Checking against NULL will be eliminated?

[Terry Reedy]

On 3/3/2011 10:39 AM, Jean-Paul Calderone wrote:

> C and C++ have standards, and the standards describe what they don't
> define.
>
> Python has implementations.

Python also has standards, the language and library manuals.


> The defined behavior is whatever the implementation does.

No, the defined behavior for a particular version of Python
is whatever the manuals for that version say.
Ambiguities are generally removed as discovered.
Some of the ambiguites have been pointed out by authors
of alternate implementations.

The docs have a few notes as to what is not defined or
what is behavior spedific to CPython, and therefore not
part of the *language* definition.
The intent is to have a rather complete definition.

Discrepancies between doc and implementation are bugs.
The usual resolution is a behavior change.

>  Until someone changes it to do something else.

At which point the change is documented with a note as to when the 
change was made. But this is never intentionally done as casually as you 
imply. Contributions to improve test coverage so as to prevent 
unintended changes are welcome.

-- 
Terry Jan Reedy