Path: csiph.com!v102.xanadu-bbs.net!xanadu-bbs.net!feeder.erje.net!us.feeder.erje.net!news.linkpendium.com!news.linkpendium.com!panix!roy
From: Roy Smith <roy@panix.com>
Newsgroups: comp.lang.python
Subject: Re: Correct type for a simple “bag of attributes” namespace object
Date: Sun, 03 Aug 2014 09:25:53 -0400
Organization: PANIX Public Access Internet and UNIX, NYC
Lines: 42
Message-ID: <roy-E2F830.09255203082014@news.panix.com>
References: <7ef67ccc-3fc3-47dd-b858-09ef3b57a497@googlegroups.com> <87r40zmkhr.fsf@elektro.pacujo.net> <53dd0717$0$29973$c3e8da3$5496439d@news.astraweb.com> <857g2qd5oc.fsf_-_@benfinney.id.au> <mailman.12580.1407068955.18130.python-list@python.org> <roy-AC5165.08404803082014@news.panix.com> <mailman.12581.1407070605.18130.python-list@python.org>
NNTP-Posting-Host: localhost
X-Trace: reader1.panix.com 1407072354 4618 127.0.0.1 (3 Aug 2014 13:25:54 GMT)
X-Complaints-To: abuse@panix.com
NNTP-Posting-Date: Sun, 3 Aug 2014 13:25:54 +0000 (UTC)
User-Agent: MT-NewsWatcher/3.5.3b3 (Intel Mac OS X)
Xref: csiph.com comp.lang.python:75595

In article <mailman.12581.1407070605.18130.python-list@python.org>,
 Chris Angelico <rosuav@gmail.com> wrote:

> On Sun, Aug 3, 2014 at 10:40 PM, Roy Smith <roy@panix.com> wrote:
> >  I usually just do:
> >
> > class Data:
> >    pass
> > my_obj = Data()
> >
> > That's all you really need.  It's annoying that you can't just do:
> >
> > my_obj = object()
> >
> > which would be even simpler, because (for reasons I don't understand),
> > you can't create new attributes on my_obj.
> 
> [...]
> The only reason I can think of for expecting a
> basic object to work this way is because of the parallel with
> ECMAScript; it's not a fundamental part of the type hierarchy.

I don't know about that.  I agree that your explanation about object 
size makes sense for why it wasn't built to work that way, but I don't 
think the expectation should be surprising.  When I write:

class Foo(Bar):
   # stuff

I'm saying, "make Foos be just like Bars, except for this stuff".  I can 
do:

>>> class Foo(object):
...     pass
... 
>>> f = Foo()
>>> f.x = 1

in which case, I've said, "make Foos just like objects, except for, oh, 
never mind, there aren't any differences".  But, in reality, the system 
bolted on the ability to have user-defined attributes without telling 
me.  I don't think it's unreasonable to be surprised at that.