Commit Graph

1271 Commits

Author SHA1 Message Date
Barry Warsaw
b80667d5f3 pcre_exec(): Andrew Kuchling's patch for pcre memory leak. 1999-01-27 21:41:08 +00:00
Barry Warsaw
105906ff6e initerrno(): Nailed a not-so-tiny memory leak. The de dictionary is
put into the module dict, but is never DECREF'd in this function, so
it and all its contents leak.
1999-01-27 18:04:05 +00:00
Guido van Rossum
54ecc3d24f Patches by William Lewis for Nextstep descendants. 1999-01-27 17:53:11 +00:00
Barry Warsaw
5b97716caf expandtabs__doc__: blank line which was not terminated with \n\ caused
the SunPro C compiler to choke.  Removed this redundant line.
1999-01-26 02:15:50 +00:00
Guido van Rossum
a106568eed Add spawnv and spawnve functions for Win32 platforms. 1999-01-25 23:20:23 +00:00
Guido van Rossum
54ec2884b8 A gift from Fredrik Lundh: fast C implementation of expandtabs().
I've reformatted it, added a few comments, a test for tabsize <= 0,
and used the AS_STRING macro.
1999-01-25 22:36:24 +00:00
Guido van Rossum
aa8d16761b Make sure not to call realloc() with a NULL pointer -- call malloc()
in that case.  Tamito Kajiyama.
1999-01-25 21:43:51 +00:00
Guido van Rossum
11801859e0 Include myselect.h -- needed on some platforms. 1999-01-25 21:39:03 +00:00
Guido van Rossum
8bf6dc77d1 Rearrange the -I flags for compiling _tkinter.c so that
/usr/local/include comes before /usr/X11R1/include.  On some Linux
distributions the latter apparently contains (standard!) a bad tcl.h
or tk.h.
1999-01-25 21:38:29 +00:00
Guido van Rossum
d371ff17a1 Only do ttyname() when HAVE_TTYNAME is defined.
Also define F_OK etc. when not already defined, when doing access().
1999-01-25 16:12:23 +00:00
Guido van Rossum
ca6954a979 Should include "Python.h", not <Python.h> -- it's not a standard
header file (yet :-).
1999-01-14 19:31:42 +00:00
Guido van Rossum
11a5071ed6 Jim Ahlstrom patch: Watcom chokes on a long expression in c_asinh().
Break it up.
1999-01-14 19:11:11 +00:00
Guido van Rossum
5d00b6d459 Need extern decl. for fdatasync() in case it exists but isn't declared
anywhere (or, more likely, the declaration requires a magical
combination of _POSIX defines).
1999-01-08 21:28:05 +00:00
Guido van Rossum
21142a09f3 Added fsync() and fdatasync(). Patches by Scott Cotton. Requires
HAVE_* macros set by configure script.
1999-01-08 21:05:37 +00:00
Guido van Rossum
446ccfe1ac The doc string for strptime had the arguments reversed -- the string
comes first, the format second!  Scott Cotton discovered this.
1999-01-07 18:29:26 +00:00
Jeremy Hylton
6892aa372d fix bug in PyZlib_flush.
patch from Grzegorz Makarewicz & Rafal Smotrzyk.
1999-01-06 22:56:24 +00:00
Guido van Rossum
015f22a0fa Change the access() code to return 1 if granted, 0 if not granted.
Patch (again) by Sean Reifschneider.
1999-01-06 22:52:38 +00:00
Guido van Rossum
763737be52 Changes for long file support by Steve Clift. 1999-01-06 18:44:57 +00:00
Guido van Rossum
056bad993c Changes for long file support by Steve Clift.
He also fixes thread-related evil that caused core dumps.
1999-01-06 18:44:23 +00:00
Guido van Rossum
94f6f72c05 Changes for long file support by Steve Clift.
Also added access() and ttyname() by Sean Reifschneider.
1999-01-06 18:42:14 +00:00
Guido van Rossum
716a89c606 Patch by Charles Waldman to implement an optional nlines argument to
w.scroll().  (It then calls wscrl(win, nlines) instead of scoll(win).)
1999-01-05 17:16:46 +00:00
Guido van Rossum
0a6363d7e1 Get rid of the strptype() declaration -- on some BSD systems, it's a
conflict, and it should be declared in time.h anyway.
(Too bad if gcc -Wall won't be happy if it isn't declared...)
1999-01-03 13:00:34 +00:00
Guido van Rossum
11847896ed Chris Herborth discovered a typo in the arrow key symbols. 1998-12-23 04:45:04 +00:00
Guido van Rossum
65d5b5763c Thanks to Chris Herborth, the thread primitives now have proper Py*
names in the source code (they already had those for the linker,
through some smart macros; but the source still had the old, un-Py names).
1998-12-21 19:32:43 +00:00
Guido van Rossum
95864d35c6 Add dummy variable to avoid optimizer bug on OS/2 -- patch by Jeff Rush.
(AMK: you should probably copy this into your patch set for pcre.)
1998-12-21 18:35:49 +00:00
Jeremy Hylton
5bc4abe006 replace missing zalloc initialization (test_zlib now runs
successfully)
1998-12-21 17:15:00 +00:00
Jeremy Hylton
36252022ec remove debugging fprintf (should have checked this before previous
checkin)
1998-12-21 16:40:21 +00:00
Jeremy Hylton
a37e244536 patches from Andrew
NOTE: There is still a bug of some sort in the behavior of zlib.  In
at least one case, inflate returns Z_OK (which is typically
interpreted to mean that more output space is needed) when it has
finished inflating a buffer.  This has been reported as a bug to the
zlib maintainers; we may need to change the Python interface.
1998-12-18 22:13:11 +00:00
Fred Drake
6de7d0c338 When _PyString_Resize() reports failure, the variable referring to the
string we wanted to resize is set to NULL.  Don't Py_DECREF() those
variables!  (5 places)
1998-12-18 19:46:59 +00:00
Guido van Rossum
476e49f055 Jim Fulton writes:
This fixes a bug that can cause core dumps when doing seeks in input
StringIO objects. This has a number of other clean-ups.
1998-12-15 21:43:15 +00:00
Andrew M. Kuchling
4c07f81d60 Fixed bug reported to Gregor Hoffleit:
> mpz.mpz('\xff') should return mpz(255).  Instead it returns
> mpz(4294967295L). Looks like the constructor doesn't work with strings
> containing characters above chr(128).
Caused by using just 'char' where 'unsigned char' should have been used.
1998-12-14 19:36:14 +00:00
Guido van Rossum
21ef088265 Need to initialize self->safe_constructors early on to prevent crash
in early dealloc.  Patch by Andrew Dalke.
1998-12-11 03:20:00 +00:00
Guido van Rossum
732aa2f987 Gregor Hoffleit writes:
But IMHO, this problem really reveals an annoyance in Python's
makesetup. makesetup puts the global include directories "$(INCLUDEPY)
$(EXECINCLUDEPY)" in front of the directories defined by the module in
Setup. Therefore global (potentially older) header files are preferred
over the ones set by the module, which makes it hard to compile new
versions of modules when the old versions are installed. AFAIK, the
other way around is common practice for most other software.

This patch to makesetup would be an potential fix for this problem,
though I don't know if it breaks anything else.
1998-12-11 01:10:29 +00:00
Guido van Rossum
46e9705eca Remove prototypes for PyOS_strto[u]l -- Chris Herborth. 1998-12-10 16:57:44 +00:00
Guido van Rossum
5de54208a5 Needed to add DL_EXPORT to (redundant?) extern decl of module init function. 1998-12-10 16:49:28 +00:00
Guido van Rossum
e94e3fbb72 Make VC++ 5.0 compiler happy. 1998-12-08 17:37:19 +00:00
Guido van Rossum
b057dd8efa Correctly document atan2. 1998-12-08 16:27:10 +00:00
Guido van Rossum
9c1201fe39 Py_Main() must be DL_EXPORT too. 1998-12-07 14:28:47 +00:00
Guido van Rossum
c6e22902ad Added Doc strings -- by Chris Petrilli. 1998-12-04 19:26:43 +00:00
Guido van Rossum
3886bb6997 Add DL_EXPORT() to all modules that could possibly be used
on BeOS or Windows.
1998-12-04 18:50:17 +00:00
Guido van Rossum
50f385c197 Fix two small bugs; add DL_EXPORT() to initcPickle decl. 1998-12-04 18:48:44 +00:00
Guido van Rossum
3b5330ef2d Bernard Herzog pointed out that rl_parse_and_bind modifies its
argument string (bad function!), so we make a temporary copy.
1998-12-04 15:34:39 +00:00
Guido van Rossum
053b8dfcde New version from Jim Fulton:
- New copyright. (Open source)

  - Added new protocol for binary string pickles that
    takes out unneeded puts:

      p=Pickler()
      p.dump(x)
      p.dump(y)
      thePickle=p.getvalue()

    This has little or no impact on pickling time, but
    often reduces unpickling time and pickle size, sometimes
    significantly.

  - Changed unpickler to use internal data structure instead
    of list to reduce unpickling times by about a third.

  - Many cleanups to get rid of obfuscated error handling
    involving 'goto finally' and status variables.

  - Extensive reGuidofication. (formatting :)

  - Fixed binary floating-point pickling bug. 0.0 was not
    pickled correctly.

  - Now use binary floating point format when saving
    floats in binary mode.

  - Fixed some error message spelling error.
1998-11-25 16:18:00 +00:00
Guido van Rossum
7d9b4133eb New version from Jim Fulton:
- New copyright. (Open source)

  - Fixed problem in seek method.  The seek method should (and now does)
    fill with nulls when seeking past the end of the "file".
1998-11-25 16:17:32 +00:00
Guido van Rossum
185ead6f37 Doc strings by Chris Petrilli. 1998-11-23 15:32:55 +00:00
Guido van Rossum
e9bc62d993 RajGopal Srinivasan noted that the latest code doesn't work when
running in a non-threaded environment.  He added some #ifdefs that fix
this.
1998-11-17 03:45:24 +00:00
Barry Warsaw
b0d1b068d9 audio(7I) suggests that applications do the following to get the
device and control pseudo-device:

- first look for the device filename in the environment variable
  AUDIODEV.

- if not found, use /dev/audio

- calculate the control device by tacking "ctl" onto the base device
  name.

We now do this.  Also, if the open fails, we call
PyErr_SetFromErrnoWithFilename() to give a more informative error
message.

Added a fileno() method to the audio object returned from open().
This returns the file descriptor which can be used by applications to
set up SIGPOLL notification, as per the manpage.
1998-10-31 22:52:54 +00:00
Guido van Rossum
3f200d76df Add note about compiling FORMS with -Dclear=__GLclear.
(The GLHACK variable isn't really needed since the change to
glmodule.c, but I'm too busy to remove it.  It's harmless.)
1998-10-21 17:03:36 +00:00
Guido van Rossum
89733a8bd6 Check in the changed version after running the stubber again -- this
solves the conflict with curses over the 'clear' entry point much
nicer.  (Jack had checked in the changes to cstubs eons ago, but I
never regenrated glmodule.c :-( )
1998-10-21 16:10:40 +00:00
Guido van Rossum
605b93dee2 On a recommendation from Sjoerd Mullender, add -Dclear=__GLclear to
the compilation flags for the gl, fl and fm modules.  This avoids a
name conflict with the curses module (both gl and curses have an entry
point called 'clear').
1998-10-21 15:08:40 +00:00