| msg122665 - (view) |
Author: Gerrit Holl (Gerrit.Holl) * |
Date: 2010-11-28 15:47 |
| When the time passed to time.ctime is large, it adds a newline: >>> import time >>> time.ctime(2<<36) 'Wed Apr 8 17:04:32 6325' >>> time.ctime(2<<37) 'Wed Jul 14 08:09:04 10680\n' |
|
|
| msg122670 - (view) |
Author: Eric V. Smith (eric.smith) *  |
Date: 2010-11-28 16:09 |
| There's an error in time_ctime where it expects the length of the string to be fixed: if (p[24] == '\n') p[24] = '\0'; It doesn't count on the year having 5 digits. It should probably say (untested): l = len(p); if (l > 0 && p[l-1] == '\n') p[l-1] = '\0'; I'll whip up a patch and some tests. I'm not sure if the tests will work on platforms other than Linux. I'll see what I can find out about ctime and large values. |
|
|
| msg122671 - (view) |
Author: Eric V. Smith (eric.smith) * |
Date: 2010-11-28 16:10 |
| That should be strlen(), of course. |
|
|
| msg122673 - (view) |
Author: Eric V. Smith (eric.smith) * |
Date: 2010-11-28 16:12 |
| What platform are you running this on? My Fedora 32 bit system won't support a time_t that large. |
|
|
| msg122674 - (view) |
Author: Gerrit Holl (Gerrit.Holl) * |
Date: 2010-11-28 16:14 |
| I'm on a 64bit system (kubuntu lucid lynx) $ uname -a Linux sjisjka 2.6.32-25-generic #45-Ubuntu SMP Sat Oct 16 19:52:42 UTC 2010 x86_64 GNU/Linux |
|
|
| msg122675 - (view) |
Author: Eric V. Smith (eric.smith) * |
Date: 2010-11-28 16:17 |
| Can you try this diff and see if it solves the problem: Index: Modules/timemodule.c =================================================================== --- Modules/timemodule.c (revision 86848) +++ Modules/timemodule.c (working copy) @@ -639,6 +639,7 @@ PyObject *ot = NULL; time_t tt; char *p; + Py_ssize_t len; if (!PyArg_UnpackTuple(args, "ctime", 0, 1, &ot)) return NULL; @@ -657,8 +658,10 @@ PyErr_SetString(PyExc_ValueError, "unconvertible time"); return NULL; } - if (p[24] == '\n') - p[24] = '\0'; + len = strlen(p); + if (len > 0 && p[len-1] == '\n') + /* Remove the trailing newline, if present. */ + p[len-1] = '\0'; return PyUnicode_FromString(p); } |
|
|
| msg122678 - (view) |
Author: Gerrit Holl (Gerrit.Holl) * |
Date: 2010-11-28 16:36 |
| Yes, this solves the issue: $ ./python Python 3.1.3 (r313:86834, Nov 28 2010, 17:34:23) [GCC 4.4.3] on linux2 Type "help", "copyright", "credits" or "license" for more information. >>> import time >>> time.ctime(2<<40) 'Fri Apr 10 03:12:32 71654' |
|
|
| msg122761 - (view) |
Author: Alexander Belopolsky (belopolsky) * |
Date: 2010-11-29 03:18 |
| This looks like a bug in the underlying platform. POSIX requires [1] that the output of ctime() fits in a 26-character buffer. Note that a change has been recently made to time.asctime() to reject year > 9999. See r85137 and . I think for consistency, we should reject huge timestamps in time ctime() as well. [1] http://www.opengroup.org/onlinepubs/009695399/functions/ctime.html |
|
|
| msg125223 - (view) |
Author: Georg Brandl (georg.brandl) * |
Date: 2011-01-03 20:02 |
| This has now been superseded by the changes made for issue #8013. |
|
|