alex Gaynor in bpo-21307)

PEP 477: Backport ensurepip (PEP 453) to Python 2.7

PEP 477 approves the inclusion of the PEP 453 ensurepip module and the improved documentation that was enabled by it in the Python 2.7 maintenance releases, appearing first in the Python 2.7.9 release.

Bootstrapping pip By Default

The new ensurepip module (defined in PEP 453) provides a standard cross-platform mechanism to bootstrap the pip installer into Python installations. The version of pip included with Python 2.7.9 is pip 1.5.6, and future 2.7.x maintenance releases will update the bundled version to the latest version of pip that is available at the time of creating the release candidate.

By default, the commands pip, pipX and pipX.Y will be installed on all platforms (where X.Y stands for the version of the Python installation), along with the pip Python package and its dependencies.

For CPython source builds on POSIX systems, the make install and make altinstall commands do not bootstrap pip by default. This behaviour can be controlled through configure options, and overridden through Makefile options.

On Windows and Mac OS X, the CPython installers now default to installing pip along with CPython itself (users may opt out of installing it during the installation process). Window users will need to opt in to the automatic PATH modifications to have pip available from the command line by default, otherwise it can still be accessed through the Python launcher for Windows as py -m pip.

As discussed in the PEP, platform packagers may choose not to install these commands by default, as long as, when invoked, they provide clear and simple directions on how to install them on that platform (usually using the system package manager).

Documentation Changes

As part of this change, the Instalando módulos de Python and Distribuir módulos de Python sections of the documentation have been completely redesigned as short getting started and FAQ documents. Most packaging documentation has now been moved out to the Python Packaging Authority maintained Python Packaging User Guide and the documentation of the individual projects.

However, as this migration is currently still incomplete, the legacy versions of those guides remaining available as Instalación de Módulos de Python (versión antigua) and Distribución de módulos de Python (versión heredada).

Ver también

PEP 453 – Explicit bootstrapping of pip in Python installations

PEP written by Donald Stufft and Nick Coghlan, implemented by Donald Stufft, Nick Coghlan, Martin von Löwis and Ned Deily.

PEP 476: Enabling certificate verification by default for stdlib http clients

PEP 476 updated httplib and modules which use it, such as urllib2 and xmlrpclib, to now verify that the server presents a certificate which is signed by a Certificate Authority in the platform trust store and whose hostname matches the hostname being requested by default, significantly improving security for many applications. This change was made in the Python 2.7.9 release.

For applications which require the old previous behavior, they can pass an alternate context:

import urllib2
import ssl

# This disables all verification
context = ssl._create_unverified_context()

# This allows using a specific certificate for the host, which doesn't need
# to be in the trust store
context = ssl.create_default_context(cafile="/path/to/file.crt")

urllib2.urlopen("https://invalid-cert", context=context)

PEP 493: HTTPS verification migration tools for Python 2.7

PEP 493 provides additional migration tools to support a more incremental infrastructure upgrade process for environments containing applications and services relying on the historically permissive processing of server certificates when establishing client HTTPS connections. These additions were made in the Python 2.7.12 release.

These tools are intended for use in cases where affected applications and services can’t be modified to explicitly pass a more permissive SSL context when establishing the connection.

For applications and services which can’t be modified at all, the new PYTHONHTTPSVERIFY environment variable may be set to 0 to revert an entire Python process back to the default permissive behaviour of Python 2.7.8 and earlier.

For cases where the connection establishment code can’t be modified, but the overall application can be, the new ssl._https_verify_certificates() function can be used to adjust the default behaviour at runtime.

New make regen-all build target

To simplify cross-compilation, and to ensure that CPython can reliably be compiled without requiring an existing version of Python to already be available, the autotools-based build system no longer attempts to implicitly recompile generated files based on file modification times.

Instead, a new make regen-all command has been added to force regeneration of these files when desired (e.g. after an initial version of Python has already been built based on the pregenerated versions).

More selective regeneration targets are also defined - see Makefile.pre.in for details.

(Contributed by Victor Stinner in bpo-23404.)

Nuevo en la versión 2.7.14.

Removal of make touch build target

The make touch build target previously used to request implicit regeneration of generated files by updating their modification times has been removed.

It has been replaced by the new make regen-all target.

(Contributed by Victor Stinner in bpo-23404.)

Distinto en la versión 2.7.14.

Acknowledgements

The author would like to thank the following people for offering suggestions, corrections and assistance with various drafts of this article: Nick Coghlan, Philip Jenvey, Ryan Lovett, R. David Murray, Hugh Secker-Walker.