Python 2.4 removes None data type?

[gaudetteje]

I just read in the 'What's New in Python 2.4' document that the None
data type was converted to a constant:
http://python.org/doc/2.4/whatsnew/node15.html

"""
# None is now a constant; code that binds a new value to the name
"None" is now a syntax error.
"""

So, what's the implications of this? I find the lack of explanation a
little puzzling, since I've written code that compares a variable's
type with the 'None' type. For example, a variable would be
initialized to 'None' and if it went through a loop unchanged, I could
determine this at the end by using a conditional type(var) ==
type(None). What will type(None) return now?

 

[Tim Peters]

[[email protected]]

I just read in the 'What's New in Python 2.4' document that the None
data type was converted to a constant:
http://python.org/doc/2.4/whatsnew/node15.html

"""
# None is now a constant; code that binds a new value to the name
"None" is now a syntax error.
"""

So, what's the implications of this?

No implications, for any sane code. You can no longer do things like this:
SyntaxError: assignment to None

That's all.

I find the lack of explanation a little puzzling, since I've written code that
compares a variable's type with the 'None' type. For example, a variable
would be initialized to 'None' and if it went through a loop unchanged, I could
determine this at the end by using a conditional type(var) == type(None).

Python's None is a singleton, meaning that there is a single, unique
instance of type type(None). So in the case you describe, it's
correct, and idiomatic, to test

if var is None:

instead. Checking type(var) against type(None) will still work, but
was never the best way to do it.

If you have an expression like

var = None

that's fine. You're not changing the binding of name "None" there,
you're changing the binding of name "var".

What will type(None) return now?

That hasn't changed:
<type 'NoneType'>

 

[Gary D. Duzan]

I just read in the 'What's New in Python 2.4' document that the None
data type was converted to a constant:
http://python.org/doc/2.4/whatsnew/node15.html

"""
# None is now a constant; code that binds a new value to the name
"None" is now a syntax error.
"""

So, what's the implications of this? I find the lack of explanation a
little puzzling, since I've written code that compares a variable's
type with the 'None' type. For example, a variable would be
initialized to 'None' and if it went through a loop unchanged, I could
determine this at the end by using a conditional type(var) ==
type(None). What will type(None) return now?

Empirically, NoneType, same as before. The change you refer to is
to prevent the user from doing silly things like:

==========================================================================
Python 2.3.4 (#1, Feb 8 2005, 13:07:40)
[GCC 3.3.3 (NetBSD nb3 20040520)] on netbsd2
Type "help", "copyright", "credits" or "license" for more information.==========================================================================
Python 2.4 (#1, Mar 4 2005, 16:55:16)
[GCC 3.3.3 (NetBSD nb3 20040520)] on netbsd2
Type "help", "copyright", "credits" or "license" for more information.==========================================================================

So your idiom should still work. Note that since there is only
one NoneType instance (i.e. None), you can just test for "var is None"
and save yourself some work.

Gary Duzan
BBN Technologies

 

[Warren Postma]

I just read in the 'What's New in Python 2.4' document that the None
data type was converted to a constant:
http://python.org/doc/2.4/whatsnew/node15.html

Implication: A long standing wart in Python now gone. Its time to
gloat. Are there any really evil glitches LEFT in Python? Now go look at
Perl and come back and say "Thank-deity-of-my-choice-I'm-using-Python".


Warren

 

[Steve Holden]

Implication: A long standing wart in Python now gone. Its time to
gloat. Are there any really evil glitches LEFT in Python? Now go look at
Perl and come back and say "Thank-deity-of-my-choice-I'm-using-Python".

Remaining warts that won't disappear:

print >> file, stuff
@decorator

regards
Steve

 

[Michael Hoffman]

Implication: A long standing wart in Python now gone. Its time to
gloat. Are there any really evil glitches LEFT in Python? Now go look at
Perl and come back and say "Thank-deity-of-my-choice-I'm-using-Python".

The fact that True and False are not constants?

 

[Roy Smith]

I just read in the 'What's New in Python 2.4' document that the None
data type was converted to a constant:
http://python.org/doc/2.4/whatsnew/node15.html

"""
# None is now a constant; code that binds a new value to the name
"None" is now a syntax error.
"""

So, what's the implications of this? I find the lack of explanation a
little puzzling, since I've written code that compares a variable's
type with the 'None' type. For example, a variable would be
initialized to 'None' and if it went through a loop unchanged, I could
determine this at the end by using a conditional type(var) ==
type(None). What will type(None) return now?

Just out of curiosity, *why* did you test against type(None). What did it
buy you compared to the simpler var == None'?

In any case, it looks like it does the right thing:

Python 2.4 (#1, Jan 17 2005, 14:59:14)
[GCC 3.3.3 (NetBSD nb3 20040520)] on netbsd2
Type "help", "copyright", "credits" or "license" for more information.<type 'NoneType'>

 

[John Machin]

I just read in the 'What's New in Python 2.4' document that the None
data type was converted to a constant:
http://python.org/doc/2.4/whatsnew/node15.html

"""
# None is now a constant; code that binds a new value to the name
"None" is now a syntax error.
"""

So, what's the implications of this?

Relax. The sky is not falling.

I find the lack of explanation a
little puzzling, since I've written code that compares a variable's
type with the 'None' type.

It is saying that if you had been silly enough to do

None = "fubar"

you will now get a syntax error.

It said absolutely nothing about "removes None data type".

There are (at least) two possible reasons for a lack of explanation:

(1) Python is one of those expensive software products that hit you
with "upgrades" where you have difficulty finding about the changes in
advance (let alone participarting in the design process) and require
you to recode large chunks of your application but you can't find out
with any sort of precision the necessary changes might be from the
vague descriptions on a complex and ever-changing website.

(2) Unless you have done something completely idiotic [which the change
is designed to prevent], no change is required to your code. OR, like
for the changes to integer division, the change is announced for a
future version, you can do "from __future__ inport whatever". Any
furore (and there have been some doozies) noticeable in the newsgroup
is part of the design process, and has died down well before the
changes are released.

Which do you think is more likely?

For example, a variable would be
initialized to 'None' and if it went through a loop unchanged, I could
determine this at the end by using a conditional type(var) ==
type(None).

or by var == None or var is None

What will type(None) return now?

Constants have types. What made you think that its type would change?

 

[John J. Lee]

Warren Postma wrote: [...]

gloat. Are there any really evil glitches LEFT in Python? Now go
look at Perl and come back and say
"Thank-deity-of-my-choice-I'm-using-Python".

Remaining warts that won't disappear:

print >> file, stuff

Not beautiful, yes, but evil? Why?

@decorator

Evil in the wrong hands...


John

 

[gaudetteje]

Sheesh... I didn't actually pull the type(None) comparison out of my
code. I was simply throwing together a possible situation. Had I
thought about it for more than a second I would have remembered how I
would actually use it.

Thanks for clearing up the confusion. I wasn't previously aware that
you could assign anything to None. Glad it's been changed, because
this would be the kind of thing to spawn evil code.

 

[Warren Postma]

The fact that True and False are not constants?

Yowza.

a = True
b = False
False = a
True = b
if (1==2)==True:
print "Doom"

 

[Warren Postma]

The fact that True and False are not constants?

Yowza.

a = True
b = False
False = a
True = b
if (1==2)==True:
print "Doom"

 

[Steven Bethard]

Yowza.

a = True
b = False
False = a
True = b
if (1==2)==True:
print "Doom"

Why stop there when you can really cause some doom:

py> import __builtin__ as bltin
py> bltin.True, bltin.False = bltin.False, bltin.True

As an example of what doom this causes, try typing it at the interactive
prompt, and then see if you can enter anything else. I can't -- the
prompt just locks up. Woo-hoo! Another way to shoot myself (and my
users) in the foot! =)

STeVe

 

[Brian van den Broek]

Steven Bethard said unto the world upon 2005-03-07 11:55:

Why stop there when you can really cause some doom:

py> import __builtin__ as bltin
py> bltin.True, bltin.False = bltin.False, bltin.True

As an example of what doom this causes, try typing it at the interactive
prompt, and then see if you can enter anything else. I can't -- the
prompt just locks up. Woo-hoo! Another way to shoot myself (and my
users) in the foot! =)

STeVe

Hi all,

just tried this in IDLE:

IDLE 1.1================================

The restart was done by IDLE itself -- poor thing just gave up. So, it
looks like Steve has found a `commit suicide' command for an IDLE
shell

Best to all,

Brian vdB

 

[John Machin]

Sheesh... I didn't actually pull the type(None) comparison out of my
code.

Sheesh yourself. Newsgroup readers can't do inspect.hasaclue('J. Random
Poster'); they rely on duck-typing or goose-typing.

I was simply throwing together a possible situation. Had I
thought about it for more than a second I would have remembered how I
would actually use it.

Indeed.

 

[Jacek Generowicz]

Implication: A long standing wart in Python now gone. Its time to
gloat. Are there any really evil glitches LEFT in Python?

Python 2.4 (#1, Dec 1 2004, 14:23:15)
[GCC 3.2.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.

 

[Jarek Zgoda]

Jacek Generowicz napisa³(a):

Implication: A long standing wart in Python now gone. Its time to
gloat. Are there any really evil glitches LEFT in Python?

Python 2.4 (#1, Dec 1 2004, 14:23:15)
[GCC 3.2.3] on linux2
Type "help", "copyright", "credits" or "license" for more information.

False

Ugh, today I've found this in some legacy code (yes, there *is* legacy
code in Python). And I will not touch this module until such syntax
become illegal.