]> granicus.if.org Git - python/log
python
27 years agoGive more detailed error message when the argument count isn't right.
Guido van Rossum [Wed, 19 Nov 1997 16:05:40 +0000 (16:05 +0000)]
Give more detailed error message when the argument count isn't right.

27 years agoAdd declaration for PyNumber_CoerceEx().
Guido van Rossum [Wed, 19 Nov 1997 16:04:54 +0000 (16:04 +0000)]
Add declaration for PyNumber_CoerceEx().

27 years agoAdd a new function PyNumber_CoerceEx() which works just like
Guido van Rossum [Wed, 19 Nov 1997 16:03:17 +0000 (16:03 +0000)]
Add a new function PyNumber_CoerceEx() which works just like
PyNumber_Coerce() except that when the coercion can't be done and no
other exceptions happen, it returns 1 instead of raising an
exception.

Use this function in PyObject_Compare() to avoid raising an exception
simply because two objects with numeric behavior can't be coerced to a
common type; instead, proceed with the non-numeric default comparison.

Note that this is a somewhat questionable practice -- comparisons for
numeric objects shouldn't default to random behavior like this, but it
is required for backward compatibility.  (Case in point, it broke
comparison of kjDict objects to integers in Aaron Watters' kjbuckets
extension.)  A correct fix (for python 2.0) should involve a different
definiton of comparison altogether.

27 years agoMartin von Loewis' _locale module (locale.py follows tomorrow).
Guido van Rossum [Tue, 18 Nov 1997 21:03:39 +0000 (21:03 +0000)]
Martin von Loewis' _locale module (locale.py follows tomorrow).

27 years agoUndo another glitch of the automatic not-so-Grand Renaming; some local
Guido van Rossum [Tue, 18 Nov 1997 19:23:07 +0000 (19:23 +0000)]
Undo another glitch of the automatic not-so-Grand Renaming; some local
variables called 'coerce' were accidentally renamed to
'PyNumber_Coerce'.  Rename them back to coercefunc.

27 years agostart numbering things 1.5b1 now
Guido van Rossum [Tue, 18 Nov 1997 19:21:47 +0000 (19:21 +0000)]
start numbering things 1.5b1 now

27 years agoAssert that the proxies object passed in to the URLopener constructor
Guido van Rossum [Tue, 18 Nov 1997 15:50:39 +0000 (15:50 +0000)]
Assert that the proxies object passed in to the URLopener constructor
is indeed a dictionary (or a mapping).

27 years agoAdded docstrings (contributed by Martin von Loewis).
Guido van Rossum [Tue, 18 Nov 1997 15:47:55 +0000 (15:47 +0000)]
Added docstrings (contributed by Martin von Loewis).

27 years agoFix the release date for 1.5b1 to be Nov. 26
Guido van Rossum [Tue, 18 Nov 1997 15:32:15 +0000 (15:32 +0000)]
Fix the release date for 1.5b1 to be Nov. 26

27 years agoAdd support for NULL macro (like UNIX, ABC etc.).
Guido van Rossum [Tue, 18 Nov 1997 15:31:16 +0000 (15:31 +0000)]
Add support for NULL macro (like UNIX, ABC etc.).

27 years agoAdd xmllib; remove (slow) from description of htmllib.
Guido van Rossum [Tue, 18 Nov 1997 15:30:40 +0000 (15:30 +0000)]
Add xmllib; remove (slow) from description of htmllib.

27 years agodis is now documented; bisect is deemed useful.
Guido van Rossum [Tue, 18 Nov 1997 15:30:13 +0000 (15:30 +0000)]
dis is now documented; bisect is deemed useful.

27 years agoDocument s.connect_ex(), which does not raise an exception on error.
Guido van Rossum [Tue, 18 Nov 1997 15:29:20 +0000 (15:29 +0000)]
Document s.connect_ex(), which does not raise an exception on error.

27 years agoClarified reference to old profiler.
Guido van Rossum [Tue, 18 Nov 1997 15:28:46 +0000 (15:28 +0000)]
Clarified reference to old profiler.
Mention conversion to Perl-style regular expressions.

27 years agoAdded doc for errorcode dictionary.
Guido van Rossum [Tue, 18 Nov 1997 15:27:51 +0000 (15:27 +0000)]
Added doc for errorcode dictionary.

27 years agoFixed case sensitivity of attributes (they are case *sensitive*).
Guido van Rossum [Tue, 18 Nov 1997 15:27:20 +0000 (15:27 +0000)]
Fixed case sensitivity of attributes (they are case *sensitive*).

27 years agoAdding dis and xmllib docs
Guido van Rossum [Tue, 18 Nov 1997 15:11:47 +0000 (15:11 +0000)]
Adding dis and xmllib docs

27 years agoDocu for xmllib.py, by Sjoerd Mullender.
Guido van Rossum [Tue, 18 Nov 1997 15:11:22 +0000 (15:11 +0000)]
Docu for xmllib.py, by Sjoerd Mullender.

27 years agoDocu for dis.py, written by Martin von Loewis.
Guido van Rossum [Tue, 18 Nov 1997 15:10:53 +0000 (15:10 +0000)]
Docu for dis.py, written by Martin von Loewis.

27 years agoSjoerd Mullender's xml parser (based on sgmllib, somewhat).
Guido van Rossum [Tue, 18 Nov 1997 15:09:54 +0000 (15:09 +0000)]
Sjoerd Mullender's xml parser (based on sgmllib, somewhat).

27 years agoThis BUGS list hasn't been updated since the release of Python 1.2!
Guido van Rossum [Fri, 14 Nov 1997 23:32:19 +0000 (23:32 +0000)]
This BUGS list hasn't been updated since the release of Python 1.2!
It is useless.  My real bugs database is currently being maintained
with GNATS.

27 years agoThis fix (across 4 files in 3 directories) solves a subtle problem with
Guido van Rossum [Fri, 14 Nov 1997 22:24:28 +0000 (22:24 +0000)]
This fix (across 4 files in 3 directories) solves a subtle problem with
signal handlers in a fork()ed child process when Python is compiled with
thread support.  The bug was reported by Scott <scott@chronis.icgroup.com>.

What happens is that after a fork(), the variables used by the signal
module to determine whether this is the main thread or not are bogus,
and it decides that no thread is the main thread, so no signals will
be delivered.

The solution is the addition of PyOS_AfterFork(), which fixes the signal
module's variables.  A dummy version of the function is present in the
intrcheck.c source file which is linked when the signal module is not
used.

27 years agoSuggestion by David Ascher: set EXE macro to empty string and use it
Guido van Rossum [Tue, 11 Nov 1997 18:29:22 +0000 (18:29 +0000)]
Suggestion by David Ascher: set EXE macro to empty string and use it
in a few places -- so it can be set to .exe on GNUWIN32 platforms and
do the right thing.  Whatever.  (This was already done in
Modules/Makefile* but wasn't carried over here.)

27 years agoUse a %s format for the exit status -- through the magical workings of
Guido van Rossum [Tue, 11 Nov 1997 17:18:48 +0000 (17:18 +0000)]
Use a %s format for the exit status -- through the magical workings of
the aptly named MagicDict class, the value is a string, not a number,
by the time it is printed...

27 years agoAs Mark Hammond found out, it was a bad idea to add "set -e" to the
Guido van Rossum [Tue, 11 Nov 1997 17:17:55 +0000 (17:17 +0000)]
As Mark Hammond found out, it was a bad idea to add "set -e" to the
check in command -- this fails for new files!

27 years agoNew version from Neale... He promised it's the last.
Guido van Rossum [Tue, 11 Nov 1997 16:36:14 +0000 (16:36 +0000)]
New version from Neale...  He promised it's the last.

27 years agoFix memory leak in exec statement with code object -- the None returned
Guido van Rossum [Tue, 11 Nov 1997 16:29:38 +0000 (16:29 +0000)]
Fix memory leak in exec statement with code object -- the None returned
by PyEval_EvalCode() on success was never DECREF'ed.

Fix by Bernhard Herzog.

27 years agov1.1; added attribution to Neale Pickett.
Guido van Rossum [Sat, 8 Nov 1997 07:16:19 +0000 (07:16 +0000)]
v1.1; added attribution to Neale Pickett.

27 years agoSome restructuring.
Guido van Rossum [Fri, 7 Nov 1997 20:38:49 +0000 (20:38 +0000)]
Some restructuring.

All geometry manager methods that apply to a master widget instead of
to a slave widget have been moved to the Misc class, which is
inherited by all of Tk(), Toplevel() and Widget().  They have been
renamed to have their geometry manager name as a prefix,
e.g. pack_propagate(); the short names can still be used where
ambiguities are resolved so that pack has priority over place has
priority over grid (since this was the old rule).

Also, the method definitions in the Pack, Place and Grid classes now
all have their respective geometry manager name as a prefix
(e.g. pack_configure); the shorter names are aliases defined through
assignment.

A similar renaming has been done for all config() methods found
elsewhere; these have been renamed to configure() with config being
the alias (instead of the other way around).  (This may not make much
of a difference but the official Tk command name is now 'configure'
and it may help in debugging tracebacks.)

Finally, a new base class BaseWidget has been introduced, which
implements the methods common between Widget and Toplevel (the
difference between those two classes is that Toplevel has a different
__init__() but also that Toplevel doesn't inherit from Pack, Place or
Grid.

27 years agoAdded vgrindefs.
Guido van Rossum [Fri, 7 Nov 1997 19:58:10 +0000 (19:58 +0000)]
Added vgrindefs.

27 years agoPlucked this from the net.
Guido van Rossum [Fri, 7 Nov 1997 19:57:54 +0000 (19:57 +0000)]
Plucked this from the net.

27 years agoFix problem discovered by Barry: if you hit ^C to
Guido van Rossum [Fri, 7 Nov 1997 19:20:34 +0000 (19:20 +0000)]
Fix problem discovered by Barry: if you hit ^C to
sys.stdin.readline(), you get a fatal error (no current thread).  This
is because there was a call to PyErr_CheckSignals() while there was no
current thread.  I wonder how many more of these we find...  I bnetter
go hunting for PyErr_CheckSignals() now...

27 years agoSeparate out a function pystones(loops=LOOPS) which runs the benchmark
Guido van Rossum [Thu, 6 Nov 1997 15:45:05 +0000 (15:45 +0000)]
Separate out a function pystones(loops=LOOPS) which runs the benchmark
and returns a (benchtime, stones) tuple.  The main() function now
calls this and prints the report.  Fred Drake's code.

27 years agoAdd __init__.py to test package.
Guido van Rossum [Thu, 6 Nov 1997 15:41:23 +0000 (15:41 +0000)]
Add __init__.py to test package.

27 years ago#(py-emacs-features): Fixed typo in docstring.
Barry Warsaw [Thu, 6 Nov 1997 14:36:49 +0000 (14:36 +0000)]
#(py-emacs-features): Fixed typo in docstring.

27 years ago(python-font-lock-keywords): Pick up block introducing keywords with
Barry Warsaw [Thu, 6 Nov 1997 14:35:15 +0000 (14:35 +0000)]
(python-font-lock-keywords): Pick up block introducing keywords with
immediately following colons.  Sjoerd noticed this one too.  Here's a
nonsense.py file that flexes all the font-lock keyword combinations.

class A:
    class B(A):
pass
    def __init__(self):
if i == 2 and j == 3 or k == 4:
    import stuff
    from otherstuff import cool
    for i in range(cool.count):
if i == j:
    break
elif j == 1:
    continue
print i
    else:
return not i
elif q is not i:
    return lambda x: x + 1
else:
    try:
try:
    raise stuff.error
except stuff.error, v:
    print v
except:
    global q
    finally:
while q > 0:
    q = q - 1
assert q == 0

def make():
    a = A()
    exec "nonsense"
    del a

27 years ago(python-font-lock-keywords): Fix for consecutive keyword font locking,
Barry Warsaw [Wed, 5 Nov 1997 18:55:50 +0000 (18:55 +0000)]
(python-font-lock-keywords): Fix for consecutive keyword font locking,
given by Sjoerd Mullender <Sjoerd.Mullender@cwi.nl>.

27 years ago(py-emacs-features): Test for working make-temp-name, which is broken
Barry Warsaw [Wed, 5 Nov 1997 18:41:11 +0000 (18:41 +0000)]
(py-emacs-features): Test for working make-temp-name, which is broken
on NTEmacs 19.34.6.

(py-serial-number): New variable.

(py-execute-region): If make-temp-name is broken, simply append a
serial number to the string "python-" to get a temporary file name.
It's possible concurrent NTEmacs can step on each others toes, but it
makes no sense to further coddle a busted NTEmacs.

27 years agoA nice blurb that Mark Hammond wrote, aimed at Windows users.
Guido van Rossum [Wed, 5 Nov 1997 17:15:02 +0000 (17:15 +0000)]
A nice blurb that Mark Hammond wrote, aimed at Windows users.

27 years ago(py-execute-region): Fixed small bug with queuing file for execution
Barry Warsaw [Wed, 5 Nov 1997 16:56:51 +0000 (16:56 +0000)]
(py-execute-region): Fixed small bug with queuing file for execution
in a py-shell.  Temp files now get cleaned up.

27 years ago(mark-python-def-or-class): Renamed to py-mark-def-or-class globally.
Barry Warsaw [Wed, 5 Nov 1997 00:51:08 +0000 (00:51 +0000)]
(mark-python-def-or-class): Renamed to py-mark-def-or-class globally.

(py-mode-map): Moved py-mark-def-or-class from M-C-h to C-c C-m since
the old binding conflicts with the standard global backward-kill-word
binding, and this new binding is more conformant with other language
modes.  Moved py-mark-block to C-c C-k.

27 years agoGot rid of the errorstr dictionary, which is redundant now that
Guido van Rossum [Tue, 4 Nov 1997 20:22:24 +0000 (20:22 +0000)]
Got rid of the errorstr dictionary, which is redundant now that
there's os.strerror() -- also, it would form a locale liability.

27 years agoUndo half of the previous change :-(
Guido van Rossum [Tue, 4 Nov 1997 19:36:18 +0000 (19:36 +0000)]
Undo half of the previous change :-(

Setting interp->builtins to the __builtin__ module instead of to its
dictionary had the unfortunate side effect of always running in
restricted execution mode :-(

I will check in a different way of setting __main__.__builtins__ to
the __builtin__ module later.

Also, there was a typo -- a comment was unfinished, and as a result
some finalizations were not being executed.

In Bart Simpson style,

I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.
I Will Not Check In Untested Changes.

27 years ago(py-backspace-function): New variable.
Barry Warsaw [Tue, 4 Nov 1997 19:21:50 +0000 (19:21 +0000)]
(py-backspace-function): New variable.

(py-electric-backspace, py-electric-delete): Support the XEmacs 20 Way
for backspace and delete mappings.  In XEmacs 19, Emacs 19, and Emacs
20, both backspace and delete keysyms are bound to
py-electric-backspace.  In XEmacs 20, backspace and delete keysyms are
bound separately, allowing the user to specify forward or backward
deletion of the delete keysym through the variable
delete-key-deletes-forward.  All this is the Right Way To Do It and
this implementation was largely ripped from CC Mode.

27 years agoFirst round of changes, mostly subprocess stuff.
Barry Warsaw [Tue, 4 Nov 1997 18:47:06 +0000 (18:47 +0000)]
First round of changes, mostly subprocess stuff.

(py-execute-file): Better interaction with comint.  Set
comint-scroll-to-bottom-on-output to t.  Wrapper buffer change in
unwind-protect in case process filter fails.

(py-shell): Start Python with -i flag to fix tty problem on Windows;
presumably -- not yet tested.

(py-clear-queue): New function to clear the pending exec file queue.
Not currently keybound.

(py-execute-region, py-execute-buffer): Added optional async flag (use
via C-u prefix) to execute the region in a new asynchrous buffer, even
if the Python shell is running.

(py-append-to-process-buffer): Removed as obsolete.  Comint provides
this functionality.

Removed fbound test defun of match-string.  All modern X/Emacsen have
this function.

27 years agoislink() returns false, but there is no constant false! Return 0 instead.
Guido van Rossum [Tue, 4 Nov 1997 18:40:53 +0000 (18:40 +0000)]
islink() returns false, but there is no constant false!  Return 0 instead.

27 years agoWhen checking for equality, open the slave file in text mode so it
Guido van Rossum [Tue, 4 Nov 1997 17:35:43 +0000 (17:35 +0000)]
When checking for equality, open the slave file in text mode so it
will compare equal even if the master file uses only \n to terminate
lines (this is by far the most common situation).  Also, check for the
case where the master file is missing, and print the time difference
in seconds when the slave file appears newer than the master (for
debugging).

27 years agoUse ``0'' instead of ``None'' to reset the underlying object in close
Guido van Rossum [Tue, 4 Nov 1997 17:32:59 +0000 (17:32 +0000)]
Use ``0'' instead of ``None'' to reset the underlying object in close
methods.  Using None causes problems if the destructor is called after
the __builtin__ module has already been destroyed (unfortunately, this
can happen!).  I can't just delete the object because it is actually
tested for (if self._sock: ...).  Setting it to 0 is a bit weird but
works.

27 years agoFix due to Bill Noon for problem discovered by Ken Manheimer: packing
Guido van Rossum [Tue, 4 Nov 1997 17:12:33 +0000 (17:12 +0000)]
Fix due to Bill Noon for problem discovered by Ken Manheimer: packing
0.0 as float or double would yield the representation for 1.0!

27 years agoGet rid of another reference to _PyImport_Inittab (now a static array)
Guido van Rossum [Tue, 4 Nov 1997 17:04:34 +0000 (17:04 +0000)]
Get rid of another reference to _PyImport_Inittab (now a static array)
that should be PyImport_Inittab (a new pointer initialized to point to
the array).

27 years agoThe warning about thread still having a frame now only happens in
Guido van Rossum [Mon, 3 Nov 1997 22:08:36 +0000 (22:08 +0000)]
The warning about thread still having a frame now only happens in
verbose mode.

27 years agoMove the Py_{{BEGIN,END}_ALLOW,BLOCK}_THREADS macros in time_sleep()
Guido van Rossum [Mon, 3 Nov 1997 22:04:46 +0000 (22:04 +0000)]
Move the Py_{{BEGIN,END}_ALLOW,BLOCK}_THREADS macros in time_sleep()
to inside floatsleep().  This is necessary because floatsleep() does
the error handling and it must have grabbed the interpreter lock and
thread state before it can do so.

27 years agoTwo independent changes (alas):
Guido van Rossum [Mon, 3 Nov 1997 21:58:47 +0000 (21:58 +0000)]
Two independent changes (alas):

- The interp->builtins variable (and hence, __main__.__builtins__) is
once again initialized to the built-in *module* instead of its
dictionary.

- The finalization order is once again changed.  Signals are finalized
relatively early, because (1) it DECREF's the signal handlers, and if
a signal handler happens to be a bound method, deleting it could cause
problems when there's no current thread around, and (2) we don't want
to risk executing signal handlers during finalization.

27 years agoChange the signal finialization so that it also resets the signal
Guido van Rossum [Mon, 3 Nov 1997 21:53:55 +0000 (21:53 +0000)]
Change the signal finialization so that it also resets the signal
handlers.  After this has been called, our signal handlers are no
longer active!

27 years agoDoing the most complicated CVS dance I've ever seen. Barry has had
Guido van Rossum [Mon, 3 Nov 1997 17:14:19 +0000 (17:14 +0000)]
Doing the most complicated CVS dance I've ever seen.  Barry has had
his *own* RCS file for python-mode.el, and I've agreed that it would
be better if his version was in the Python source tree.  However I
don't want to totally get rid of the old RCS file (which has
interesting info such as which version was in which Python release).
So I've moved the old one to python-mode-old.el behind the scenes,
and this checkin message indicates that I'm now deleting it.
If you do an update, you will actually get Barry's *new* version!

27 years agoChanges in preparation for integrating into the Python CVS tree.
Barry Warsaw [Mon, 3 Nov 1997 16:59:38 +0000 (16:59 +0000)]
Changes in preparation for integrating into the Python CVS tree.

Introductory comment updates.

(python-font-lock-keywords): Added "assert"

(py-block-closing-keywords-re): New variable.

(py-no-outdent-re): Rewrite to use py-block-closing-keywords-re.

(py-shell): py-process-filter should no longer be necessary.  Comint
should do all the work.  Note that more fixes to the py-shell process
mechanism need to be done.

(py-execute-region): Check for empty region.  Some questionable
changes to set-buffer after shell-command-on-region.  Again, this all
needs to be closely examined for X/Emacs 19/20 compatibility.

(py-goto-beyond-final-line): py-parse-partial-sexp-works-p should no
longer be necessary.

(py-statement-closes-block-p): Use py-block-closing-keywords-re.

27 years agoAdd cast to realloc/malloc call to shut up AIX compiler. (Vladimir Marangozov)
Guido van Rossum [Fri, 31 Oct 1997 20:32:13 +0000 (20:32 +0000)]
Add cast to realloc/malloc call to shut up AIX compiler.  (Vladimir Marangozov)

27 years agoNew policy for package imports: only a directory containing
Guido van Rossum [Fri, 31 Oct 1997 18:38:52 +0000 (18:38 +0000)]
New policy for package imports: only a directory containing
__init__.py (or __init__.pyc/.pyo, whichever applies) is considered a
package.  All other subdirectories are left alone.  Should make Konrad
Hinsen happy!

27 years agoInstead of using _PyImport_Inittab[] directly, use the new "official"
Guido van Rossum [Fri, 31 Oct 1997 18:37:24 +0000 (18:37 +0000)]
Instead of using _PyImport_Inittab[] directly, use the new "official"
pointer *PyImport_Inittab which is initialized to _PyImport_Inittab.

27 years agoRemove the declaration of "internal" table _PyImport_Inittab[]; add
Guido van Rossum [Fri, 31 Oct 1997 18:36:47 +0000 (18:36 +0000)]
Remove the declaration of "internal" table _PyImport_Inittab[]; add
new "official" pointer *PyImport_Inittab.

27 years agoAdd empty __init__.py files to the test packages so the new policy
Guido van Rossum [Fri, 31 Oct 1997 18:33:41 +0000 (18:33 +0000)]
Add empty __init__.py files to the test packages so the new policy
will recognize them.

27 years agoTake out the setlocale() call. It affects atof() and will break
Guido van Rossum [Fri, 31 Oct 1997 18:25:15 +0000 (18:25 +0000)]
Take out the setlocale() call.  It affects atof() and will break
floating point literals.  Need to do this differently...

27 years agoHack suggested by Matthias Klose to pull in all relevant entry points
Guido van Rossum [Fri, 31 Oct 1997 17:00:30 +0000 (17:00 +0000)]
Hack suggested by Matthias Klose to pull in all relevant entry points
in libmath.a so they are available to mathmodule.so (in case it is
shared).  While this still gets triggered on Solaris 2.x, this appears
to be harmless there.

27 years agoAdd #include "mytime.h", needed for SunOS 4.1 (Matthias Klose).
Guido van Rossum [Fri, 31 Oct 1997 16:55:51 +0000 (16:55 +0000)]
Add #include "mytime.h", needed for SunOS 4.1 (Matthias Klose).

27 years agoNew version from Sjoerd, small bugfix + optimizations.
Guido van Rossum [Thu, 30 Oct 1997 15:27:37 +0000 (15:27 +0000)]
New version from Sjoerd, small bugfix + optimizations.

27 years agoInstall Misc/Makefile.pre.in in the config/ install directory.
Guido van Rossum [Wed, 29 Oct 1997 19:07:30 +0000 (19:07 +0000)]
Install Misc/Makefile.pre.in in the config/ install directory.
Suggestion by Konrad Hinsen.

27 years agoTests for tokenize.py (Ka-Ping Yee)
Guido van Rossum [Mon, 27 Oct 1997 22:15:06 +0000 (22:15 +0000)]
Tests for tokenize.py (Ka-Ping Yee)

27 years agoNew, fixed version with proper r"..." and R"..." support from Ka-Ping.
Guido van Rossum [Mon, 27 Oct 1997 21:17:24 +0000 (21:17 +0000)]
New, fixed version with proper r"..." and R"..." support from Ka-Ping.

27 years agoRedone (by Ka-Ping) using the new re module, and adding recognition
Guido van Rossum [Mon, 27 Oct 1997 20:44:15 +0000 (20:44 +0000)]
Redone (by Ka-Ping) using the new re module, and adding recognition
for r"..." raw strings.  (And R"..." string support added by Guido.)

27 years agoSmall wording change about global and local trace functions.
Guido van Rossum [Mon, 27 Oct 1997 19:57:20 +0000 (19:57 +0000)]
Small wording change about global and local trace functions.

27 years agoThis should hopefully finally clean up the remaining __del__ related
Guido van Rossum [Mon, 27 Oct 1997 18:56:19 +0000 (18:56 +0000)]
This should hopefully finally clean up the remaining __del__ related
problems with this module, even if an instance of a derived class is
kept alive longer than the urllib module itself...

27 years agoBugfix in match() -- the number of registers shouldn't be divided by two!
Guido van Rossum [Mon, 27 Oct 1997 18:17:19 +0000 (18:17 +0000)]
Bugfix in match() -- the number of registers shouldn't be divided by two!

27 years agoUse __dict__.update(state) instead of for loop over state.items() and
Guido van Rossum [Sun, 26 Oct 1997 17:00:25 +0000 (17:00 +0000)]
Use __dict__.update(state) instead of for loop over state.items() and
call to setattr().  This changes semantics, following the change
already implemented in pickle.

Also reindented a few lines properly.

27 years agoOoops, found too more references to old conditions on floating point values.
Fred Drake [Fri, 24 Oct 1997 21:15:55 +0000 (21:15 +0000)]
Ooops, found too more references to old conditions on floating point values.

27 years agoRemoved obsolete discussion on _xdr helper module.
Fred Drake [Fri, 24 Oct 1997 21:14:36 +0000 (21:14 +0000)]
Removed obsolete discussion on _xdr helper module.

27 years agore -> regex conversions by Sjoerd.
Guido van Rossum [Fri, 24 Oct 1997 14:46:16 +0000 (14:46 +0000)]
re -> regex conversions by Sjoerd.

27 years agoNew module -- converts regex regular expressions to re style.
Guido van Rossum [Thu, 23 Oct 1997 22:43:50 +0000 (22:43 +0000)]
New module -- converts regex regular expressions to re style.
There are two ways to use this -- as a filter (e.g. using C-U M-| on a
regex string literal in an Emacs buffer) or from a Python program
which imports this as a module.  Read the doc string for more info,
and also some caveats (some cases aren't handled right).

27 years agoAlthough it's hard to be sure, I *think* this is a working conversion
Guido van Rossum [Thu, 23 Oct 1997 19:09:21 +0000 (19:09 +0000)]
Although it's hard to be sure, I *think* this is a working conversion
from regex to re style regular expressions.  This should make sgmllib
and htmllib threadsafe, so I can now create a threaded version of
webchecker...

27 years agoRemove redundant import regsub.
Guido van Rossum [Thu, 23 Oct 1997 17:50:10 +0000 (17:50 +0000)]
Remove redundant import regsub.

27 years agoDaniel Larsson discovered another cut-n-paste bug in the Olimit story...
Guido van Rossum [Thu, 23 Oct 1997 17:43:11 +0000 (17:43 +0000)]
Daniel Larsson discovered another cut-n-paste bug in the Olimit story...

27 years agoConvert all remaining *simple* cases of regex usage to re usage.
Guido van Rossum [Wed, 22 Oct 1997 21:00:49 +0000 (21:00 +0000)]
Convert all remaining *simple* cases of regex usage to re usage.

27 years agoAdd pcre to the list of safe modules.
Guido van Rossum [Wed, 22 Oct 1997 20:56:32 +0000 (20:56 +0000)]
Add pcre to the list of safe modules.

27 years agoRemove outdated comments.
Guido van Rossum [Wed, 22 Oct 1997 20:54:50 +0000 (20:54 +0000)]
Remove outdated comments.

27 years agoRemoved some unneeded imports, moved others around.
Guido van Rossum [Wed, 22 Oct 1997 20:52:53 +0000 (20:52 +0000)]
Removed some unneeded imports, moved others around.

27 years agoRemoved unneeded imports of regex and regsub.
Guido van Rossum [Wed, 22 Oct 1997 20:50:37 +0000 (20:50 +0000)]
Removed unneeded imports of regex and regsub.

27 years agoUse re instead of regex. Also remove bogus return statement from __init__().
Guido van Rossum [Wed, 22 Oct 1997 20:49:52 +0000 (20:49 +0000)]
Use re instead of regex.  Also remove bogus return statement from __init__().

27 years agoAdded pgrep() function, which is like grep/egrep/emgrep but uses Perl
Guido van Rossum [Wed, 22 Oct 1997 20:44:58 +0000 (20:44 +0000)]
Added pgrep() function, which is like grep/egrep/emgrep but uses Perl
syntax, by virtue of the new re module.

27 years agoDeleted this long-obsolete module
Guido van Rossum [Wed, 22 Oct 1997 16:29:53 +0000 (16:29 +0000)]
Deleted this long-obsolete module

27 years agoAdded "new" symbol defns for RE_ANSI_HEX and RE_NO_GNU_EXTENSIONS --
Guido van Rossum [Wed, 22 Oct 1997 16:28:53 +0000 (16:28 +0000)]
Added "new" symbol defns for RE_ANSI_HEX and RE_NO_GNU_EXTENSIONS --
hopefully the last maintenance on this module, ever.

27 years agoPut {} around \^ in sets to force latex2html to do the right thing...
Guido van Rossum [Wed, 22 Oct 1997 03:03:44 +0000 (03:03 +0000)]
Put {} around \^ in sets to force latex2html to do the right thing...

27 years agoNew CObject from Jim Fulton, adds PyCObject_FromVoidPtrAndDesc() and
Guido van Rossum [Tue, 21 Oct 1997 19:48:35 +0000 (19:48 +0000)]
New CObject from Jim Fulton, adds PyCObject_FromVoidPtrAndDesc() and
PyCObject_GetDesc().

27 years agoForgot to add .cxx and .cpp to the second case statement.
Guido van Rossum [Tue, 21 Oct 1997 19:30:29 +0000 (19:30 +0000)]
Forgot to add .cxx and .cpp to the second case statement.
Thanks to Daniel Larsson.

27 years agoSome patches to Lee Busby's fpectl mods that accidentally didn't make it
Guido van Rossum [Mon, 20 Oct 1997 23:50:01 +0000 (23:50 +0000)]
Some patches to Lee Busby's fpectl mods that accidentally didn't make it
into 1.5a4.

27 years agoRemove "." or "" from front of sys.path (this bit me once).
Guido van Rossum [Mon, 20 Oct 1997 23:48:01 +0000 (23:48 +0000)]
Remove "." or "" from front of sys.path (this bit me once).

27 years agoCatch KeyboardInterrupt separately and propagate it, instead of
Guido van Rossum [Mon, 20 Oct 1997 23:46:54 +0000 (23:46 +0000)]
Catch KeyboardInterrupt separately and propagate it, instead of
reporting a "crash".

Use sys.exc_info() instead of sys.exc_type and sys.exc_value.

27 years agoAdd optional 4th argument to count(), matching find() etc.
Guido van Rossum [Mon, 20 Oct 1997 23:31:15 +0000 (23:31 +0000)]
Add optional 4th argument to count(), matching find() etc.
Also change all occurrences of "x == None" to "x is None" (not that it
matters much, these functions are all reimplemented in strop -- but
count() is not).

27 years agoPatch by Charles G. Waldman to add optional user and password
Guido van Rossum [Mon, 20 Oct 1997 23:29:44 +0000 (23:29 +0000)]
Patch by Charles G. Waldman to add optional user and password
arguments to NNTP.__init__(), for nntp servers that need them.

27 years agoWrite a str() function for class objects that returns
Guido van Rossum [Mon, 20 Oct 1997 23:26:11 +0000 (23:26 +0000)]
Write a str() function for class objects that returns
"modulename.classname" instead of returning the same as repr().

27 years agoDon't use sscanf(s, "%x", &c) to parse \xX... escapes; hardcode it.
Guido van Rossum [Mon, 20 Oct 1997 23:24:07 +0000 (23:24 +0000)]
Don't use sscanf(s, "%x", &c) to parse \xX... escapes; hardcode it.

27 years agoAdd getintarg(), getlongarg(), getstrarg() to macros since these no
Guido van Rossum [Mon, 20 Oct 1997 23:22:07 +0000 (23:22 +0000)]
Add getintarg(), getlongarg(), getstrarg() to macros since these no
longer exist in the general headers.