pydoc creates a strange description for the __doc__
member: it prints the doc-string of 'str'. Example:


C:\>c:\sf\python\dist\src\PCBuild\python.exe
Python 2.3a0 (#29, Apr 15 2002, 18:33:31) [MSC 32 bit
(Intel)] on win32
Type "help", "copyright", "credits" or "license" for
more information.
>>> class O(object):
...   "some text"
...
>>> import pydoc
>>> pydoc.help(O)
Help on class O in module __main__:

class O(__builtin__.object)
 |  some text
 |
 |  Data and non-method functions defined here:
 |
 |  __dict__ = <dict-proxy object at 0x0080D410>
 |
 |  __doc__ = 'some text'
 |      str(object) -> string
 |
 |      Return a nice string representation of the object.
 |      If the argument is a string, the return value
is the same object.
 |

The attached patch prints the following (also for the
HTML output):
 |  __doc__ = 'some text'
 |      The documentation string
 |

Logged In: YES 
user_id=11105

The same problem exists for the __module__ attribute. I can
update the patch if needed.

Logged In: YES 
user_id=11105

Please ignore my previous comments.

It turns out that the problem is not in pydoc, it is in
Python itself. "spam".__doc__ is the same as str.__doc__
(which describes what the callable 'str' does.
Is this intended?

The side-effect is that pydoc prints unexpected output for
class variables. Execute this code to find out:

class X:
    a = "a string"
    b = 42

import pydoc
pydoc.help(X)

Logged In: YES 
user_id=11105

Please ignore my previous comments.

It turns out that the problem is not in pydoc, it is in
Python itself. "spam".__doc__ is the same as str.__doc__
(which describes what the callable 'str' does.
Is this intended?

The side-effect is that pydoc prints unexpected output for
class variables. Execute this code to find out:

class X:
    a = "a string"
    b = 42

import pydoc
pydoc.help(X)

Logged In: YES 
user_id=6380

The instance doc is supposed to be the same as the class
doc, yes.

I suppose the str() __doc__ is not particularly appropriate
for string instances.

Can you suggest a better wording? (This problem is
widespread, I presume, so a generic solution may be in
order.)

Logged In: YES 
user_id=11105

No, I cannot think of a better wording.
The problem (as you know) is that we have to cover three 
cases: the str() function, the str type, and the string 
instances (same for all the other 'simple' types).
Would it be worth to add a _doc field to the PyStringObject 
structure and add a __doc__ entry to tp_members?

Logged In: YES 
user_id=11105

Or maybe add a __doc__ tp_member (or whatever) which always 
returns None?

Logged In: YES 
user_id=11105

The attatched patch (stringobject.diff) implements this
behaviour: The doc string of 'str' is unchanged, but string
objects have a __doc__ attribute of None.
The same could (and should) be done for int, float, and
probably more types as well.

Logged In: YES 
user_id=80475

Since this would also go into int, float, and other places, 
consider factoring it to abstract.c as 
PyObject_GenericGetNoneDoc or some such.

Logged In: YES 
user_id=6380

I'm very sympathetic, but have no time to review this just
yet. Please don't check in without my approval -- this is
awfully deep shit (see the history of the __doc__ descriptor
in typeobject.c).

Logged In: YES 
user_id=6380

I think there are actually two issues here: pydoc shows the
docstrings for what it calls "data and non-method
functions", which usually aren't helpful; and the docstrings
for basic types are confusing when retrieved through an
instance.

For the pydoc issue, I have a different patch
(pydoc2.diff).  This suppresses the docstring completely
unless the value is callable. (Now, I know that callable()
is not 100% reliable, but it's close enough for pydoc.)

For the other issue, I'm not sure *what* to do. Tim suggests
that perhaps for the basic types, x.__doc__ should return
None. Opinions?

Logged In: YES 
user_id=11105

> I think there are actually two issues here: pydoc shows the
> docstrings for what it calls "data and non-method
> functions", which usually aren't helpful; and the docstrings
> for basic types are confusing when retrieved through an
> instance.
> 
> For the pydoc issue, I have a different patch
> (pydoc2.diff).  This suppresses the docstring completely
> unless the value is callable. (Now, I know that callable()
> is not 100% reliable, but it's close enough for pydoc.)
Hm, I see no pydoc2.diff ;-)
IMO, pydoc should try to document the _purpose_ of the
special attributes (__doc__, __module__, ...) instead
of their _value_. 

> For the other issue, I'm not sure *what* to do. Tim suggests
> that perhaps for the basic types, x.__doc__ should return
> None. Opinions?
That's what my sample patch stringobject.diff does.
+1 from me.

Logged In: YES 
user_id=6380

> Hm, I see no pydoc2.diff ;-)

Oops, here's pydoc2.diff

> IMO, pydoc should try to document the _purpose_ of the
> special attributes (__doc__, __module__, ...) instead
> of their _value_.

Maybe, but aren't there lots of these?  Maybe this ought to
be table-driven instead of an if-statement for each one.
I'll call YAGNI on this one for now; at best it's a separate
feature request.

> > For the other issue, I'm not sure *what* to do. Tim
suggests
> > that perhaps for the basic types, x.__doc__ should
return
> > None. Opinions?
> That's what my sample patch stringobject.diff does.
> +1 from me.

That's what I realized after looking at your string patch.
:-)

Raymond, do you want to work on finishing this?

Logged In: YES 
user_id=11105

Guido, after applying pydoc2.diff I'm happy. I retract 
the "feature request".

Logged In: YES 
user_id=6380

OK, I've checked in pydoc2.diff.

I'm not sure if we need to do anything about the docstrings
of basic objects; I'll close this bug report, if there's a
desire for a change there, somebody can open a new bug
report.