Strange __future__ behavior in Python 2.5
mdst...@gmail.com
something like:
from __future__ import foo, bar
to enable more than one feature. However, this does not seem to be
working properly in 2.5; it behaves as expected when typed into the
interactive interpreter, but not when it is in a module. When I try to
import the following module:
from __future__ import with_statement, division, absolute_import
def bar():
print 5/3
with open('asdf') as f:
for line in f: print line.strip()
I get a warning that 'with' will soon be a reserved keyword, and a
SyntaxError on the line with the with statement, so obviously, the
__future__ statement is not working. When I change the first line to:
from __future__ import with_statement
from __future__ import division,absolute_import
then the with statement works fine. However, the true division also
works fine, so apparently making multiple __future__ imports on one
line works for division, but not for with_statement.
Is this a bug, or am I misunderstanding something? I'm using the final
release of Python 2.5 (r25:51918, Sep 19 2006, 08:49:13) on Mac OS X.
Avi...@gmail.com
> My understanding of the __future__ statement is that you may say
> something like:
>
> from __future__ import foo, bar
>
> to enable more than one feature. However, this does not seem to be
> working properly in 2.5; it behaves as expected when typed into the
> interactive interpreter, but not when it is in a module. When I try to
> import the following module:
*snip*
> Is this a bug, or am I misunderstanding something? I'm using the final
> release of Python 2.5 (r25:51918, Sep 19 2006, 08:49:13) on Mac OS X.
Only one "from __future__" can be imported per line.
So,
from __future__ import foo
from __future__ import bar
etc.
It will only import the first if you give multiple.
Why this is, I don't know.
Georg Brandl
This is a bug and has now been fixed in the SVN repo.
Thanks for bringing it up.
Georg
Avi...@gmail.com
Georg Brandl wrote:
> This is a bug and has now been fixed in the SVN repo.
> Thanks for bringing it up.
Ouch, I feel bad now. I've been noticing this behavior since 2.5B1 but
I didn't realize it was a bug.