mirror of https://github.com/python/cpython
Some more new stuff.
This commit is contained in:
parent
68fede768a
commit
974f295dd4
56
Misc/NEWS
56
Misc/NEWS
|
@ -31,6 +31,13 @@ in the same archive file, and perhaps not on the same date.
|
|||
- All bugs noted on the errors page (and many unnoted) are fixed. All
|
||||
new bugs take their places.
|
||||
|
||||
- No longer a core dump when attempting to print (or repr(), or str())
|
||||
a list or dictionary that contains an instance of itself; instead, the
|
||||
recursive entry is printed as [...] or {...}. See Py_ReprEnter() and
|
||||
Py_ReprLeave() below. Comparisons of such objects still go beserk,
|
||||
since this requires a different kind of fix; fortunately, this is a
|
||||
less common scenario in practice.
|
||||
|
||||
Syntax change
|
||||
-------------
|
||||
|
||||
|
@ -52,13 +59,9 @@ careful with the order in which modules are destroyed. Destructors
|
|||
pwill now generally be able to reference built-in names such as None
|
||||
without trouble.
|
||||
|
||||
- On case-insensitive platforms such as Mac and Windows, require the
|
||||
case of a module's filename, e.g. "SocketServer.py", to match the case
|
||||
of the module name as specified in the import statement. This is an
|
||||
experimental feature -- if it turns out to break in too many
|
||||
situations, it will be removed (or disabled by default) in the future.
|
||||
On Windows, it can be disabled on a per-case basis by setting the
|
||||
environment variable PYTHONCASEOK (to any value).
|
||||
- Case-insensitive platforms such as Mac and Windows require the case
|
||||
of a module's filename to match the case of the module name as
|
||||
specified in the import statement (see below).
|
||||
|
||||
- The code for figuring out the default path now distinguishes between
|
||||
files, modules, executable files, and directories. When expecting a
|
||||
|
@ -68,9 +71,10 @@ Parser/tokenizer changes
|
|||
------------------------
|
||||
|
||||
- The tokenizer can now warn you when your source code mixes tabs and
|
||||
spaces for indentation in a manner that depends on the worth of a tab
|
||||
in spaces. Use "python -t" or "python -v" to enable this option. Use
|
||||
"python -tt" to turn the warnings into errors.
|
||||
spaces for indentation in a manner that depends on how much a tab is
|
||||
worth in spaces. Use "python -t" or "python -v" to enable this
|
||||
option. Use "python -tt" to turn the warnings into errors. (See also
|
||||
tabnanny.py and tabpolice.py below.)
|
||||
|
||||
- Return unsigned characters from tok_nextc(), so '\377' isn't
|
||||
mistaken for an EOF character.
|
||||
|
@ -161,7 +165,9 @@ Standard library modules
|
|||
|
||||
- All standard library modules have been converted to an indentation
|
||||
style using either only tabs or only spaces -- never a mixture -- if
|
||||
they weren't already consistent according to tabnanny.
|
||||
they weren't already consistent according to tabnanny. This means
|
||||
that the new -t option (see above) won't complain about standard
|
||||
library modules.
|
||||
|
||||
- New standard library modules:
|
||||
|
||||
|
@ -199,8 +205,8 @@ the re module and the supporting pcre extension) by Andrew Kuchling.
|
|||
Incompatible new feature in re.sub(): the handling of escapes in the
|
||||
replacement string has changed.
|
||||
|
||||
- Interface change in copy.py: a __deepcopy__ method is now called
|
||||
with the memo dictionary as an argument.
|
||||
- Interface change in the copy module: a __deepcopy__ method is now
|
||||
called with the memo dictionary as an argument.
|
||||
|
||||
- Feature change in the tokenize module: differentiate between NEWLINE
|
||||
token (an official newline) and NL token (a newline that the grammar
|
||||
|
@ -323,6 +329,13 @@ so Tcl will be finalized when Python exits.
|
|||
The Python/C API
|
||||
----------------
|
||||
|
||||
- New function PyThreadState_GetDict() returns a per-thread dictionary
|
||||
intended for storing thread-local global variables.
|
||||
|
||||
- New functions Py_ReprEnter() and Py_ReprLeave() use the per-thread
|
||||
dictionary to allow recursive container types to detect recursion in
|
||||
their repr(), str() and print implementations.
|
||||
|
||||
- New function PyObject_Not(x) calculates (not x) according to Python's
|
||||
standard rules (basically, it negates the outcome PyObject_IsTrue(x).
|
||||
|
||||
|
@ -369,6 +382,23 @@ of the file.
|
|||
|
||||
- Make sure setvbuf() isn't used unless HAVE_SETVBUF is defined.
|
||||
|
||||
Windows 95/NT
|
||||
-------------
|
||||
|
||||
- The .lib files are now part of the distribution; they are collected
|
||||
in the subdirectory "libs" of the installation directory.
|
||||
|
||||
- The extension modules (.pyd files) are now collected in a separate
|
||||
subdirectory of the installation directory named "DLLs".
|
||||
|
||||
- The case of a module's filename must now match the case of the
|
||||
module name as specified in the import statement. This is an
|
||||
experimental feature -- if it turns out to break in too many
|
||||
situations, it will be removed (or disabled by default) in the future.
|
||||
It can be disabled on a per-case basis by setting the environment
|
||||
variable PYTHONCASEOK (to any value).
|
||||
|
||||
|
||||
|
||||
======================================================================
|
||||
|
||||
|
|
Loading…
Reference in New Issue