]> granicus.if.org Git - python/log
python
26 years agoPatch by Greg Stein to document the 'P' flag.
Guido van Rossum [Mon, 21 Sep 1998 14:44:34 +0000 (14:44 +0000)]
Patch by Greg Stein to document the 'P' flag.

26 years agore.findall(): Mark as added in 1.5.2.
Fred Drake [Fri, 18 Sep 1998 17:11:49 +0000 (17:11 +0000)]
re.findall():  Mark as added in 1.5.2.

26 years agoPatches from Greg Stein to support 'P' format in struct module's
Guido van Rossum [Fri, 18 Sep 1998 14:14:13 +0000 (14:14 +0000)]
Patches from Greg Stein to support 'P' format in struct module's
native format, as void* (translated to Python int or long).
Also adds PyLong_FromVoidPtr and PyLong_AsVoidPtr to longobject.c.

26 years agoGet rid of 'ppp' that accidentally crept in.
Guido van Rossum [Thu, 17 Sep 1998 17:07:15 +0000 (17:07 +0000)]
Get rid of 'ppp' that accidentally crept in.

26 years agoAdded optional mouseregion parameter to WaitNextEvent (which is now
Jack Jansen [Thu, 17 Sep 1998 15:28:58 +0000 (15:28 +0000)]
Added optional mouseregion parameter to WaitNextEvent (which is now
manually generated).

26 years agoRichard Wolff's changes, documenting his changes to pdb.py.
Guido van Rossum [Thu, 17 Sep 1998 15:11:51 +0000 (15:11 +0000)]
Richard Wolff's changes, documenting his changes to pdb.py.

26 years agoUntabified and applied Richard Wolff's changes (plus my own reflowing
Guido van Rossum [Thu, 17 Sep 1998 15:01:38 +0000 (15:01 +0000)]
Untabified and applied Richard Wolff's changes (plus my own reflowing
of some paragraphs).

26 years agoRichard Wolff's additional changes; some layout nits, and change the
Guido van Rossum [Thu, 17 Sep 1998 15:00:30 +0000 (15:00 +0000)]
Richard Wolff's additional changes; some layout nits, and change the
alias delimiter to ';;'.

26 years agoDuplicate the decls for PySys_WriteStd{out,err} here so the VC++
Guido van Rossum [Thu, 17 Sep 1998 14:00:22 +0000 (14:00 +0000)]
Duplicate the decls for PySys_WriteStd{out,err} here so the VC++
compiler doesn't grumble.  Greg Stein's suggestion.

26 years agoAt Greg Stein's request, add PyLong_*LongLong entry points.
Guido van Rossum [Thu, 17 Sep 1998 13:58:37 +0000 (13:58 +0000)]
At Greg Stein's request, add PyLong_*LongLong entry points.

26 years agoDefine SIZEOF_LONG_LONG (as 8) at Greg Stein's request.
Guido van Rossum [Thu, 17 Sep 1998 13:19:36 +0000 (13:19 +0000)]
Define SIZEOF_LONG_LONG (as 8) at Greg Stein's request.

26 years agoIn load_inst(), when instantiating an instance the old way (i.e. when
Guido van Rossum [Tue, 15 Sep 1998 20:25:57 +0000 (20:25 +0000)]
In load_inst(), when instantiating an instance the old way (i.e. when
there's an __getinitargs__() method), if a TypeError occurs, catch and
reraise it but add info to the error about the class name being
instantiated.  This makes debugging a lot easier if __getinitargs__()
returns something bogus (e.g. a string instead of a singleton tuple).

26 years agoIgnore the TclError exception raised when deleting the registration
Guido van Rossum [Mon, 14 Sep 1998 19:06:39 +0000 (19:06 +0000)]
Ignore the TclError exception raised when deleting the registration
for callit, used by the after() command.  This can happen when the
callback deletes the window.

26 years agoSeveral changes that Python carry on in the face of errors in the
Barry Warsaw [Mon, 14 Sep 1998 18:51:11 +0000 (18:51 +0000)]
Several changes that Python carry on in the face of errors in the
initialization of class exceptions.  Specifically:

init_class_exc(): This function now returns an integer status of the
class exception initialization.  No fatal errors in this method now.
Also, use PySys_WriteStderr() when writing error messages.  When an
error occurs in this function, 0 is returned, but the partial creation
of the exception classes is not undone (this happens elsewhere).

Things that could trigger the fallback:

    - exceptions.py fails to be imported (due to syntax error, etc.)

    - one of the exception classes is missing (e.g. due to library
      version mismatch)

    - exception class can't be inserted into __builtin__'s dictionary

    - MemoryError instance can't be pre-allocated

    - some other PyErr_Occurred

newstdexception(): Changed the error message.  This is still a fatal
error because if the string based exceptions can't be created, we
really can't continue.

initerrors(): Be sure to xdecref the .exc field, which might be
non-NULL if class exceptions init was aborted.

_PyBuiltin_Init_2(): If class exception init fails, print a warning
message and reinstate the string based exceptions.

26 years agoAdded the mainloop() call.
Guido van Rossum [Mon, 14 Sep 1998 17:52:29 +0000 (17:52 +0000)]
Added the mainloop() call.

26 years agonannified
Guido van Rossum [Mon, 14 Sep 1998 17:51:14 +0000 (17:51 +0000)]
nannified

26 years agoThere was a confusion in my checkin of the code to support list() with
Guido van Rossum [Mon, 14 Sep 1998 17:36:51 +0000 (17:36 +0000)]
There was a confusion in my checkin of the code to support list() with
and without a message number argument: the argument was called 'msg'
but the code expected it to be called 'which'.  In line with the other
methods, I've renamed the argument to 'which', and adapted the doc
string not to refer to 'msg'.

26 years agonannified
Guido van Rossum [Mon, 14 Sep 1998 16:44:15 +0000 (16:44 +0000)]
nannified

26 years agoPrint serious errors to stderr instead of stdout.
Guido van Rossum [Mon, 14 Sep 1998 16:22:21 +0000 (16:22 +0000)]
Print serious errors to stderr instead of stdout.

26 years ago#Code rearranging to quiet byte-compiler
Barry Warsaw [Mon, 14 Sep 1998 16:16:18 +0000 (16:16 +0000)]
#Code rearranging to quiet byte-compiler

26 years agonannified
Guido van Rossum [Mon, 14 Sep 1998 16:03:02 +0000 (16:03 +0000)]
nannified

26 years agoNannified.
Guido van Rossum [Mon, 14 Sep 1998 16:02:28 +0000 (16:02 +0000)]
Nannified.

26 years agonannified
Guido van Rossum [Mon, 14 Sep 1998 15:57:09 +0000 (15:57 +0000)]
nannified

26 years agonannified.
Guido van Rossum [Mon, 14 Sep 1998 15:56:40 +0000 (15:56 +0000)]
nannified.

26 years agoPatch by Marc-Andre Lemburg: use re module to compare test results.
Guido van Rossum [Mon, 14 Sep 1998 15:50:40 +0000 (15:50 +0000)]
Patch by Marc-Andre Lemburg: use re module to compare test results.
This makes it possible to accept that on Linux %w returns "01" instead
of "1", for example.

26 years agoUtility to replace CRLF with LF in argument files.
Guido van Rossum [Mon, 14 Sep 1998 15:46:41 +0000 (15:46 +0000)]
Utility to replace CRLF with LF in argument files.

26 years agoUtility to replace LF with CRLF in argument files.
Guido van Rossum [Mon, 14 Sep 1998 15:46:15 +0000 (15:46 +0000)]
Utility to replace LF with CRLF in argument files.

26 years agoUtility to untabify stubber results.
Guido van Rossum [Mon, 14 Sep 1998 15:43:38 +0000 (15:43 +0000)]
Utility to untabify stubber results.

26 years agonannified.
Guido van Rossum [Mon, 14 Sep 1998 15:34:45 +0000 (15:34 +0000)]
nannified.

26 years agoMac-specific mod to enable aliases on import paths.
Guido van Rossum [Mon, 14 Sep 1998 13:40:53 +0000 (13:40 +0000)]
Mac-specific mod to enable aliases on import paths.
(Jack Jansen and/or Just van Rossum)

26 years agoRemove some unused variables from gethostbyaddr_ex and gethostbyaddr,
Guido van Rossum [Sun, 13 Sep 1998 15:52:13 +0000 (15:52 +0000)]
Remove some unused variables from gethostbyaddr_ex and gethostbyaddr,
discovered by Marc Lemburg.

26 years agoUndo victim of careless global substitute ("long long_hash" was
Guido van Rossum [Sun, 13 Sep 1998 15:37:20 +0000 (15:37 +0000)]
Undo victim of careless global substitute ("long long_hash" was
changed to "LONG_LONG_hash" in the list of forward decls).  Discovered
by Jason Harper.

26 years agoDon't remember...
Jack Jansen [Sun, 13 Sep 1998 15:03:55 +0000 (15:03 +0000)]
Don't remember...

26 years agoRichard Wolff's changes:
Guido van Rossum [Sat, 12 Sep 1998 14:42:23 +0000 (14:42 +0000)]
Richard Wolff's changes:

pdb.doc Updated to reflect better the various changes.

26 years agoRichard Wolff's changes:
Guido van Rossum [Fri, 11 Sep 1998 22:50:09 +0000 (22:50 +0000)]
Richard Wolff's changes:

pdb.py  Uses the Breakpoint class so one can enable/disable breakpoints,
set temporary ones, set ignore counts, and conditions.  The last
can be set using the 'b' command
b 243 , i>4 ( b 243,i>4 if you are space adverse)
or with the condition command so conditions can be changed
for a particular breakpoint.

Breakpoints are numbered from 1 on, and if a breakpoint is deleted,
the number is not reused.  All the breakpoint handling commands
refer to breakpoints by number.  To be consistent, the clear command
does so as well, which is the one change from the original pdb that
is not transparent.  Thus only the breakpoint command 'b' uses a
line number or file:line or method.  You can also give
b whrandom.random    and the method will be searched for along
sys.path.  This is implemented with an 'egrep' command and so
is not as portable as it might be.  [ see  lineinfo() and
lineinfoCmd ]

Breakpoints cannot be set at a line that is blank or a '#' comment
or starts a triply quoted comment.  This is because I would like
this behavior in my DDD interface and think it reasonable for
pdb as well.  It can be removed readily, however as it is all
incorporated in the routine checkline().  If one attempts to
set a breakpoint at a 'def' line, the breakpoint is automatically
moved to the first executable line after the 'def'.  This too is
in checkline().

do_EOF() returns zero so typing an end-of-file character as a command
does nothing.  'quit' does the quitting.

The routine defaultFile() is present so as to preserve the current
pdb behavior and yet allow me to override it in pydb.

There's some code in lineinfo() that is probably mainly useful only
for pydb and if you prefer, much up to the comment "Best first guess"
could be removed.

Keith Davidson provided the code for handling $HOME/.pdbrc and
./.pdbrc, and it has been incorporated.  He also provided the
alias handling routine.  I modified it a bit so it could live
nicely in precmd().  He and I have been in contact; he has the
new pdb (and pydb) with his code incorporated.  He also asked
about the possibility of allowing multiple commands on one
line, such as step;step  or s;s  or with an alias such as
alias ct tbreak %1 ; continue
and since it was so easy, that's in place as well.  It's a simple
'split the line at the first ";"' operation and puts the second
half in the command queue (self.cmdqueue).  This has the unfortunate
effect of destroying a line like   print "i: "+i+"; j: "+j
but either there's a simple way to deal with this, or my attitude
will remain that pdb is a debugger, not a compiler/parser/etc.
An alias like   alias 4s  s;;s;
will work because the adjacent and trailing ";" act like a <cr> which
repeats the last command.  Of course, either s;s;s;s or s;;;  would be
a bit more sensible.

The help commands have been updated.

26 years agoRichard Wolff's changes:
Guido van Rossum [Fri, 11 Sep 1998 22:38:35 +0000 (22:38 +0000)]
Richard Wolff's changes:

bdb.py  now has a class definition called Breakpoint along with
associated methods.  There's no reason why this class has to
be there; if you prefer it elsewhere, 'tis easily done.

(Minor reformatting by GvR; e.g. moved Breakpoint's doc string to
proper point.)

26 years agoRichard Wolff's changes:
Guido van Rossum [Fri, 11 Sep 1998 22:33:08 +0000 (22:33 +0000)]
Richard Wolff's changes:

cmd.py  has incorporated the changes we discussed a couple of weeks ago
(a command queue, returning line from precmd, and stop from postcmd)
and some changes to help that were occasioned because I wanted to
inherit from pdb which inherits from cmd.py and the help routine
didn't look for commands or the associated help deeply enough.

26 years agoAdd a \label to the section "Fancier Output Formatting" (it's referenced from
Fred Drake [Fri, 11 Sep 1998 16:21:55 +0000 (16:21 +0000)]
Add a \label to the section "Fancier Output Formatting" (it's referenced from
the KOE docs).

Some markup nits.

26 years agoMarkup nits.
Fred Drake [Thu, 10 Sep 1998 20:21:00 +0000 (20:21 +0000)]
Markup nits.

26 years agocorrect bogus instructions 'ccoment out' -> 'uncomment'
Jeremy Hylton [Thu, 10 Sep 1998 20:18:09 +0000 (20:18 +0000)]
correct bogus instructions 'ccoment out' -> 'uncomment'

26 years agodo_cmd_methodlineni(): New function, needed because the ref. manual now
Fred Drake [Thu, 10 Sep 1998 18:59:13 +0000 (18:59 +0000)]
do_cmd_methodlineni():  New function, needed because the ref. manual now
actually uses this markup.

26 years agoAdded comment about support for rational numbers, and pointer to
Fred Drake [Thu, 10 Sep 1998 18:42:55 +0000 (18:42 +0000)]
Added comment about support for rational numbers, and pointer to
Demos/classes/Rat.py.

26 years agoMarkup nits.
Fred Drake [Thu, 10 Sep 1998 18:25:58 +0000 (18:25 +0000)]
Markup nits.

Fixes/index improvements from Michael Ernst <mernst@cs.washington.edu>.

26 years agoFixed the words in the comment and error message about defining FULL_PATH.
Guido van Rossum [Thu, 10 Sep 1998 18:22:18 +0000 (18:22 +0000)]
Fixed the words in the comment and error message about defining FULL_PATH.
(It's not SCRIPTPATH!)

26 years agoinclude <string.h> to get prototype for strcmp
Jeremy Hylton [Thu, 10 Sep 1998 18:10:59 +0000 (18:10 +0000)]
include <string.h> to get prototype for strcmp

change error messages to be a little more straightforward

change definition of FULL_PATH so that an error is raised if the
setuid wrapper is used un-edited

26 years agoFix typo in URL; reported by Michael Ernst <mernst@cs.washington.edu>.
Fred Drake [Thu, 10 Sep 1998 17:55:32 +0000 (17:55 +0000)]
Fix typo in URL; reported by Michael Ernst <mernst@cs.washington.edu>.

26 years agoApparently on AIX when using gcc you need to call pthread_init()
Guido van Rossum [Thu, 10 Sep 1998 03:04:40 +0000 (03:04 +0000)]
Apparently on AIX when using gcc you need to call pthread_init()
(which is not a POSIX threads call!).  Reported and confirmed by Brad
Howes.

26 years ago\itembreak,
Fred Drake [Wed, 9 Sep 1998 15:27:43 +0000 (15:27 +0000)]
\itembreak,
\itemjoin:  Remove obsolete macros.

\itemnewline:  Rename to \py@itemnewline, since it's only used internally.

26 years agoFix a bug where comparison of a rational with a float failed because
Guido van Rossum [Wed, 9 Sep 1998 14:07:06 +0000 (14:07 +0000)]
Fix a bug where comparison of a rational with a float failed because
the difference got converted to float.
Put brackets around the string representation of (non-integer)
rationals.
(Sjoerd Mullender.)

26 years agoMichael P. Reilly suggested this fix: makesetup wants to prepend
Guido van Rossum [Wed, 9 Sep 1998 14:04:51 +0000 (14:04 +0000)]
Michael P. Reilly suggested this fix: makesetup wants to prepend
"$(srcdir)/" before all source files even when is starts with "/".

26 years agoPatch suggested by Perry Stoll -- os.path.normpath(".//x") returned
Guido van Rossum [Tue, 8 Sep 1998 22:28:06 +0000 (22:28 +0000)]
Patch suggested by Perry Stoll -- os.path.normpath(".//x") returned
"/x", should return "x".

26 years agoSplit tcl/tk libs out of other libraries (don't want to build these always)
Jack Jansen [Tue, 8 Sep 1998 19:43:32 +0000 (19:43 +0000)]
Split tcl/tk libs out of other libraries (don't want to build these always)

26 years agoMods by Just to allow aliases in sys.path entries
Jack Jansen [Mon, 7 Sep 1998 11:36:17 +0000 (11:36 +0000)]
Mods by Just to allow aliases in sys.path entries

26 years agoBeh, new version must be edited in a separate file.
Guido van Rossum [Fri, 4 Sep 1998 21:20:29 +0000 (21:20 +0000)]
Beh, new version must be edited in a separate file.

26 years agoUse SCRIPT_NAME (if defined) instead of hardcoding /cgi-bin/ when
Guido van Rossum [Fri, 4 Sep 1998 21:19:55 +0000 (21:19 +0000)]
Use SCRIPT_NAME (if defined) instead of hardcoding /cgi-bin/ when
sending the cookie.  (Suggestion by AMK.)

26 years agoCorrect typo in #ifdef: PY_THREAD_D4, should be PY_PTHREAD_D4.
Guido van Rossum [Fri, 4 Sep 1998 13:38:32 +0000 (13:38 +0000)]
Correct typo in #ifdef: PY_THREAD_D4, should be PY_PTHREAD_D4.
Reported by Jonathan Giddy.

26 years agoShould no longer surround PyOS_Readline() call with
Guido van Rossum [Thu, 3 Sep 1998 22:25:05 +0000 (22:25 +0000)]
Should no longer surround PyOS_Readline() call with
Py_{BEGIN,END}_ALLOW_THREADS macros.  Also get rid of the declaration
for it (it's now in pythonrun.h).

26 years agoEasy optimizations of urlparse for the common case of parsing an http URL.
Jeremy Hylton [Wed, 2 Sep 1998 21:53:16 +0000 (21:53 +0000)]
Easy optimizations of urlparse for the common case of parsing an http URL.

1. use dict.get instead of try/except KeyError
2. if the url scheme is 'http' then avoid the series of
   'if var in [someseq]:'.  instead, inline all of the code.
3. find = string.find

26 years agoFix suggested by movits@lockstar.com (plus doc string by myself)
Guido van Rossum [Wed, 2 Sep 1998 14:42:02 +0000 (14:42 +0000)]
Fix suggested by movits@lockstar.com (plus doc string by myself)
for LIST command with msg argument.

26 years agoMake bind variants without a sequence return a tuple of sequences
Guido van Rossum [Mon, 31 Aug 1998 16:54:33 +0000 (16:54 +0000)]
Make bind variants without a sequence return a tuple of sequences
(formerly it returned a string, which wasn't very convenient).

Add image commands to the Text widget (these are new in Tk 8.0).

26 years agoNow that we have standard (optional) long long support, the long long
Guido van Rossum [Mon, 31 Aug 1998 14:14:09 +0000 (14:14 +0000)]
Now that we have standard (optional) long long support, the long long
support in this module can go.  The patch only deletes code
(PyLong_FromLongLong() and PyLong_AsLongLong()).  By Sjoerd Mullender.

26 years agoThe previous fix was still broken; the Py_END_ALLOW_THREADS macro was
Guido van Rossum [Sat, 29 Aug 1998 16:03:27 +0000 (16:03 +0000)]
The previous fix was still broken; the Py_END_ALLOW_THREADS macro was
never executed because of a return statement.  Sigh.

26 years ago(py-in-literal): How'd this get through? c-point => py-point
Barry Warsaw [Sat, 29 Aug 1998 06:32:31 +0000 (06:32 +0000)]
(py-in-literal): How'd this get through?  c-point => py-point

26 years agoIgnore the generated README file.
Fred Drake [Fri, 28 Aug 1998 21:34:50 +0000 (21:34 +0000)]
Ignore the generated README file.

26 years agoWhen building the tarball of the PostScript files, also generate & include a
Fred Drake [Fri, 28 Aug 1998 21:17:58 +0000 (21:17 +0000)]
When building the tarball of the PostScript files, also generate & include a
little README giving the page count of each file when printed.

26 years agoTarget to generate a little README file with the page counts of the PostScript
Fred Drake [Fri, 28 Aug 1998 21:17:07 +0000 (21:17 +0000)]
Target to generate a little README file with the page counts of the PostScript
files.

26 years agoScript to generate page counts of the PostScript files.
Fred Drake [Fri, 28 Aug 1998 21:16:28 +0000 (21:16 +0000)]
Script to generate page counts of the PostScript files.

26 years agoSupport long options that take arguments of the form --option=value as well
Fred Drake [Fri, 28 Aug 1998 20:21:04 +0000 (20:21 +0000)]
Support long options that take arguments of the form --option=value as well
as --option value.

Minor nits cleaned up.

26 years agoMarkup changes in the section on disciplines to match method descriptions
Fred Drake [Fri, 28 Aug 1998 20:03:12 +0000 (20:03 +0000)]
Markup changes in the section on disciplines to match method descriptions
a little better, and produce better HTML.

Add some index entries.

26 years agoWe now assume that PyOS_Readline() is called with the interpreter lock
Guido van Rossum [Thu, 27 Aug 1998 19:43:43 +0000 (19:43 +0000)]
We now assume that PyOS_Readline() is called with the interpreter lock
held.  It releases the lock around the call to the function pointed to by
PyOS_ReadlineFunctionPointer (default PyOS_StdioReadline()).

26 years agoAdd a 'volatile' to the declaration of threadid in get_thread_ident().
Guido van Rossum [Thu, 27 Aug 1998 19:21:53 +0000 (19:21 +0000)]
Add a 'volatile' to the declaration of threadid in get_thread_ident().
According to Vladimir Marangozov, this is necessary for AIX, where
high optimization levels inline this function and then get it wrong :-(

26 years agoDon't release the interpreter lock around PyParser_ParseFile().
Guido van Rossum [Thu, 27 Aug 1998 19:14:49 +0000 (19:14 +0000)]
Don't release the interpreter lock around PyParser_ParseFile().
It is needed so that tokenizer.c can use PySys_WriteStderr().

26 years agoChanges by Richard Wolff:
Guido van Rossum [Thu, 27 Aug 1998 19:02:51 +0000 (19:02 +0000)]
Changes by Richard Wolff:

1) I added a command queue which is helpful to me (at least so far) and
   would also allow syntax like 's;s' (step; step) in conjunction with precmd
2) doc_leader allows the derived class to print a message before the help
   output.  Defaults to current practise of a blank line
3) nohelp allows one to override the 'No help on' message.  I need
    'Undefined command: "%s".  Try "help".'
4) Pass line to self.precmd to allow one to do some parsing: change first
   word to lower case, strip out a leading number, whatever.
5) Pass the result of onecmd and the input line to postcmd.  This allows
   one to ponder the stop result before it is effective.
6) emptyline() requires a   if self.lastcmd:  conditional because if the
   first command is null (<cr>), you get an infinite recursion with the
   code as it stands.

26 years agoAdded a note about the buildlibs script.
Jack Jansen [Thu, 27 Aug 1998 13:39:01 +0000 (13:39 +0000)]
Added a note about the buildlibs script.

26 years agofix typo in keyword argument 'allow_frament' should be 'allow_fragment'
Jeremy Hylton [Tue, 25 Aug 1998 19:45:24 +0000 (19:45 +0000)]
fix typo in keyword argument 'allow_frament' should be 'allow_fragment'

26 years ago__file__ used to be always set to the .pyc (or .pyo) file, even if
Guido van Rossum [Tue, 25 Aug 1998 18:44:34 +0000 (18:44 +0000)]
__file__ used to be always set to the .pyc (or .pyo) file, even if
that file in fact did not exist or at least was not used.  Change this
so that __file__ is *only* set to the .pyc/.pyo file when it actually
read the code object from it; otherwise __file__ is set to the .py
file.

26 years agoComment out the print statement about underflow. (This only seems to
Guido van Rossum [Tue, 25 Aug 1998 18:22:17 +0000 (18:22 +0000)]
Comment out the print statement about underflow.  (This only seems to
happen when you use a non-keyword argument after a keyword argument,
and in this case you also get a syntax error.  I fully suspect that
the underflow is caused by the code that stops generating code when it
detects the syntax error, but I can't find the culprit right now.  I
know, I know.)

26 years agoAdd the type of the object to the error message about calling a non-function.
Guido van Rossum [Tue, 25 Aug 1998 18:16:54 +0000 (18:16 +0000)]
Add the type of the object to the error message about calling a non-function.

26 years agoReplace all calls to fprintf(stderr, ...) with PySys_WriteStderr(...).
Guido van Rossum [Tue, 25 Aug 1998 18:13:04 +0000 (18:13 +0000)]
Replace all calls to fprintf(stderr, ...) with PySys_WriteStderr(...).

26 years agoAdd a version of PySys_WriteStderr() that writes to stderr, so we can
Guido van Rossum [Tue, 25 Aug 1998 18:12:36 +0000 (18:12 +0000)]
Add a version of PySys_WriteStderr() that writes to stderr, so we can
use it in tokenizer.c.

26 years agoRestructure the file so that it is never empty. No longer needs
Guido van Rossum [Tue, 25 Aug 1998 17:48:25 +0000 (17:48 +0000)]
Restructure the file so that it is never empty.  No longer needs
Metrowerks specific #ifdef.

26 years agoPatch by Mark Hammond to support 64-bit ints on MS platforms.
Guido van Rossum [Tue, 25 Aug 1998 16:07:15 +0000 (16:07 +0000)]
Patch by Mark Hammond to support 64-bit ints on MS platforms.
The MS compiler doesn't call it 'long long', it uses __int64,
so a new #define, LONG_LONG, has been added and all occurrences
of 'long long' are replaced with it.

26 years agoIgnore cruft generating by the test run for hello.py.
Guido van Rossum [Tue, 25 Aug 1998 15:27:36 +0000 (15:27 +0000)]
Ignore cruft generating by the test run for hello.py.

26 years agoY2K fix affecting asctime(), mktime(), strftime().
Guido van Rossum [Tue, 25 Aug 1998 14:51:12 +0000 (14:51 +0000)]
Y2K fix affecting asctime(), mktime(), strftime().

2-digit years are now converted using rules that are (according to
Fredrik Lundh) recommended by POSIX or X/Open: 0-68 mean 2000-2068,
69-99 mean 1969-1999.

2-digit years are now only accepted if time.accept2dyear is set to a
nonzero integer; if it is zero or not an integer or absent, only year
values >= 1900 are accepted.  Year values 100-1899 and negative year
values are never accepted.

The initial value of time.accept2dyear depends on the environment
variable PYTHONY2K: if PYTHONY2K is set and non-empty,
time.accept2dyear is initialized to 0; if PYTHONY2K is empty or not
set, time.accept2dyear is initialized to 0.

26 years agoDescribe what happens when ``raise C, x'' is called where C is a class
Guido van Rossum [Tue, 25 Aug 1998 14:45:41 +0000 (14:45 +0000)]
Describe what happens when ``raise C, x'' is called where C is a class
and x is not an instance of C (nor of a class derived of C).

26 years agoClarify Y2K behavior when a tuple with a 2-digit date is passed to
Guido van Rossum [Tue, 25 Aug 1998 14:44:49 +0000 (14:44 +0000)]
Clarify Y2K behavior when a tuple with a 2-digit date is passed to
mktime() and such.

26 years agoNew version, with contributions from Sjoerd Mullender and Mark Hammond.
Guido van Rossum [Tue, 25 Aug 1998 14:06:55 +0000 (14:06 +0000)]
New version, with contributions from Sjoerd Mullender and Mark Hammond.
Sjoerd writes:

This version of freeze creates one file per Python module, instead of
one humongous file for all Python modules.
bkfile: new module to used to write files with backups.  No new file
is produced if the new contents is identical to the old.
New option "-x excluded-module" for modulefinder test program.
New option "-i filename" for freeze main program to include a list of
options in place of the -i option.

26 years agoThere was still something wrong. The original NOTTESTS are replaced
Guido van Rossum [Tue, 25 Aug 1998 12:29:08 +0000 (12:29 +0000)]
There was still something wrong.  The original NOTTESTS are replaced
by the new '-x' arguments, losing the previous items.  Thus,
test_support, test_b1 & test_b2 are executed (and warnings issued).
(Discovered by Vladimir Marangozov.)

26 years agoPatch by Chris Herborth (posted to comp.lang.python)to make it behave
Guido van Rossum [Mon, 24 Aug 1998 20:59:13 +0000 (20:59 +0000)]
Patch by Chris Herborth (posted to comp.lang.python)to make it behave
with tags that have - or . in their names.

26 years agoround(): Mark the second (optional) parameter as optional, since it's
Fred Drake [Mon, 24 Aug 1998 20:30:07 +0000 (20:30 +0000)]
round():  Mark the second (optional) parameter as optional, since it's
described that way.

setattr():  Clarify that the attribute doesn't need to exist to be set.

26 years agoAdded some cross-references to the end.
Fred Drake [Mon, 24 Aug 1998 18:46:14 +0000 (18:46 +0000)]
Added some cross-references to the end.

26 years agospoking --> speaking
Fred Drake [Mon, 24 Aug 1998 17:57:20 +0000 (17:57 +0000)]
spoking --> speaking

object whose method this is --> object on which the method operates

26 years agoTed Horst writes in psa-members@python.org:
Guido van Rossum [Mon, 24 Aug 1998 14:15:44 +0000 (14:15 +0000)]
Ted Horst writes in psa-members@python.org:

This is a patch that Bill Bummgarner did for 1.4 that hasn't made its
way into the distribution yet.  This is important if you want to use
the ObjC module.

26 years agoShould pass explicit arguments to findtests(). Should initialize 'nottests'.
Guido van Rossum [Mon, 24 Aug 1998 13:48:36 +0000 (13:48 +0000)]
Should pass explicit arguments to findtests().  Should initialize 'nottests'.

26 years agoMove an indented #define to column 1.
Guido van Rossum [Sun, 23 Aug 1998 22:06:59 +0000 (22:06 +0000)]
Move an indented #define to column 1.

26 years agoThere is no <ISINPUT> element, just <ISINDEX>. ;-)
Fred Drake [Fri, 21 Aug 1998 20:02:06 +0000 (20:02 +0000)]
There is no <ISINPUT> element, just <ISINDEX>.  ;-)

26 years agoThe .subn() method wasn't setting _num_regs, which is used by the .groups()
Andrew M. Kuchling [Fri, 21 Aug 1998 18:39:38 +0000 (18:39 +0000)]
The .subn() method wasn't setting _num_regs, which is used by the .groups()
method, so .groups() didn't work inside the replacement function
called by re.sub.  One-line fix: set self._num_regs inside subn().

26 years agoUpdate commentary. For most stuff, point people to the web page
Barry Warsaw [Thu, 20 Aug 1998 22:10:46 +0000 (22:10 +0000)]
Update commentary.  For most stuff, point people to the web page

26 years ago(py-shell-map): New variable contains the keymap used in *Python*
Barry Warsaw [Thu, 20 Aug 1998 21:51:27 +0000 (21:51 +0000)]
(py-shell-map): New variable contains the keymap used in *Python*
shell buffers.

(py-shell): Moved the require of comint to the top level.  Also
use-local-map py-shell-map instead of hacking on the comint-mode-map.
This eliminates breakage of other comint-mode buffers (e.g. shell).

26 years ago(py-shell): comint-output-filter-functions is already buffer-local
Barry Warsaw [Thu, 20 Aug 1998 20:00:53 +0000 (20:00 +0000)]
(py-shell): comint-output-filter-functions is already buffer-local