- file_util.py: operations on single files
- dir_util.py: operations on whole directories or directory trees
- dep_util.py: simple timestamp-based dependency analysis
- archive_util.py: creation of archive (tar, zip, ...) files
The functions left in util.py are miscellany that don't fit in any of the
new files.
the Command class from core.py to cmd.py. No other code needs changing
though; distutils.core still provides the Command and Distribution classes,
although indirectly now.
HKEY_* and Reg* names once, rather than having near-duplicate code
in the two import attempts.
Also dropped the leading underscore on all the imported symbols,
as it's not appropriate (they're not local to this module).
* build to "Debug" or "Release" temp directory
* put linker turds (.lib and .exp files) in the build temp directory
* tack on "_d" to extensions built with debugging
* added 'get_ext_libname()' help in putting linker turds to temp dir
Also, moved the code that simplifies None to empty list for a bunch
of options to 'finalize_options()' instead of 'run()'.
Simplified 'Command.get_peer_option()' a tad -- just call 'find_peer()'
to get the peer command object.
Updated 'Command.copy_file()' to take a 'link' parameter, just like
'util.copy_file()' does now.
Added 'Command.make_archive()' to wrap 'util.make_archive()'.
Changes to 'copy_file()':
* added support for making hard links and symlinks
* noted that it silently clobbers existing files when copying, but
blows up if destination exists when linking -- hmmm...
* error message tweak
Added 'base_name' parameter to 'make_tarball()' and 'make_zipfile()'.
Added 'make_archive()' -- wrapper around 'make_tarball()' or
'make_zipfile()' to take care of the archive "root directory".
the command that actually creates "dumb" binary distributions, ie.
tarballs and zip files that you just unpack under <prefix> or <exec-prefix>.
Very limited, but it's a start.
Initial revision is pretty limited; it only knows how to generate "dumb"
binary distributions, i.e. a tarball on Unix and a zip file on Windows.
Also, due to limitations in the installation code, it only knows how to
distribute Python library code. But hey, it's a start.
provided by Distribution.
Cosmetic and error message tweaks.
Simplified 'make_release_tree()':
* extracted 'distutils.util.create_tree()'
* don't have to do hard-linking ourselves -- it's now handled by
'distutils.util.copy_file()' (although the detection of
whether hard linking is available still needs to be factored out)
Removed 'make_tarball()' and 'make_zipfile()' entirely -- their role
is now amply filled by 'distutils.util.make_archive()'.
Simplified 'make_distribution()':
* use Distribution's new 'get_full_name()' method
* use 'make_archive()' instead of if/elif/.../else on the archive format
If a non-tuple sequence is passed as the *arg, convert it to a tuple
before checking its length.
If named keyword arguments are used in combination with **kwargs, make
a copy of kwargs before inserting the new keys.
is responsible for installing all Python modules (pure and extensions).
Added 'get_outputs()' in preparation for the 'bdist' command, and
'_mutate_outputs()' to support 'get_outputs()'.
in a class attribute 'sub_commands', rather than hard-coded in 'run()'.
This should make it easier to subclass 'install', and also makes it
easier to keep 'run()' and the new 'get_outputs()' consistent.
Added 'get_outputs()' in preparation for the 'bdist' command.
Changed signature of 'build_extensions()': no longer takes the extension
list, but uses 'self.extensions' (just like 'get_outputs()' has to)
Moved call to 'check_extensions_list()' from 'run()' to 'build_extensions()',
again for consistency with 'get_outputs()'.
A host of improvements in preparation for the 'bdist' command:
- added 'get_outputs()' method (all the other improvements were to support
this addition)
- made 'find_package_modules()' and 'find_modules()' return similar
values (list of (package, module, module_filename) tuples)
- factored 'find_all_modules()' out of 'get_source_files()' (needed
by 'get_outputs()')
- factored 'get_module_outfile()' out of 'build_module()' (also needed
by 'get_outputs()')
- various little tweaks, improvements, comment/doc updates
executive summary:
Instead of typing 'apply(f, args, kwargs)' you can type 'f(*arg, **kwargs)'.
Some file-by-file details follow.
Grammar/Grammar:
simplify varargslist, replacing '*' '*' with '**'
add * & ** options to arglist
Include/opcode.h & Lib/dis.py:
define three new opcodes
CALL_FUNCTION_VAR
CALL_FUNCTION_KW
CALL_FUNCTION_VAR_KW
Python/ceval.c:
extend TypeError "keyword parameter redefined" message to include
the name of the offending keyword
reindent CALL_FUNCTION using four spaces
add handling of sequences and dictionaries using extend calls
fix function import_from to use PyErr_Format
The new filecmp module has an optional argument called use_statcache
which is documented as a true/false value, but used as an tuple index.
This patches replaces the tuple stuff with a good old if- statement,
and also removes a few other tuple pack/unpack constructs (if not
else, this saves a few bytes in the PYC file, and a few microseconds
when using the module ;-).
The attached patch set includes a workaround to get Python with
Unicode compile on BSDI 4.x (courtesy Thomas Wouters; the cause
is a bug in the BSDI wchar.h header file) and Python interfaces
for the MBCS codec donated by Mark Hammond.
Also included are some minor corrections w/r to the docs of
the new "es" and "es#" parser markers (use PyMem_Free() instead
of free(); thanks to Mark Hammond for finding these).
The unicodedata tests are now in a separate file
(test_unicodedata.py) to avoid problems if the module cannot
be found.
This patch fixes the "search" command in imaplib. The problem
was that a search can take multiple arguments, but as defined,
would only accept one.
I have also made changes to the test code at the end to be less
verbose by default, but to accept a verbosity argument.
The robotparser.py module currently lives in Tools/webchecker. In
preparation for its migration to Lib, I made the following changes:
* renamed the test() function _test
* corrected the URLs in _test() so they refer to actual documents
* added an "if __name__ == '__main__'" catcher to invoke _test()
when run as a main program
* added doc strings for the two main methods, parse and can_fetch
* replaced usage of regsub and regex with corresponding re code
we don't know what to do with it when we see it.
Call '_fix_object_args()' and/or '_fix_lib_args()' as appropriate, rather
than just '_fix_link_args()'.
Split '_fix_link_args()' up into '_fix_object_args()' (for use of
'create_static_lib() and link methods) and '_fix_lib_args()' (for the
link methods only).
Attached you find the latest update of the Unicode implementation.
The patch is against the current CVS version.
It includes the fix I posted yesterday for the core dump problem
in codecs.c (was introduced by my previous patch set -- sorry),
adds more tests for the codecs and two new parser markers
"es" and "es#".
these must come from the 'build' command. This means we no longer need
the misconceived 'set_peer_option()' method in Command and, more importantly,
sweeps away a bunch of potential future complexity to handle this tricky
case.
Fix 'Command.set_undefined_option()' to call 'ensure_ready()' rather
than 'finalize_options()' (which is only supposed to be called once,
which is the whole point of 'ensure_ready()').
Added comment to 'set_peer_option()' to remind myself that this method
cannot work and is fundamentally wrong-headed.
this command for a while; this implements roughly the plan cooked up by
Guido, Fred, and me. Seems to strike a nice balance between usability in
the common cases (just set one option), expandability for more types of
files to install in future, and customizability of installation
directories.
This revision isn't completely working: standard and alternate
installations work fine, but there are still some kinks to work out of
customized installations.
Attached you find an update of the Unicode implementation.
The patch is against the current CVS version. I would appreciate
if someone with CVS checkin permissions could check the changes
in.
The patch contains all bugs and patches sent this week and also
fixes a leak in the codecs code and a bug in the free list code
for Unicode objects (which only shows up when compiling Python
with Py_DEBUG; thanks to MarkH for spotting this one).
* improve help strings
* warn if user supplies non-existing directories
* don't try to 'remove_tree()' non-existing directories
* try to remove the build_base after cleanup (but don't do or say
anything if it fails -- this is just in case we made it empty)
option in the 'build_ext' command):
* in ccompiler.py: 'gen_lib_options()' now takes 'runtime_library_dirs'
parameter
* in unixccompiler.py and msvccompiler.py: now pass
'self.runtime_library_dirs' to 'gen_lib_options()', and define
'runtime_library_dir_option()' (although in msvccompiler.py it
blows up with a DistutilsPlatformError right now!)
code generator uses flowgraph as intermediate representation. the old
rep uses a list with explicit "StackRefs" to indicate the target
of jumps.
pyassem converts flowgraph to bytecode, breaks up individual steps of
generating bytecode
sure it's imported! ;)
Re-wrap the docstrings on get_python_inc() and get_python_lib() to be
closer to the "normal" Python style. See GvR's "style guide" on the
essays page (http://www.python.org/doc/essays/).
There should never be a space between a function name and the '(' that
opens the argument list (see the style guide again).
the platform-neutral include dir by default and with Mac support.
Added 'get_python_lib()', inspired by 'get_python_inc()'.
Rewrote 'get_config_h_filename()' and 'get_makefile_filename()'
in terms of 'get_python_inc()' and 'get_python_lib()'.
Changed '_init_nt()' and '_init_mac()' to use 'get_python_inc()' and
'get_python_lib()' for directory names.
When you set a breakpoint on a function with a multi-line argument
list, the breakpoint is actually set on the second line of the
arguments instead of the first line of the body. This patch fixes
that.
fix imports
remove parse functions and visitor code
track name change: Classdef to Class
add some comments and tweak order of visitXXX methods
get rid of if __name__ == "__main__ section
implement it (so far):
* moved filename generation methods into CCompiler base class,
driven by data supplied by implementation classes
* moved a bunch of common code from UnixCCompiler to convenience
methods in CCompiler
* overhauled MSVCCompiler's compile/link methods to look and act
as much as possible like UnixCCompiler's, in order to regularize
both interface and behaviour (especially by using those new
convenience methods)
* replaced build_lib.py with build_clib.py
* renamed the class in build_clib.py
* changed all references to 'build_lib' command in other command classes
when building extensions (uses build_lib's 'get_library_names()' method).
Ensure that the relative structure of source filenames is preserved in
the temporary build tree, eg. foo/bar.c compiles to
build/temp.<plat>/foo/bar.o.
Added 'build_clib' and 'build_temp' options (where to put C libraries
and where to put temporary compiler by-products, ie. object files).
Moved the call to 'check_library_list()' from 'run()' to 'finalize_options()'
-- that way, if we're going to crash we do so earlier, and we guarantee
that the library list is valid before we do anything (not just run).
Disallow directory separators in library names -- the compiled library
always goes right in 'build_clib'.
Added 'get_library_names()', so the "build_ext" command knows what
libraries to link every extension with.
filenames when constructing object filenames, even if output_dir given --
eg. "foo/bar.c" will compile to "foo/bar.o" without an output_dir, and to
"temp/foo/bar.o" if output_dir is "temp".
* 'build_dir' -> 'build_lib', which by default takes its value
straight from 'build_lib' in the 'build' command
* added 'build_temp' and 'inplace' options
* change 'build_extensions()' to put object files (compiler turds) in
'build_temp' dir
* complicated the name-of-extension-file shenanigans in
'build_extensions()' to support "in-place" extension building, i.e.
put the extension right into the source tree (handy for developers)
* added 'get_ext_fullname()', renamed 'extension_filename()' to
'get_ext_filename()', and tweaked the latter a bit -- all to support
the new filename shenanigans
This patch is re: Lucas.Dejonge@awtpl.com.au: [Python-bugs-list] imaplib -
not complying with RFC (PR#218)
Lucas de Jonge reported that the code in imaplib that detects a read-write
to read-only change doesn't comply with RFC 2060.
discussed on c.l.py last January. Specifically:
- more characters allowed in section & option names
- if '=' is used to separate the option & value, the value can be
followed by a comment of the form '\s;'
OptionMenu is modified. Somewhat rewritten and elaborated by myself.
class _setit: The constructor now takes an optional argument
`callback' and stashes this in a private variable. If set, the
__call__() method will invoke this callback after the variable's value
has changed. It will pass the callback the value, followed by any
args passed to __call__().
class OptionMenu: The constructor now takes keyword arguments, the
only one that's legally recognized is `command', which can be set to a
callback. This callback is invoked when the OptionMenu value is set.
Any other keyword argument throws a TclError.
A change in my last patch could, under certain circumstances,
cause a loop if the connection to the server dropped while
waiting for a command completion. I've changed the code to
re-raise the error after possible debugging output.
standard library. Added some comments:
# XXX Note: this is now a standard library module.
# XXX The API needs to undergo changes however; the current code is too
# XXX script-like. This will be addressed later.
add clsFilesystemImporter class attribute, alter handling of suffix list
convert suffix importers to funcs rather than instances
remove backwards compat code: Importer.install and 2-tuple get_code()
result values
switch to isinstance() rather than direct type comparisons
removing chaining concept
update ImportManager.install() to take an optional namespace to install
itself in. this will be useful for setting up rexec environments.
minor comment nits
old 'dist' command, but the code for dealing with manifests is completely
redone -- and renaming the command to 'sdist' is more symmetric with the
soon-to-exist 'bdist' command.
- removed now (happily) unused second arg
- need to verify results of [].index are correct; for building consts,
need to have same value and same type, e.g. 2 not the same as 2L
(big surprise). new solution is a little less hackish.
Code gen adds a TupleArg instance in the argument slot. The tuple arg
includes a copy of the names that it is responsble for binding. The
PyAssembler uses this information to calculate the correct argcount.
all fix this wacky case: del (a, ((b,), c)), d
which is the same as: del a, b, c, d
(Can't wait for Guido to tell me why.)
solution uses findOp which walks a tree to find out whether it
contains OP_ASSIGN or OP_DELETE or ...
Reasons for patches:
1st patch (15,21):
version change
2nd patch (66,72):
This is a patch I found in a Zope product release (quite by accident!).
It relaxes the conditions for matching a literal. I've looked over the
logic, and tested it, and it seems sensible.
3rd patch (117,123):
It appears the quoting matcher was too general, and that the IMAP4
protocol requires characters like ':' in commands to be unquoted.
(This is the patch already sent to Guido.)
4th patch (699,705):
Spelling correction in comment.
5th patch (753,761):
Another patch from the Zope product. It seems that some IMAP4 servers
produce unexpected responses in the middle of valid command/response
sequences. This patch ignores the unexpected responses in this
situation. (How I wish users would send me bug reports with examples!).
last 2 patches: (1015,1028) (1038,1044):
Minor improvements to test code.
- added a number of support methods to generate code just before the
body
- hack protocol for communicating number of args to PyAssembler
fix TryExcept generation for case where exception handler has no body
fix visitAssAttr
add comment about incomplete visitAssName
stop using the ExampleASTVisitor
change script invocation to accept a list of .py files (e.g. Lib/*.py)
named OPTIMIZED, which matches compile.c and makes more sense for
classes
revamp call signature for PythonVMCode.__init__; doesn't really matter
'cuz this code is going to be refactored out of existence
add generateClassCode and modify Func & Lambda generation
add support for nodes Classdef, Keyword,
fix CallFunc to generate right op arg when calling w/ keywords
add ugly hack to properly compute offsets when the same stack ref is
used multiple times
change resolution of local name ops (LOAD_FAST). i think it makes
sense now. if it is an argument or a local var name that it used, it
must be in varnames. if it is a local var name that is used, it must
also be in names
'get_msvc_paths()'.
Renamed '_do_SET()' to 'set_path_env_var()', tweaked docstring, and
cosmetically tweaked code.
Stylistic changes to MSVCCompiler constructor (variable renaming
and type consistency).
* tweak my docstrings
* fix None returns to empty list
* reshuffle responsibilities between '_find_exe()', '_find_SET()', and
the MSVCCompiler constructor -- now the constructor worries about
fetching the version list and determining the most recent one
* added "/W3" compile option
Also, I added/tweaked some docstrings.
NNTPError - derived from Exception, it's the base class for all
other exceptions in this module
NNTPReplyError - what used to be error_reply
NNTPTemporaryError - what used to be error_temp
NNTPPermanentError - what used to be error_perm
NNTPProtocolError - what used to be error_proto
NNTPDataError - what used to be error_data
All the old names are retained for backwards compatibility; they point
to the class that replaces them. Also, any code in this module that
raises an exception, now does so with the exception class.
NNTP.__init__(): Added a new optional argument `readermode', which is
a flag that defaults to false. When set to true, the "mode reader"
command is sent to the NNTP server before user authentication. Reader
mode is sometimes necessary if you are connecting to an NNTP server on
the local machine and intend to call reader-specific comamnds, such as
`group'. If you get unexpected NNTPPermanentErrors, you might need to
set readermode. Patch provided by Thomas Wouters (who include the
standard disclaimer on is patches@python.org submission), and inspired
by Jim Tittsler.
FUNCTION_NAMESPACE. initialize in __init__ and reset in
generateFunctionCode.
replace direct issue of STORE_FAST, STORE_GLOBAL, etc. with call to
storeName; same for loadName and deleteName
the new {store,load,delete}Name methods use the namespace attr and the
local variable stack to determine the correct bytecode to issue
* don't need to mention python<ver>.lib -- it's done by a pragma
* add debug flags for compile and link, and use them
* fix 'link_shared_library()' to pass everything to 'link_shared_object()'
* change filename when shared object with debug info (ugh)
* prints out examples of nodes that are handled by visitor. simply a
development convenience
remove NestedCodeGenerator -- it was bogus after all
replace with generateFunctionCode, a method to call to generate code
for a function instead of a top-level module
fix impl of visitDiscard (most pop stack)
emit lineno for pass
handle the following new node types: Import, From, Getattr, Subscript,
Slice, AssAttr, AssTuple, Mod, Not, And, Or, List
LocalNameFinder: remove names declared as globals for locals
PythonVMCode: pass arg names to constructor, force varnames to contain
them all (even if they aren't referenced)
add -q option on command line to disable stdout
someone who knows Windows/MSVC++ to come along and add the right flags.
Comment noting that 'link_static_lib()' signature is inconsistent with
the other compiler classes (uh-oh!)
VERBOSE setting for the ASTVisitor
add getopt handling for one or more -v args
rename ForwardRef to StackRef, because it isn't necessarily directional
CodeGenerator:
* add assertStackEmpty method. prints warning if stack is not empty
when it should be
* define methods for AssName, UNARY_*, For
PythonVMCode:
* fix mix up between hasjrel and hasjabs for address calculation
language.
CodeGenerator:
* modify to track stack depth
* add emit method that call's PythonVMCode's makeCodeObject
* thread filenames through in hackish way
* set flags for code objects for modules and functions
XXX the docs for the flags seem out of date and/or incomplete
PythonVMCode:
* add doc string describing the elements of a real code object
LineAddrTable:
* creates an lnotab (no quite correctly though)
Greg Ward
Fred Drake
Andrew M. Kuchling
Guido van Rossum
Jeremy Hylton
Barry Warsaw
Greg Stein