[issue4315] On some Python builds, exec in a function can't create shadows of variables if these are declared "global" in another function of the same module
New submission from Silas S. Brown <[EMAIL PROTECTED]>:
Here's the example code:
setting1 = "val1"
setting2 = "val2"
def dummy():
global setting1
def f(x):
d ={"setting1":setting1,"setting2":setting2}
exec(x) in d
return d['setting1'], d['setting2']
print f("setting1=setting2='new'")
Expected result: ('new', 'new')
Actual result: ('val1', 'new')
The presence of "global setting1" in a different function effectively
stops a shadowed setting1 from being created by the exec.
Workaround: Add a real assignment before the exec, i.e.:
def f(x):
setting1 = 0
exec(x)
return setting1, setting2
or do the exec in a dictionary instead of in the current scope.
Observed in:
Python 2.5.1 (r251:54863, May 18 2007, 16:56:43) on Cygwin
Python 2.5.2 on 2.6.26-gentoo-r1 (by Christopher Faylor
http://www.cygwin.com/ml/cygwin/2008-11/msg00168.html )
Not observed in:
Python 2.5.1 (r251:54863, Aug 1 2008, 00:32:16) on SUSE Linux
Python 2.4.4 (#2, Apr 17 2008, 01:58:28) (Debian etch, ARM)
Python 2.5.2 (r252:60911, Jul 31 2008, 17:31:22) (Ubuntu)
--
components: None
messages: 75829
nosy: ssb22
severity: normal
status: open
title: On some Python builds, exec in a function can't create shadows of
variables if these are declared "global" in another function of the same module
type: behavior
versions: Python 2.5
___
Python tracker <[EMAIL PROTECTED]>
<http://bugs.python.org/issue4315>
___
___
Python-bugs-list mailing list
Unsubscribe:
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue4315] On some Python builds, exec in a function can't create shadows of variables if these are declared "global" in another function of the same module
Silas S. Brown <[EMAIL PROTECTED]> added the comment:
Sorry, I accidentally posted the workaround code instead of the bug
example. This is what I should have posted:
setting1 = "val1"
setting2 = "val2"
def dummy():
global setting1
def f(x):
exec(x)
return setting1,setting2
print f("setting1='new' ; setting2='new'")
Expected result: ('new', 'new')
Actual result: ('val1', 'new')
The presence of "global setting1" in a different function effectively
stops a shadowed setting1 from being created by the exec.
Workaround: Add a real assignment before the exec, i.e.:
def f(x):
setting1 = 0
exec(x)
return setting1, setting2
or do the exec in a dictionary instead of in the current scope, i.e.:
def f(x):
d ={"setting1":setting1,"setting2":setting2}
exec(x) in d
return d['setting1'], d['setting2']
Observed in:
Python 2.5.1 (r251:54863, May 18 2007, 16:56:43) on Cygwin
Python 2.5.2 on 2.6.26-gentoo-r1 (by Christopher Faylor
http://www.cygwin.com/ml/cygwin/2008-11/msg00168.html )
Not observed in:
Python 2.5.1 (r251:54863, Aug 1 2008, 00:32:16) on SUSE Linux
Python 2.4.4 (#2, Apr 17 2008, 01:58:28) (Debian etch, ARM)
Python 2.5.2 (r252:60911, Jul 31 2008, 17:31:22) (Ubuntu)
___
Python tracker <[EMAIL PROTECTED]>
<http://bugs.python.org/issue4315>
___
___
Python-bugs-list mailing list
Unsubscribe:
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue1384175] random module - Provider DLL failed to initialize correctly
Silas S. Brown added the comment:
I got a very similar error on an Otek Pocket PC running Windows Mobile
2003 SE and the latest version of pythonce from
pythonce.sourceforge.net. The error is:
File
"C:\devl\release\PythonCE-2.5-20061219\Python-2.5-wince\Lib\random.py",
line 108, in seed
: [Error 87] Provider DLL failed to
initialize correctly
Although this was thrown up at the "import random" at the start of my
program, the actual change I made that resulted in this error was much
later in the program, and it was to change the lines
try: justSynthesize=raw_input("Say: ")
except EOFError: break
into
try:
justSynthesize=raw_input(cond(winCEsound,"".join(warnings_printed)+cond(warnings_printed,"\n",""),"")+"Say:
") # (WinCE uses an input box so need to repeat the warnings if any)
except EOFError: break
where "cond" is an "if a then b else c" function.
A little more investigation showed that the culprit was the comment!
Removing the comment after the raw_input() call (or putting it on a
different line) causes the program to work again.
I confirmed that adding any raw_input() call to any function, with a
prompt parameter and a comment afterwards, causes this error to happen
on the "import random" near the top of the program. This is a very
strange bug.
--
nosy: +ssb22
___
Python tracker
<http://bugs.python.org/issue1384175>
___
___
Python-bugs-list mailing list
Unsubscribe:
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
[issue1384175] random module - Provider DLL failed to initialize correctly
Silas S. Brown added the comment: After further investigation I'm suspecting that this issue is actually due to the process running out of RAM. ___ Python tracker <http://bugs.python.org/issue1384175> ___ ___ Python-bugs-list mailing list Unsubscribe: http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
