]> granicus.if.org Git - python/log
python
24 years agoMinor nit to make this work a little better with the PDF generation fix.
Fred Drake [Tue, 5 Sep 2000 15:11:56 +0000 (15:11 +0000)]
Minor nit to make this work a little better with the PDF generation fix.
This does not change the HTML or PostScript output.

24 years agoDon't emit 8 lines of (c) information at startup -- use the same text
Guido van Rossum [Tue, 5 Sep 2000 13:51:14 +0000 (13:51 +0000)]
Don't emit 8 lines of (c) information at startup -- use the same text
as main.c.

24 years agoMinor markup revisions for consistency with the rest of the documentation.
Fred Drake [Tue, 5 Sep 2000 13:50:21 +0000 (13:50 +0000)]
Minor markup revisions for consistency with the rest of the documentation.

24 years agoAdded entry for mmap module.
Fred Drake [Tue, 5 Sep 2000 13:49:30 +0000 (13:49 +0000)]
Added entry for mmap module.

24 years agoBarry Warsaw: Fixed -+ operator. Added some new sections. Leave some
Guido van Rossum [Tue, 5 Sep 2000 12:42:46 +0000 (12:42 +0000)]
Barry Warsaw: Fixed -+ operator.  Added some new sections.  Leave some
XXX notes for now.

I could use help here!!!!  Please mail me patches ASAP.  We may have
to put some of this off to 2.0final, but it's best to have it in shape
now...

24 years agoNow that I've got real docs, redirect _DOC_ out of the Python 1.6 tree and
Tim Peters [Tue, 5 Sep 2000 06:23:16 +0000 (06:23 +0000)]
Now that I've got real docs, redirect _DOC_ out of the Python 1.6 tree and
to the new docs.
Added a description to the Tcl/Tk file copies.

24 years agoThe usual.
Guido van Rossum [Tue, 5 Sep 2000 04:49:50 +0000 (04:49 +0000)]
The usual.

24 years agoSh*t! Should use $(srcdir)/LICENSE of course.
Guido van Rossum [Tue, 5 Sep 2000 04:46:32 +0000 (04:46 +0000)]
Sh*t!  Should use $(srcdir)/LICENSE of course.

24 years agoInstall LICENSE as LICENSE.txt in the <prefix>/lib/python<version>
Guido van Rossum [Tue, 5 Sep 2000 04:44:48 +0000 (04:44 +0000)]
Install LICENSE as LICENSE.txt in the <prefix>/lib/python<version>
directory.

24 years agoInstead of printing all 9 lines of copyright info, print
Guido van Rossum [Tue, 5 Sep 2000 04:41:18 +0000 (04:41 +0000)]
Instead of printing all 9 lines of copyright info, print

    "Type \"copyright\", \"credits\" or \"license\" for more information."

24 years agoThe GCC version is loooooooooong; put it on a new line.
Guido van Rossum [Tue, 5 Sep 2000 04:40:39 +0000 (04:40 +0000)]
The GCC version is loooooooooong; put it on a new line.

24 years agoAdd new builtin commands "copyright", "license", "credits" which
Guido van Rossum [Tue, 5 Sep 2000 04:39:55 +0000 (04:39 +0000)]
Add new builtin commands "copyright", "license", "credits" which
display the information you would expect them to display.

24 years agoAdded the 2.0b1 news.
Guido van Rossum [Tue, 5 Sep 2000 04:38:34 +0000 (04:38 +0000)]
Added the 2.0b1 news.

24 years agoUpdate license according to the LICENSE file in the source tree.
Fred Drake [Tue, 5 Sep 2000 04:32:26 +0000 (04:32 +0000)]
Update license according to the LICENSE file in the source tree.

24 years agoSet the date.
Fred Drake [Tue, 5 Sep 2000 03:13:38 +0000 (03:13 +0000)]
Set the date.

24 years agoSince it looks like the dual license clause may be neither necessary
Guido van Rossum [Tue, 5 Sep 2000 03:05:07 +0000 (03:05 +0000)]
Since it looks like the dual license clause may be neither necessary
nor sufficient to make Python 2.0 compatible with the GPL, we won't
bother with it now.

In other words, we're still where we were weeks ago -- CNRI believes
that its license is GPL-compatible, Stallman says it's not.  I'm
trying to arrange a meeting between their lawyers so they can work it
out.  Whether dual licensing is the solution is open at this point.
If it is the (only!) solution, we'll add that to the BeOpen license
for 2.0 final.

24 years agoFleshed out the section on the setup config file, setup.cfg.
Greg Ward [Mon, 4 Sep 2000 20:07:15 +0000 (20:07 +0000)]
Fleshed out the section on the setup config file, setup.cfg.
Added a few clarifying footnotes and cross-references.
Various minor tweaks.

24 years agoVarious corrections pointed out by Detlef Lannert
Andrew M. Kuchling [Mon, 4 Sep 2000 17:59:27 +0000 (17:59 +0000)]
Various corrections pointed out by Detlef Lannert

24 years agoAdded release dates (in some cases detemined by looking at the mtime
Guido van Rossum [Mon, 4 Sep 2000 17:24:24 +0000 (17:24 +0000)]
Added release dates (in some cases detemined by looking at the mtime
of distribution files in ftp://ftp.python.org/pub/python/src/).

24 years ago1.6 is history.
Guido van Rossum [Mon, 4 Sep 2000 16:26:03 +0000 (16:26 +0000)]
1.6 is history.

24 years agoFor this server to work on Windows, directories should use "/" as the
Guido van Rossum [Mon, 4 Sep 2000 15:55:31 +0000 (15:55 +0000)]
For this server to work on Windows, directories should use "/" as the
separator in the href, not os.sep.

Added a <title> tag to directory listings.

Bumped version to 0.5.

24 years agoTim Peters: "Audun S. Runde mailto:audun@mindspring.com wins a
Guido van Rossum [Mon, 4 Sep 2000 12:11:50 +0000 (12:11 +0000)]
Tim Peters: "Audun S. Runde mailto:audun@mindspring.com wins a
Fabulous Prize for being our first Windows ME tester!  Also our only,
and I think he should get another prize just for that."

24 years agotest_mmap wrote null bytes into its expected-output file; this caused me to
Tim Peters [Mon, 4 Sep 2000 07:34:06 +0000 (07:34 +0000)]
test_mmap wrote null bytes into its expected-output file; this caused me to
waste an hour tracking down an illusion; repaired it; writing/reading non-
printable characters (except \t\r\n) into/outof text-mode files ain't
defined x-platform, and at least some Windows text editors do surprising
things in their presence.
Also added a by-hand "build humber" to the Windows build, in an approximation
of Python's inexplicable BUILD-number Unix scheme.  I'll try to remember to
increment it each time I make a Windows installer available.  It's starting
at 2, cuz I've put 2 installers out so far (both with BUILD #0).

24 years agoFixes bug in --with-libdb. If --with-libdb was not specified (default is to
Skip Montanaro [Mon, 4 Sep 2000 04:06:05 +0000 (04:06 +0000)]
Fixes bug in --with-libdb.  If --with-libdb was not specified (default is to
enable it), but db.h was not found, the WITH_LIBDB macros was still being
defined, resulting in compilation errors.  Also added a short explain when
bsddb support wasn't enabled (because db.h wasn't found) when the user
explicitly used --with-libdb on the configure command line.

24 years agoCorrect docstring about return value when group didn't participate in match
Andrew M. Kuchling [Mon, 4 Sep 2000 03:19:48 +0000 (03:19 +0000)]
Correct docstring about return value when group didn't participate in match
    (pointed out by /F)

24 years agoAll right. More uniformity, and extra blank lines.
Guido van Rossum [Mon, 4 Sep 2000 01:27:04 +0000 (01:27 +0000)]
All right.  More uniformity, and extra blank lines.

24 years agoUse periods, not semicolons between Copyright and All Rights Reserved.
Guido van Rossum [Mon, 4 Sep 2000 01:22:12 +0000 (01:22 +0000)]
Use periods, not semicolons between Copyright and All Rights Reserved.

24 years agoChanges in license names by BobW.
Guido van Rossum [Mon, 4 Sep 2000 00:58:48 +0000 (00:58 +0000)]
Changes in license names by BobW.

24 years agoFix the char* vs. const char* mismatch for the argument of aix_loaderror()
Vladimir Marangozov [Mon, 4 Sep 2000 00:54:56 +0000 (00:54 +0000)]
Fix the char* vs. const char* mismatch for the argument of aix_loaderror()

24 years agoMove down the INT_MAX logic, because HAVE_LIMITS_H was always undefined
Vladimir Marangozov [Sun, 3 Sep 2000 23:47:08 +0000 (23:47 +0000)]
Move down the INT_MAX logic, because HAVE_LIMITS_H was always undefined
and this breaks the AIX build with an INT_MAX redefinition error.
"config.h" is included in pgenheaders.h, so moving this down fixes the
problem.

24 years agoHack the Windows code to use os.popen().
Guido van Rossum [Sun, 3 Sep 2000 17:12:50 +0000 (17:12 +0000)]
Hack the Windows code to use os.popen().
The returned file is assigned to an instance variable;
otherwise the implied close hangs for a long time.

24 years agoProperly name and number the BEOPEN OPEN SOURCE PYTHON LICENSE
Guido van Rossum [Sun, 3 Sep 2000 13:21:38 +0000 (13:21 +0000)]
Properly name and number the BEOPEN OPEN SOURCE PYTHON LICENSE
AGREEMENT VERSION 1.

trade name -> trade names.

Note: depending on community feedback, we may end up taking the dual
licensing clause out for 2.0b1, and put it back into 2.0final, if
there's no other solution for assuring GPL compatibility by then.

See my message to python-dev and license-py20.

24 years agochanged \x to consume exactly two hex digits, also for unicode
Fredrik Lundh [Sun, 3 Sep 2000 11:29:49 +0000 (11:29 +0000)]
changed \x to consume exactly two hex digits, also for unicode
strings.  closes PEP-223.

also added \U escape (eight hex digits).

24 years agoupdated SRE test suite (fixes PEP223 problem, shows syntax errors)
Fredrik Lundh [Sun, 3 Sep 2000 10:43:16 +0000 (10:43 +0000)]
updated SRE test suite (fixes PEP223 problem, shows syntax errors)

24 years agoRepair failing test_sre.py.
Tim Peters [Sun, 3 Sep 2000 08:15:19 +0000 (08:15 +0000)]
Repair failing test_sre.py.
This was a funny one!  The test very subtly relied on 1.5.2's
behavior of treating "\x%" as "\x%", i.e. ignoring that was an
\x escape that didn't make sense.  But /F implemented PEP 223,
which causes 2.0 to raise an exception on the bad escape.
Fixed by merely making the 3 such strings of this kind into
raw strings.

24 years agoRemove a comma from CWI's copyright notice.
Guido van Rossum [Sun, 3 Sep 2000 04:47:47 +0000 (04:47 +0000)]
Remove a comma from CWI's copyright notice.

24 years agoChange the copyright notice according to CNRI's wishes, with
Guido van Rossum [Sun, 3 Sep 2000 03:35:50 +0000 (03:35 +0000)]
Change the copyright notice according to CNRI's wishes, with
BeOpen.com added to the front.

(Even if maybe we won't print this long banner at startup, the string
must still be defined for sys.copyright.)

24 years agoVarious edits. Most importantly, added dual licensing. Also some
Guido van Rossum [Sun, 3 Sep 2000 03:13:44 +0000 (03:13 +0000)]
Various edits.  Most importantly, added dual licensing.  Also some
changes suggested by BobW.

24 years agochanged \x to consume exactly two hex digits. implements PEP-223
Fredrik Lundh [Sat, 2 Sep 2000 20:11:27 +0000 (20:11 +0000)]
changed \x to consume exactly two hex digits.  implements PEP-223
for 8-bit strings.

24 years agoreturn -1 for undefined groups (as implemented in 1.5.2) instead of
Fredrik Lundh [Sat, 2 Sep 2000 16:36:57 +0000 (16:36 +0000)]
return -1 for undefined groups (as implemented in 1.5.2) instead of
None (as documented) from start/end/span.  closes bug #113254

24 years agoPatch by Martin von Löwis to give him his umlaut, to remove Christian
Guido van Rossum [Sat, 2 Sep 2000 15:01:50 +0000 (15:01 +0000)]
Patch by Martin von Löwis to give him his umlaut, to remove Christian
Tismer's clone, and to list Hajime Saitou's real name.

Added a note that the file uses Latin-1 (as distributed).

24 years ago-- tightened up parsing of octal numbers
Fredrik Lundh [Sat, 2 Sep 2000 11:03:34 +0000 (11:03 +0000)]
-- tightened up parsing of octal numbers

-- improved the SRE test harness: don't use asserts, test a few more
   things (including more boundary conditions)

24 years agoPyInterpreterState_New is not thread-safe, and the recent fix to _PyPclose
Tim Peters [Sat, 2 Sep 2000 09:16:15 +0000 (09:16 +0000)]
PyInterpreterState_New is not thread-safe, and the recent fix to _PyPclose
can cause it to get called by multiple threads simultaneously.

Ditto for PyInterpreterState_Delete.

Of the former, the docs say "The interpreter lock need not be held, but may
be held if it is necessary to serialize calls to this function".  This
kinda implies it both is and isn't thread-safe.

Of the latter, the docs merely say "The interpreter lock need not be
held.", and the clause about serializing is absent.

I expect it was *believed* these are both thread-safe, and the bit about
serializing via the global lock was meant as a permission rather than a
caution.

I also expect we've never seen a problem here because the Python core
(prior to the _PyPclose fix) only calls these functions once per run.
The Py_NewInterpreter subsystem exposed by the C API (but not used by
Python itself) also calls them, but that subsystem appears to be very
rarely used.

Whatever, they're both thread-safe now.

24 years agoCosmetic cleanup by Vladimir.
Thomas Wouters [Sat, 2 Sep 2000 08:34:40 +0000 (08:34 +0000)]
Cosmetic cleanup by Vladimir.

24 years agoSourceForge patch 101396, by an anonymous friend.
Tim Peters [Sat, 2 Sep 2000 07:44:32 +0000 (07:44 +0000)]
SourceForge patch 101396, by an anonymous friend.
"sre_parse.py missing '7' in DIGITS"

24 years agoREMOVED all CWI, CNRI and BeOpen copyright markings.
Guido van Rossum [Fri, 1 Sep 2000 23:29:29 +0000 (23:29 +0000)]
REMOVED all CWI, CNRI and BeOpen copyright markings.
This should match the situation in the 1.6b1 tree.

24 years agoRewritten some pieces of PyNumber_InPlaceAdd() for clarity.
Guido van Rossum [Fri, 1 Sep 2000 23:27:32 +0000 (23:27 +0000)]
Rewritten some pieces of PyNumber_InPlaceAdd() for clarity.

24 years agoUpdated to include all files here.
Guido van Rossum [Fri, 1 Sep 2000 23:22:12 +0000 (23:22 +0000)]
Updated to include all files here.

24 years agoDelete too.
Guido van Rossum [Fri, 1 Sep 2000 23:21:04 +0000 (23:21 +0000)]
Delete too.

24 years agoDeleted. Long obsolete.
Guido van Rossum [Fri, 1 Sep 2000 23:18:22 +0000 (23:18 +0000)]
Deleted.  Long obsolete.

24 years agoDeleted. We no longer require disclaimers on contributions.
Guido van Rossum [Fri, 1 Sep 2000 23:16:41 +0000 (23:16 +0000)]
Deleted.  We no longer require disclaimers on contributions.

24 years agoDeleted; this is ancient info.
Guido van Rossum [Fri, 1 Sep 2000 23:15:38 +0000 (23:15 +0000)]
Deleted; this is ancient info.

24 years agoDeleted. This is obsoleted by Tools/scripts/fixnotice.py.
Guido van Rossum [Fri, 1 Sep 2000 23:12:13 +0000 (23:12 +0000)]
Deleted.  This is obsoleted by Tools/scripts/fixnotice.py.

24 years agoReady to go.
Guido van Rossum [Fri, 1 Sep 2000 22:50:02 +0000 (22:50 +0000)]
Ready to go.

24 years agoRemove the 1.5.2 news. 2.0 news is still to be done.
Guido van Rossum [Fri, 1 Sep 2000 22:34:33 +0000 (22:34 +0000)]
Remove the 1.5.2 news.  2.0 news is still to be done.

24 years agoMoved NEWS from 1.5.2 here.
Guido van Rossum [Fri, 1 Sep 2000 22:26:44 +0000 (22:26 +0000)]
Moved NEWS from 1.5.2 here.

24 years agoDeleted. This information is now in ../LICENSE.
Guido van Rossum [Fri, 1 Sep 2000 22:17:32 +0000 (22:17 +0000)]
Deleted.  This information is now in ../LICENSE.

24 years agoThe second argument to start_new_thread() is not a tuple, as pointed
Guido van Rossum [Fri, 1 Sep 2000 20:47:58 +0000 (20:47 +0000)]
The second argument to start_new_thread() is not a tuple, as pointed
out by Curtis Jensen on c.l.py and forwarded by Fredrik Lundh.

24 years agopatch by Neil Schemenauer to improve (fix?) line number generation
Jeremy Hylton [Fri, 1 Sep 2000 20:47:37 +0000 (20:47 +0000)]
patch by Neil Schemenauer to improve (fix?) line number generation

24 years agoThe "more" cmd varies across Windows flavors, sometimes adding stray
Tim Peters [Fri, 1 Sep 2000 20:38:55 +0000 (20:38 +0000)]
The "more" cmd varies across Windows flavors, sometimes adding stray
newlines at the start or end.  Fiddle test_popen2 and popen2._test() to
tolerate this.  Also change all "assert"s in these tests to raise
explicit exceptions, so that python -O doesn't render them useless.
Also, in case of error, make the msg display the reprs of what we
wrote and what we read, so we can tell exactly why it's failing.

24 years agoClean up a bit. Drop -X option, update date, remove documentation
Guido van Rossum [Fri, 1 Sep 2000 20:36:34 +0000 (20:36 +0000)]
Clean up a bit.  Drop -X option, update date, remove documentation
URLs, add various useful URLs.  Update address and email.  Drop PSA
and copyright.  Add license info.

24 years agoUpdate magic number.
Jeremy Hylton [Fri, 1 Sep 2000 20:33:26 +0000 (20:33 +0000)]
Update magic number.

Fix import support to work with import as variant of Python 2.0.  The
grammar for import changed, requiring changes in transformer and code
generator, even to handle compilation of imports with as.

24 years agoInstall LICENSE.txt and README.txt.
Tim Peters [Fri, 1 Sep 2000 19:59:11 +0000 (19:59 +0000)]
Install LICENSE.txt and README.txt.
In the Welcome dialog:
    Reworded reference to non-existent "Exit Setup" button.
    Removed useless "Back" button.
    Changed "push" to "click".

24 years agoTentative license. Could still change for the 2.0b1 release and will
Guido van Rossum [Fri, 1 Sep 2000 19:51:14 +0000 (19:51 +0000)]
Tentative license.  Could still change for the 2.0b1 release and will
definitely change for the 2.0 final release.

24 years agoGuido pointed out that the "non-admin install" blurb got displayed
Tim Peters [Fri, 1 Sep 2000 19:30:26 +0000 (19:30 +0000)]
Guido pointed out that the "non-admin install" blurb got displayed
very late in the process when running on Windows 2000 without admim
privs.  Rearranged so that the admin check is done at the start
instead.  Added words to the end of the blurb to make it very clear
how to abort the install (wasn't obvious to me that "Cancel" was
the right thing to click).

24 years agoAdding new files, removing some.
Guido van Rossum [Fri, 1 Sep 2000 19:27:34 +0000 (19:27 +0000)]
Adding new files, removing some.

24 years agoRepaired damaged comments, and extra spaces in fatal error msgs we'd better
Tim Peters [Fri, 1 Sep 2000 19:26:36 +0000 (19:26 +0000)]
Repaired damaged comments, and extra spaces in fatal error msgs we'd better
not ever see!

24 years agoThe usual
Guido van Rossum [Fri, 1 Sep 2000 19:25:51 +0000 (19:25 +0000)]
The usual

24 years agoCorrect configure.in version.
Guido van Rossum [Fri, 1 Sep 2000 19:22:55 +0000 (19:22 +0000)]
Correct configure.in version.

24 years agoMove the Py_DECREF(x) after the error: label so that we don't leak x
Guido van Rossum [Fri, 1 Sep 2000 15:35:12 +0000 (15:35 +0000)]
Move the Py_DECREF(x) after the error: label so that we don't leak x
when PyDict_SetItemString() fails.

24 years agoRob Hooft, Moshe Zadka: converted to 4 space indents and re instead of regex.
Guido van Rossum [Fri, 1 Sep 2000 13:41:37 +0000 (13:41 +0000)]
Rob Hooft, Moshe Zadka: converted to 4 space indents and re instead of regex.

24 years agoBack off to HKCU (instead of HKLM) if user doesn't have "NT adminstrator
Tim Peters [Fri, 1 Sep 2000 11:45:02 +0000 (11:45 +0000)]
Back off to HKCU (instead of HKLM) if user doesn't have "NT adminstrator
privileges".  Untested except on Win98SE (where Wise writes to HKLM).

24 years agoCosmetics on Py_Get/SetRecursionLimit (for the style guide)
Vladimir Marangozov [Fri, 1 Sep 2000 11:07:19 +0000 (11:07 +0000)]
Cosmetics on Py_Get/SetRecursionLimit (for the style guide)

24 years agoUpdate documentation for PyErr_Format, because the function has changed.
Moshe Zadka [Fri, 1 Sep 2000 09:47:20 +0000 (09:47 +0000)]
Update documentation for PyErr_Format, because the function has changed.
Connected to patch 100895

24 years agoDo the absolute minimal amount of modifications to eradicate
Barry Warsaw [Fri, 1 Sep 2000 09:01:32 +0000 (09:01 +0000)]
Do the absolute minimal amount of modifications to eradicate
Py_FatalError() from module initialization functions.  The importing
mechanism already checks for PyErr_Occurred() after module importation
and it Does The Right Thing.

Unfortunately, the following either were not compiled or tested by the
regression suite, due to issues with my development platform:

almodule.c
cdmodule.c
mpzmodule.c
puremodule.c
timingmodule.c

24 years agoTool to generate binary GNU .mo file from .po template files. Written
Barry Warsaw [Fri, 1 Sep 2000 08:10:08 +0000 (08:10 +0000)]
Tool to generate binary GNU .mo file from .po template files.  Written
by Martin v. Loewis, proofed by Barry Warsaw for coding standards,
typos, and to make command line options compatible with GNU msgfmt
where they overlap.

Closes patch #101295.

24 years agoAdded installation of w9xpopen.exe.
Tim Peters [Fri, 1 Sep 2000 07:54:04 +0000 (07:54 +0000)]
Added installation of w9xpopen.exe.
Removed installation of Lib/plat-win/*.py, because it no longer exists!

24 years agoFix grouping, again. This time properly :-) Sorry, guys.
Thomas Wouters [Fri, 1 Sep 2000 07:53:25 +0000 (07:53 +0000)]
Fix grouping, again. This time properly :-) Sorry, guys.

24 years agoApplying patch #100994 to allow JPython to use more of the standard
Barry Warsaw [Fri, 1 Sep 2000 06:53:52 +0000 (06:53 +0000)]
Applying patch #100994 to allow JPython to use more of the standard
Python test suite.  Specifically,

- import time instead of strop in test_b1

- test for ClassType of exceptions using isinstance instead of
  equality in test_exceptions

- remove __builtins__ from dir() output in test_pkg

test_pkg output needs to be regenerated.

24 years agoFix test_popen2 on Windows, recently broken by changes to the dict(!)
Tim Peters [Fri, 1 Sep 2000 06:51:24 +0000 (06:51 +0000)]
Fix test_popen2 on Windows, recently broken by changes to the dict(!)
implementation.  You don't want to know.  I've asked Guido to give this
a critical review (we agreed on the approach, but the implementation
proved more ... interesting ... than anticipated).  This will almost
certainly be the highlight of Mark Hammond's day <wink>.

24 years agoApplied patch #101350, closing it.
Barry Warsaw [Fri, 1 Sep 2000 06:40:07 +0000 (06:40 +0000)]
Applied patch #101350, closing it.

24 years agoDocument the new optional argument "rest" on the transfercmd(),
Barry Warsaw [Fri, 1 Sep 2000 06:32:32 +0000 (06:32 +0000)]
Document the new optional argument "rest" on the transfercmd(),
ntransfercmd(), and retrbinary() commands.  This closes SF patch
#101187.

24 years agoAdded support for RFC 959's REST command (restart), closing SF patch
Barry Warsaw [Fri, 1 Sep 2000 06:09:23 +0000 (06:09 +0000)]
Added support for RFC 959's REST command (restart), closing SF patch
#101187, which some modifications.  Specifically,

ntransfercmd(), transfercmd(), and retrbinary() all grow an optional
`rest' argument, which if not None, is used as the argument to an FTP
REST comman dbefore the socket is returned.  Differences from the SF
patch:

- always compare against None with `is' or `is not' instead of == or !=

- no parens around conditional

- RFC 959 defines the argument to REST is a string containing any
  ASCII characters in the range [33..126].  Therefore, we use the %s
  format character instead of %f or %d as suggested in the patch's
  comments.  Note that we do /not/ sanity checkthe contents of the
  rest argument (but we'll document this in the library reference
  manual).

24 years agoDocument PyImport_AppendInittab(), PyImport_ExtendInittab(), and
Fred Drake [Fri, 1 Sep 2000 05:30:00 +0000 (05:30 +0000)]
Document PyImport_AppendInittab(), PyImport_ExtendInittab(), and
struct _inittab.

This closes SourceForge bug #111499.

24 years agoset the default threshold much higher
Jeremy Hylton [Fri, 1 Sep 2000 04:01:55 +0000 (04:01 +0000)]
set the default threshold much higher
we don't need to run gc frequently

24 years agocode part of patch #100895 by Fredrik Lundh
Jeremy Hylton [Fri, 1 Sep 2000 03:49:47 +0000 (03:49 +0000)]
code part of patch #100895 by Fredrik Lundh

PyErr_Format computes size of buffer needed rather than relying on
static buffer.

24 years agoDon't call Py_FatalError in module initialization
Jeremy Hylton [Fri, 1 Sep 2000 03:46:16 +0000 (03:46 +0000)]
Don't call Py_FatalError in module initialization
(leaving the rest of the modules for Barry)

24 years agoRevert removal of void from function definition. Guido sez I can take it
Tim Peters [Fri, 1 Sep 2000 03:34:26 +0000 (03:34 +0000)]
Revert removal of void from function definition.  Guido sez I can take it
out again after we complete switching to C++ <wink>.  Thanks to Greg Stein
for hitting me.

24 years agoNow that StreamRequestHandler defaults rfile to buffered, make it
Guido van Rossum [Fri, 1 Sep 2000 03:27:34 +0000 (03:27 +0000)]
Now that StreamRequestHandler defaults rfile to buffered, make it
unbuffered (by setting the class variable rbufsize to 0), because we
(may) need to pass the file descriptor to the subprocess running the
CGI script positioned after the headers.

24 years agoIn class StreamRequestHandler, make the default buffering for rfile
Guido van Rossum [Fri, 1 Sep 2000 03:25:14 +0000 (03:25 +0000)]
In class StreamRequestHandler, make the default buffering for rfile
and wfile class variables (that the instance can also override).
Change the default for rfile to buffered, because that seems to make a
big difference in performance on some platforms.

An anti-patch is needed to revert the effect in CGIHTTPServer.py which
I'll check in momentarily.

24 years agorefactor __del__ exception handler into PyErr_WriteUnraisable
Jeremy Hylton [Fri, 1 Sep 2000 02:47:25 +0000 (02:47 +0000)]
refactor __del__ exception handler into PyErr_WriteUnraisable
add sanity check to gc: if an exception occurs during GC, call
PyErr_WriteUnraisable and then call Py_FatalEror.

24 years agoAdd three prototypes for functions in history.h to shut up gcc -Wall.
Guido van Rossum [Fri, 1 Sep 2000 02:43:38 +0000 (02:43 +0000)]
Add three prototypes for functions in history.h to shut up gcc -Wall.

24 years agoMoodules -> Modules.
Guido van Rossum [Fri, 1 Sep 2000 02:40:11 +0000 (02:40 +0000)]
Moodules -> Modules.

24 years agoAdd parens suggested by gcc -Wall.
Guido van Rossum [Fri, 1 Sep 2000 02:39:00 +0000 (02:39 +0000)]
Add parens suggested by gcc -Wall.

24 years agoRepaired comment.
Tim Peters [Fri, 1 Sep 2000 02:20:20 +0000 (02:20 +0000)]
Repaired comment.

24 years agoSet the recursion limit to 1000 -- 2500 was not enough, let's be
Guido van Rossum [Fri, 1 Sep 2000 01:52:08 +0000 (01:52 +0000)]
Set the recursion limit to 1000 -- 2500 was not enough, let's be
conservative.

24 years agoRene Liebscher/Thomas Heller:
Greg Ward [Fri, 1 Sep 2000 01:44:45 +0000 (01:44 +0000)]
Rene Liebscher/Thomas Heller:
* ensure the "dist" directory exists
* raise exception if using for modules containing compiled extensions
  on a non-win32 platform.
* don't create an .ini file anymore (it was just for debugging)

24 years agoRene Liebscher:
Greg Ward [Fri, 1 Sep 2000 01:28:33 +0000 (01:28 +0000)]
Rene Liebscher:
* reverse library names from bcpp_library to library_bcpp
* move some code to the right places, to put the def-files
  in the right directories again

24 years agoRene Liebscher: comment fixes.
Greg Ward [Fri, 1 Sep 2000 01:24:31 +0000 (01:24 +0000)]
Rene Liebscher: comment fixes.

24 years agoRene Liebscher: hack '_init_posix()' to handle the BeOS linker script.
Greg Ward [Fri, 1 Sep 2000 01:23:26 +0000 (01:23 +0000)]
Rene Liebscher: hack '_init_posix()' to handle the BeOS linker script.
(With a worry-wart comment added by me about where we *should* add the
Python library to the link.)