]> granicus.if.org Git - python/log
python
24 years agoUpdates to reflect pending changes to the XML conversion process.
Fred Drake [Wed, 22 Nov 2000 16:42:37 +0000 (16:42 +0000)]
Updates to reflect pending changes to the XML conversion process.

24 years agoDo not use \verb in the Python documentation -- it makes parsing the LaTeX
Fred Drake [Wed, 22 Nov 2000 16:06:16 +0000 (16:06 +0000)]
Do not use \verb in the Python documentation -- it makes parsing the LaTeX
sources more difficult and other tools do not always work well with it.
Since we have better markup for this case, just fix it.

24 years agoUpdate test output.
Fred Drake [Tue, 21 Nov 2000 22:03:09 +0000 (22:03 +0000)]
Update test output.

24 years agotestInsertBefore(): Rewritten to actually test insertBefore() somewhat.
Fred Drake [Tue, 21 Nov 2000 22:02:43 +0000 (22:02 +0000)]
testInsertBefore():  Rewritten to actually test insertBefore() somewhat.

testAAA(),
testAAB():  Added checks that the results are right.

testTooManyDocumentElements():  Added code to actually test this.

testCloneElementDeep()
testCloneElementShallow():  Filled these in with test code.

_testCloneElementCopiesAttributes(),
_setupCloneElement():  Helper functions used with the other
        testCloneElement*() functions.

testCloneElementShallowCopiesAttributes():  No longer a separate test;
        _setupCloneElement() uses _testCloneElementCopiesAttributes() to
        test that this is always done.

testNormalize():  Added to check Node.normalize().

24 years agoReduce the visibility of imported modules for cleaner "from ... import *"
Fred Drake [Tue, 21 Nov 2000 22:02:22 +0000 (22:02 +0000)]
Reduce the visibility of imported modules for cleaner  "from ... import *"
behavior.

Added support for the Attr.ownerElement attribute.

Everywhere:  Define constant object attributes in the classes rather than
on the instances during object construction.  This reduces the amount of
work needed for object construction and destruction; these need to be
lightweight operations on a DOM.

Node._get_firstChild(),
Node._get_lastChild():  Return None if there are no children (required for
        compliance with DOM level 1).

Node.insertBefore():  If refChild is None, append the new node instead of
        failing (required for compliance).  Also, update the sibling
        relationships.  Return the inserted node (required for compliance).

Node.appendChild():  Update the parent of the appended node.

Node.replaceChild():  Actually replace the old child!  Update the parent
        and sibling relationships of both the old and new children.  Return
        the replaced child (required for compliance).

Node.normalize():  Implemented the normalize() method.  Required for
        compliance, but missing from the release.  Useful for joining
        adjacent Text nodes into a single node for easier processing.

Node.cloneNode():  Actually make this work.  Don't let the new node share
        the instance __dict__ with the original.  Do proper recursion if
        doing a "deep" clone.  Move the attribute cloning out of the base
        class, since only Element is supposed to have attributes.

Node.unlink():  Simplify handling of child nodes for efficiency, and
        remove the attribute handling since only Element nodes support
        attributes.

Attr.cloneNode():  Extend this to clear the ownerElement attribute in
        the clone.

AttributeList.items(),
AttributeList.itemsNS():  Slight performance improvement (avoid lambda).

Element.cloneNode():  Extend Node.cloneNode() with support for the
        attributes.  Clone the Attr objects after creating the underlying
        clone.

Element.unlink():  Clean out the attributes here instead of in the base
        class, since this is the only class that will have them.

Element.toxml():  Adjust to create only one AttributeList instance; minor
        efficiency improvement.

_nssplit():  No need to re-import string.

Document.__init__():  No longer needed once constant attributes are
        initialized in the class itself.

Document.createElementNS(),
Document.createAttributeNS():  Use the defined constructors rather than
        directly access the classes.

_get_StringIO():  New function.  Create an output StringIO using the most
        efficient available flavor.

parse(),
parseString():  Import pulldom here instead of in the public namespace of
        the module.

24 years agoAdded MathLib and InterfaceLib. Added Python:Include so #include "package/file.h...
Jack Jansen [Sun, 19 Nov 2000 21:53:15 +0000 (21:53 +0000)]
Added MathLib and InterfaceLib. Added Python:Include so #include "package/file.h" works. Moved bits around so now at least Numeric with Distutils builds and instals.

24 years agoClose the project after building it, so that building umpteen projects doesn't keep...
Jack Jansen [Sun, 19 Nov 2000 21:51:06 +0000 (21:51 +0000)]
Close the project after building it, so that building umpteen projects doesn't keep all the projects open.

24 years agoDocument the new setupterm() function
Andrew M. Kuchling [Sat, 18 Nov 2000 17:57:33 +0000 (17:57 +0000)]
Document the new setupterm() function

24 years agoPatch #102412 from mwh: Add support for the setupterm() function, to
Andrew M. Kuchling [Sat, 18 Nov 2000 17:45:59 +0000 (17:45 +0000)]
Patch #102412 from mwh: Add support for the setupterm() function, to
    initialize the terminal without necessarily calling initscr()

24 years agoAdded information about the %r string formatting conversion. Added note
Fred Drake [Fri, 17 Nov 2000 19:44:14 +0000 (19:44 +0000)]
Added information about the %r string formatting conversion.  Added note
about the interpretation of radix 0 for int(), and added description of
the optional radix argument for long().  Based on comments from Reuben
Sumner <rasumner@users.sourceforge.net>.

This closes bug #121672.

24 years agoOops, back out change committed by accident! This is not ready, and
Fred Drake [Fri, 17 Nov 2000 19:09:34 +0000 (19:09 +0000)]
Oops, back out change committed by accident!  This is not ready, and
breaks things.

24 years agoNote that readframes() returns data in linear format, even if the original
Fred Drake [Fri, 17 Nov 2000 19:05:12 +0000 (19:05 +0000)]
Note that readframes() returns data in linear format, even if the original
is encoded in u-LAW format.  Based on suggestion from Anthony Baxter
<anthony_baxter@users.sourceforge.net>.

This closes bug #122273.

24 years agoCorrected a number of typos reported by Gilles Civario
Fred Drake [Fri, 17 Nov 2000 18:20:33 +0000 (18:20 +0000)]
Corrected a number of typos reported by Gilles Civario
<gcivario@users.sourceforge.net>.

This closes bug #122562.

24 years agoFixed typos and bug in the second example, reported by Scott Schram
Fred Drake [Fri, 17 Nov 2000 18:04:03 +0000 (18:04 +0000)]
Fixed typos and bug in the second example, reported by Scott Schram
<schram@users.sourceforge.net>.

This closes bug #122236.

24 years agoA solution to the classic N queens problem.
Guido van Rossum [Thu, 16 Nov 2000 21:25:51 +0000 (21:25 +0000)]
A solution to the classic N queens problem.

24 years agoFix for SF bug 122176: Error in rotor module documentation.
Tim Peters [Tue, 14 Nov 2000 21:43:01 +0000 (21:43 +0000)]
Fix for SF bug 122176: Error in rotor module documentation.

24 years agoGetting rid of 68K targets.
Jack Jansen [Tue, 14 Nov 2000 21:42:11 +0000 (21:42 +0000)]
Getting rid of 68K targets.
Started to rename projects to the de-facto standard .mcp extension.

24 years agoVerify that str(a) and repr(a) don't blow up (part of SF patch 102068).
Tim Peters [Tue, 14 Nov 2000 21:36:07 +0000 (21:36 +0000)]
Verify that str(a) and repr(a) don't blow up (part of SF patch 102068).

24 years agoSF bug 119622: compile errors due to redundant atof decls. I don't understand
Tim Peters [Tue, 14 Nov 2000 20:44:53 +0000 (20:44 +0000)]
SF bug 119622:  compile errors due to redundant atof decls.  I don't understand
the bug report (for details, look at it), but agree there's no need for Python
to declare atof itself:  we #include stdlib.h, and ANSI C sez atof is declared
there already.

24 years agoAdded a note about removing preferences and .pyc files when you move the Python insta...
Jack Jansen [Tue, 14 Nov 2000 20:37:37 +0000 (20:37 +0000)]
Added a note about removing preferences and .pyc files when you move the Python installation.

24 years agoTypo for Mac code, fixing SF bug 12195.
Guido van Rossum [Mon, 13 Nov 2000 20:30:57 +0000 (20:30 +0000)]
Typo for Mac code, fixing SF bug 12195.

24 years agoAllow new.function() called with explicit 3rd arg of None, as
Guido van Rossum [Mon, 13 Nov 2000 20:29:20 +0000 (20:29 +0000)]
Allow new.function() called with explicit 3rd arg of None, as
documented, and as is reasonable (since it is optional, but there's
another argument following it that may require you to specify a
value).  This solves SF bug 121887.

24 years agoAdded _HAVE_BSDI and __APPLE__ to the list of platforms that require a
Guido van Rossum [Mon, 13 Nov 2000 19:48:22 +0000 (19:48 +0000)]
Added _HAVE_BSDI and __APPLE__ to the list of platforms that require a
hack for TELL64()...  Sounds like there's something else going on
really.  Does anybody have a clue I can buy?

24 years agoFix syntax error. Submitted by Bill Bumgarner. Apparently this is
Guido van Rossum [Mon, 13 Nov 2000 19:45:45 +0000 (19:45 +0000)]
Fix syntax error.  Submitted by Bill Bumgarner.  Apparently this is
still in use, for Apple Mac OSX.

24 years agoRip out DOS and Win16 support.
Guido van Rossum [Mon, 13 Nov 2000 17:29:30 +0000 (17:29 +0000)]
Rip out DOS and Win16 support.

24 years agoRip out DOS-8x3 support.
Guido van Rossum [Mon, 13 Nov 2000 17:26:32 +0000 (17:26 +0000)]
Rip out DOS-8x3 support.

24 years agoRip out Win3.1 and DOS support
Guido van Rossum [Mon, 13 Nov 2000 17:24:13 +0000 (17:24 +0000)]
Rip out Win3.1 and DOS support

24 years agoRemoving DOS 8x3 support
Guido van Rossum [Mon, 13 Nov 2000 17:11:45 +0000 (17:11 +0000)]
Removing DOS 8x3 support

24 years agoJack Jansen: added 'get_command_list()' method, and Mac-specific code to
Greg Ward [Sat, 11 Nov 2000 02:47:11 +0000 (02:47 +0000)]
Jack Jansen: added 'get_command_list()' method, and Mac-specific code to
use it to generate a dialog for users to specify the command-line (because
providing a command-line with MacPython is so awkward).

24 years agoNo reason to keep "Run as Python" in binhex form anymore, just put the binary in...
Jack Jansen [Fri, 10 Nov 2000 22:38:31 +0000 (22:38 +0000)]
No reason to keep "Run as Python" in binhex form anymore, just put the binary in the repository.

24 years agoFix for SF bug 117402, crashes on str(array) and repr(array). This was an
Tim Peters [Fri, 10 Nov 2000 19:04:19 +0000 (19:04 +0000)]
Fix for SF bug 117402, crashes on str(array) and repr(array).  This was an
unfortunate consequence of somebody switching from PyArg_Parse to
PyArg_ParseTuple but without changing the argument from a NULL to a tuple.

24 years agoCommitting autoconf output for
Guido van Rossum [Thu, 9 Nov 2000 21:14:40 +0000 (21:14 +0000)]
Committing autoconf output for
Fred.

24 years agoImplement the suggestion of bug_id=122070: surround tell() call with
Guido van Rossum [Thu, 9 Nov 2000 18:05:24 +0000 (18:05 +0000)]
Implement the suggestion of bug_id=122070: surround tell() call with
try/except.

24 years agoSMTP.connect(): If the socket.connect() raises a socket.error, be sure
Barry Warsaw [Wed, 8 Nov 2000 22:19:47 +0000 (22:19 +0000)]
SMTP.connect(): If the socket.connect() raises a socket.error, be sure
to call self.close() to reclaim some file descriptors, the reraise the
exception.  Closes SF patch #102185 and SF bug #119833.

24 years agoRemove AC_C_INLINE test from configure.in, since the only place the symbol
Fred Drake [Wed, 8 Nov 2000 20:22:59 +0000 (20:22 +0000)]
Remove AC_C_INLINE test from configure.in, since the only place the symbol
occurs in the Python sources appears to be as text in comments.  We do not
want to interfere with C++ keywords!

This closes bug #119851.

24 years agoAdded test cases to detect regression on SourceForge bug #121965.
Fred Drake [Wed, 8 Nov 2000 19:51:25 +0000 (19:51 +0000)]
Added test cases to detect regression on SourceForge bug #121965.

24 years agoFixed support for containment test when a negative step is used; this
Fred Drake [Wed, 8 Nov 2000 19:42:43 +0000 (19:42 +0000)]
Fixed support for containment test when a negative step is used; this
*really* closes bug #121965.

Added three attributes to the xrange object: start, stop, and step.  These
are the same as for the slice objects.

24 years agoIn the containment test, get the boundary condition right. ">" was used
Fred Drake [Wed, 8 Nov 2000 18:37:05 +0000 (18:37 +0000)]
In the containment test, get the boundary condition right.  ">" was used
where ">=" should have been.

This closes bug #121965.

24 years agoAdd 1994 Coroutine module by Tim Peters
Guido van Rossum [Wed, 8 Nov 2000 15:17:49 +0000 (15:17 +0000)]
Add 1994 Coroutine module by Tim Peters

24 years agoLast CW Pro 5.3 projects.
Jack Jansen [Tue, 7 Nov 2000 21:07:33 +0000 (21:07 +0000)]
Last CW Pro 5.3 projects.

24 years agoMore names...
Fred Drake [Tue, 7 Nov 2000 16:09:53 +0000 (16:09 +0000)]
More names...

24 years agoFix for SF bug #117606:
Greg Ward [Tue, 7 Nov 2000 15:44:21 +0000 (15:44 +0000)]
Fix for SF bug #117606:
  - when compiling with GCC on Solaris, use "$(CC) -shared" instead
    of "$(CC) -G" to generate .so files
  - when compiling with GCC on any platform, add "-fPIC" to OPT
    (without this, "$(CC) -shared" dies horribly)

24 years agoPatch #102278: add tparm() function to _curses module
Andrew M. Kuchling [Tue, 7 Nov 2000 03:35:24 +0000 (03:35 +0000)]
Patch #102278: add tparm() function to _curses module

24 years agoDocument the proper exception to be raised by I/O operations on closed
Fred Drake [Mon, 6 Nov 2000 20:17:37 +0000 (20:17 +0000)]
Document the proper exception to be raised by I/O operations on closed
files; error reported by Ng Pheng Siong <ngps@post1.com>.

Make sure that various special object attributes are properly indexed.

24 years agoThis fixes several bug reports concering memory bloating during large
Barry Warsaw [Mon, 6 Nov 2000 18:46:09 +0000 (18:46 +0000)]
This fixes several bug reports concering memory bloating during large
file uploads.

In response to SF bugs 110674 and 119806, and discussions on
python-dev, we are removing the self.lines attribute from the
FieldStorage class.  Specifically touched where methods __init__(),
read_lines_to_eof(), and skip_lines().

No one can remember why self.lines was added.  Technically, it's part
of the public interface for the class, but it was never documented.
It's possible clever or nosy code will break because of this, but it
was decided to remove it and see who complains.

This resolution also closes the second half of the cgi.py entry in PEP
42.  The first half of that PEP concerns specifically binary file
uploads, where there may be no end-of-line marker for a very long
time.  This patch does not address that issue.

24 years agoa few small optimizations that seem to give a 5-10% speedup; the
Jeremy Hylton [Mon, 6 Nov 2000 16:03:52 +0000 (16:03 +0000)]
a few small optimizations that seem to give a 5-10% speedup; the
further optimization of com_node makes the most difference.

24 years agomove pruneNext method to correct object (doh!)
Jeremy Hylton [Mon, 6 Nov 2000 03:47:39 +0000 (03:47 +0000)]
move pruneNext method to correct object (doh!)

24 years agoChange the graph structure to contain the code generator object for
Jeremy Hylton [Mon, 6 Nov 2000 03:43:11 +0000 (03:43 +0000)]
Change the graph structure to contain the code generator object for
embedded code objects (e.g. functions) rather than the generated code
object.  This change means that the compiler generates code for
everything at the end, rather then generating code for each function
as it finds it.  Implementation note: _convert_LOAD_CONST in
pyassem.py must be change to call getCode().

Other changes follow.  Several changes creates extra edges between
basic blocks to reflect control flow for loops and exceptions.  These
missing edges had gone unnoticed because they do not affect the
current compilation process.

pyassem.py:
    Add _enable_debug() and _disable_debug() methods that print
    instructions and blocks to stdout as they are generated.

    Add edges between blocks for instructions like SETUP_LOOP,
    FOR_LOOP, etc.

    Add pruneNext to get rid of bogus edges remaining after
    unconditional transfer ops (e.g. JUMP_FORWARD)

    Change repr of Block to omit block length.

pycodegen.py:
    Make sure a new block is started after FOR_LOOP, etc.

    Change assert implementation to use RAISE_VARARGS 1 when there is
    no user-specified failure output.

misc.py:
    Implement __contains__ and copy for Set.

24 years agoIf a function contains a doc string, remove the doc string node from
Jeremy Hylton [Mon, 6 Nov 2000 03:33:52 +0000 (03:33 +0000)]
If a function contains a doc string, remove the doc string node from
the function's body.

If assert is used without an error message, make the AST node None
rather than Name('None').

24 years agoApplication of [ Patch #102226 ] freeze/modulefinder.py should use _winreg, not win32api
Mark Hammond [Mon, 6 Nov 2000 02:49:27 +0000 (02:49 +0000)]
Application of [ Patch #102226 ] freeze/modulefinder.py should use _winreg, not win32api

24 years agoAdded 38,642 missing characters to the Unicode database (first-last
Fredrik Lundh [Fri, 3 Nov 2000 20:24:15 +0000 (20:24 +0000)]
Added 38,642 missing characters to the Unicode database (first-last
ranges) -- but thanks to the 2.0 compression scheme, this doesn't add
a single byte to the resulting binaries (!)

Closes bug #117524

24 years agoFix Makefile so at least it uses Python 2.0, and compiles out of the
Guido van Rossum [Fri, 3 Nov 2000 12:58:09 +0000 (12:58 +0000)]
Fix Makefile so at least it uses Python 2.0, and compiles out of the
box on Linux.

24 years agoMove our own getopt() implementation to _PyOS_GetOpt(), and use it
Thomas Wouters [Fri, 3 Nov 2000 08:18:37 +0000 (08:18 +0000)]
Move our own getopt() implementation to _PyOS_GetOpt(), and use it
regardless of whether the system getopt() does what we want. This avoids the
hassle with prototypes and externs, and the check to see if the system
getopt() does what we want. Prefix optind, optarg and opterr with _PyOS_ to
avoid name clashes. Add new include file to define the right symbols. Fix
Demo/pyserv/pyserv.c to include getopt.h itself, instead of relying on
Python to provide it.

24 years agoMake sure we clean up the index data each time it is written by LaTeX.
Fred Drake [Fri, 3 Nov 2000 02:57:31 +0000 (02:57 +0000)]
Make sure we clean up the index data each time it is written by LaTeX.

24 years agoFix cut & paste error that describes three paramters when there are only
Fred Drake [Thu, 2 Nov 2000 21:49:17 +0000 (21:49 +0000)]
Fix cut & paste error that describes three paramters when there are only
two [bug #119729].

Update use of distutils.sysconfig that "broke" when Greg W. changed the API
[bug #119645].

24 years agoNew configure script from latest configure.in with autoconf 2.13
Guido van Rossum [Thu, 2 Nov 2000 19:33:53 +0000 (19:33 +0000)]
New configure script from latest configure.in with autoconf 2.13

24 years agoMake sure the Modules/ directory is created before writing Modules/Setup.
Fred Drake [Thu, 2 Nov 2000 17:52:56 +0000 (17:52 +0000)]
Make sure the Modules/ directory is created before writing Modules/Setup.

24 years agoPatch from Randall Hopper to fix PR #116172, "curses module fails to
Andrew M. Kuchling [Wed, 1 Nov 2000 19:59:12 +0000 (19:59 +0000)]
Patch from Randall Hopper to fix PR #116172, "curses module fails to
build on SGI":
* Check for 'sgi' preprocessor symbol, not '__sgi__'
* Surround individual character macros with #ifdef's, instead of making them
  all rely on STRICT_SYSV_CURSES

24 years agoTypo: writeable --> writable
Fred Drake [Wed, 1 Nov 2000 03:12:34 +0000 (03:12 +0000)]
Typo:  writeable --> writable
Reported by Erno Kuusela <erno@iki.fi>.

24 years agoHack ndiff to display lines w/ leading tabs more intuitively. This synchs
Tim Peters [Wed, 1 Nov 2000 02:51:27 +0000 (02:51 +0000)]
Hack ndiff to display lines w/ leading tabs more intuitively.  This synchs
ndiff w/ a custom version I made for Guido during the pre-2.0 freeze.

24 years agotrack recent change to test_extcall.py
Jeremy Hylton [Mon, 30 Oct 2000 19:41:33 +0000 (19:41 +0000)]
track recent change to test_extcall.py

24 years agoHack to force -lpthread instead instead of -lcma on HPUX, by Philipp
Guido van Rossum [Mon, 30 Oct 2000 17:45:07 +0000 (17:45 +0000)]
Hack to force -lpthread instead instead of -lcma on HPUX, by Philipp
Jocham.  Philipp asks: "Are there any success stories with HP-UX 11.00
and -lcma?  Maybe libcma is broken."

24 years agoFix for SF bug #117241
Jeremy Hylton [Mon, 30 Oct 2000 17:15:20 +0000 (17:15 +0000)]
Fix for SF bug #117241

When a method is called with no regular arguments and * args, defer
the first arg is subclass check until after the * args have been
expanded.

N.B. The CALL_FUNCTION implementation is getting really hairy; should
review it to see if it can be simplified.

24 years agodefine_module(): Change the "index sub-item" for definitions at module
Fred Drake [Mon, 30 Oct 2000 06:24:56 +0000 (06:24 +0000)]
define_module():  Change the "index sub-item" for definitions at module
                  scope to be " (in module <name>)" instead of
                  " (in <name>)" to be consistent with \withsubitem
                  usage throughout the documentation.  This achieves
                  consistency in indexing throughout the documentation.

24 years agoFix \withsubitem so that it actually works if the only indexing macro in
Fred Drake [Mon, 30 Oct 2000 06:22:22 +0000 (06:22 +0000)]
Fix \withsubitem so that it actually works if the only indexing macro in
the content body is the \ttindex macro, which seems to match actual usage.

Adjust \funcline to restore the "index sub-item" for functions to be
" (in module <name>)" instead of " (in <name>)".  This is need to match
uses of \withsubitem throughout the documentation.  (Not ideal, but
gets achieves consistency.)

24 years agoRevise the message to be a little nicer.
Fred Drake [Sun, 29 Oct 2000 13:21:45 +0000 (13:21 +0000)]
Revise the message to be a little nicer.

Have the mail sent to the python-dev and Doc-SIG lists.

24 years agoAvoid using \withsubitem and \ttindex internally; they have not proven
Fred Drake [Sun, 29 Oct 2000 05:19:16 +0000 (05:19 +0000)]
Avoid using \withsubitem and \ttindex internally; they have not proven
to be very robust.  Using \index directly fixes a lot of entries in the
index that had to be specifically read to determine that they had the
wrong parenthesized description.

24 years agoMany small markup revisions to be more consistent with markup elsewhere,
Fred Drake [Sun, 29 Oct 2000 05:10:30 +0000 (05:10 +0000)]
Many small markup revisions to be more consistent with markup elsewhere,
and to provide more consistent indexing.

Added an index entry.

Added documentation for the error and XMLParserType objects.

24 years agoParserCreate(): Added test that the namespace_separator value, if given,
Fred Drake [Sun, 29 Oct 2000 04:57:53 +0000 (04:57 +0000)]
ParserCreate():  Added test that the namespace_separator value, if given,
                 has the required length.

initpyexpat():  Provide the type object for the ParserCreate() return
                value as XMLParserType.

24 years agoDo not echo the echo command with the message telling the user that Setup
Fred Drake [Sun, 29 Oct 2000 04:28:48 +0000 (04:28 +0000)]
Do not echo the echo command with the message telling the user that Setup
may be out of date.

24 years agoDo not release unallocated Tcl objects. Closes #117278 and #117167.
Martin v. Löwis [Sun, 29 Oct 2000 00:44:43 +0000 (00:44 +0000)]
Do not release unallocated Tcl objects. Closes #117278 and  #117167.

24 years ago-- properly reset groups in findall (bug #117612)
Fredrik Lundh [Sat, 28 Oct 2000 19:30:41 +0000 (19:30 +0000)]
-- properly reset groups in findall (bug #117612)

-- fixed negative lookbehind to work correctly at the beginning
of the target string (bug #117242)

-- improved syntax check; you can no longer refer to a group
inside itself (bug #110866)

24 years agoUpdate author information.
Fred Drake [Sat, 28 Oct 2000 04:08:38 +0000 (04:08 +0000)]
Update author information.

24 years ago(py-goto-beginning-of-tqs): When searching backwards for the matching
Barry Warsaw [Fri, 27 Oct 2000 05:00:25 +0000 (05:00 +0000)]
(py-goto-beginning-of-tqs): When searching backwards for the matching
delimiter, watch out for backslash escaped delimiters.  Also use =
instead of eq for character comparison (because a character is = to
it's integer value, but not eq to it).

24 years agoAdded the -D/--docstrings option for extraction of unmarked module,
Barry Warsaw [Fri, 27 Oct 2000 04:56:28 +0000 (04:56 +0000)]
Added the -D/--docstrings option for extraction of unmarked module,
class, method, and function docstrings.

24 years agoExplained that os.path.basename() may return something different from the
Fred Drake [Thu, 26 Oct 2000 21:38:23 +0000 (21:38 +0000)]
Explained that os.path.basename() may return something different from the
basename program, as suggested by Gregor Hoffleit <gregor@hoffleit.de>.

This closes bug #119485.

24 years agoDo not require packages that are not needed.
Fred Drake [Thu, 26 Oct 2000 21:13:22 +0000 (21:13 +0000)]
Do not require packages that are not needed.

24 years agoNormalize the HTML generated for table headers.
Fred Drake [Thu, 26 Oct 2000 20:14:58 +0000 (20:14 +0000)]
Normalize the HTML generated for table headers.

24 years agoMinor simplification.
Fred Drake [Thu, 26 Oct 2000 20:01:09 +0000 (20:01 +0000)]
Minor simplification.

24 years agoUpdate the dependency information to allow the other Makefiles to handle
Fred Drake [Thu, 26 Oct 2000 19:26:47 +0000 (19:26 +0000)]
Update the dependency information to allow the other Makefiles to handle
as much of this as possible.  Avoids propogating information about how
various outputs relate (or don't!).

24 years agoNoted that building the info version requires the HTML::Element package,
Fred Drake [Thu, 26 Oct 2000 19:01:46 +0000 (19:01 +0000)]
Noted that building the info version requires the HTML::Element package,
and provided instructions for getting it installed.

24 years agoAll acknowledgements have been moved to the Doc/ACKS file.
Fred Drake [Thu, 26 Oct 2000 17:19:58 +0000 (17:19 +0000)]
All acknowledgements have been moved to the Doc/ACKS file.

Adjusted to reflect the rename of Setup.in to Setup.dist.

Added pointer to the "Distributing Python Modules" manual in the
appropriate place.

24 years agoUpdate build instructions to reflect the rename from Setup.in to Setup.dist.
Fred Drake [Thu, 26 Oct 2000 17:13:19 +0000 (17:13 +0000)]
Update build instructions to reflect the rename from Setup.in to Setup.dist.
Clarify when this file is created automatically and do not advocate
creating it unless needed.

Explain that Setup never gets overwritten.

24 years agoTest for and create Modules/Setup in the configure script if it does not
Fred Drake [Thu, 26 Oct 2000 17:09:35 +0000 (17:09 +0000)]
Test for and create Modules/Setup in the configure script if it does not
already exist.

24 years agoRename Setup.in to Setup.dist, and assume that configure will create
Fred Drake [Thu, 26 Oct 2000 17:07:40 +0000 (17:07 +0000)]
Rename Setup.in to Setup.dist, and assume that configure will create
Setup (instead of creating it from the Makefile).

24 years agoLots of small markup adjustments for consistency with the rest of the
Fred Drake [Thu, 26 Oct 2000 16:41:03 +0000 (16:41 +0000)]
Lots of small markup adjustments for consistency with the rest of the
documentation.

24 years agoUpdate a comment to be correct.
Fred Drake [Thu, 26 Oct 2000 14:57:29 +0000 (14:57 +0000)]
Update a comment to be correct.

24 years agoRemove bogus stdout redirection and use of sys.__stdout__; use
Fred Drake [Thu, 26 Oct 2000 03:56:46 +0000 (03:56 +0000)]
Remove bogus stdout redirection and use of sys.__stdout__; use
augmented print statement instead.

24 years agoRemove bogus stdout redirection and use of sys.__stdout__; use
Fred Drake [Thu, 26 Oct 2000 03:49:15 +0000 (03:49 +0000)]
Remove bogus stdout redirection and use of sys.__stdout__; use
augmented print statement instead.

24 years agoTypo: shorted --> shorter
Fred Drake [Wed, 25 Oct 2000 23:22:54 +0000 (23:22 +0000)]
Typo:  shorted --> shorter
This closes bug #117706.

24 years agoPatch 102114, Bug 11725. On OpenBSD (but apparently not on the other
Guido van Rossum [Wed, 25 Oct 2000 22:07:45 +0000 (22:07 +0000)]
Patch 102114, Bug 11725.  On OpenBSD (but apparently not on the other
BSDs) you need a leading underscore in the dlsym() lookup name.

24 years agoFix two typos in __imul__. Closes Bug #117745.
Guido van Rossum [Wed, 25 Oct 2000 21:58:20 +0000 (21:58 +0000)]
Fix two typos in __imul__.  Closes Bug #117745.

24 years agoAdd IDLE wish: access items of arrays
Guido van Rossum [Wed, 25 Oct 2000 21:18:12 +0000 (21:18 +0000)]
Add IDLE wish: access items of arrays

24 years agoSince LookupError can be raised by sys.setdefaultencoding(), we should not
Fred Drake [Wed, 25 Oct 2000 21:05:29 +0000 (21:05 +0000)]
Since LookupError can be raised by sys.setdefaultencoding(), we should not
document it as only being a base class, not matter how unlikely this is
to affect normal users.

24 years agoMarc-Andre Lemburg <mal@lemburg.com> noted that the encode() string method
Fred Drake [Wed, 25 Oct 2000 21:03:55 +0000 (21:03 +0000)]
Marc-Andre Lemburg <mal@lemburg.com> noted that the encode() string method
was added in 2.0.

24 years agoAdditions from Marc-Andre Lemburg <mal@lemburg.com>, documenting
Fred Drake [Wed, 25 Oct 2000 21:02:55 +0000 (21:02 +0000)]
Additions from Marc-Andre Lemburg <mal@lemburg.com>, documenting
getdefaultencoding() and setdefaultencoding().

24 years agoSeveral additions and updates based on text from Marc-Andre Lemburg
Fred Drake [Wed, 25 Oct 2000 20:59:52 +0000 (20:59 +0000)]
Several additions and updates based on text from Marc-Andre Lemburg
<mal@lemburg.com>.

Lots of markup reformatting to make it easier for me to read & maintain.

24 years agoMany changes.
Jeremy Hylton [Wed, 25 Oct 2000 18:10:32 +0000 (18:10 +0000)]
Many changes.

Reformatting -- long lines, "[ ]" -> "[]", a few indentation nits.

Replace calls to Node function (which constructed ast nodes) with
calls to actual constructors imported from ast module.

Optimize com_node (most frequently used method) for the common case --
the appropriate method is found in _dispatch.

Fix com_augassign to use class object's rather than node names
(rendered invalid by recent changes to ast)

Remove expensive tests for sequence-ness in com_stmt and
com_append_stmt. These tests should never fail; if they do, something
is really broken and exception will be raised elsewhere.

Fix com_stmt and com_append_stmt to use isinstance rather than
testing's type slot of ast node (this slot disappeared with recent
changes to ast).

24 years agoGenerated from rev 1.1 of ast.txt
Jeremy Hylton [Wed, 25 Oct 2000 18:02:59 +0000 (18:02 +0000)]
Generated from rev 1.1 of ast.txt

24 years agoSmall optimizations in dispatch method: 1) lookup node's __class__ once
Jeremy Hylton [Wed, 25 Oct 2000 18:02:02 +0000 (18:02 +0000)]
Small optimizations in dispatch method: 1) lookup node's __class__ once
and store in local; 2) define _preorder to be dispatch (rather than
method that called dispatch).