namespace question

Bengt Richter bokr at oz.net
Sat Sep 4 21:31:56 CEST 2004


On Fri, 3 Sep 2004 08:11:35 +0200, aleaxit at yahoo.com (Alex Martelli) wrote:

>Elaine Jackson <elainejackson7355 at home.com> wrote:
>
>> I would like to be able to write a function f, which will live in a module M,
>> and will call a function g, such that, when f is imported from M into the
>> interpreter, and invoked there, its invokation of g will return the
>> interpreter's global namespace. Is there a way to do this and, if so, how?
>> Muchas gracias for any and all assistance.
>
>There is no "interpreter global namespace".  If you mean the namespace
>of the calling module (which could be the interactive session's __main__
>if that's who did the importing) you can get there via sys._getframe --
>it's meant for debugging purposes (note the leading underscore) but it
>will work.  If you mean the namespace of __main__ (which could be the
>interactive session if there is one) you can get the __main__ module
>anywhere as sys.modules['__main__'] and a vars call around that will
>give the module's dictionary (which is, I assume, what you mean by
>namespace).
>
This brings up the question of standard terminology (e.g. for namespace).
If one googles for python glossary, the top references are mostly

    http://manatee.mojam.com/python-glossary

which seems not to have caught on so far (mea culpa too, since it takes volunteers ;-/ ).

Anyway, my take on 'namespace' is a name->value mapping. I.e., pretty abstract.
Then follows a discussion of various python mechanisms that do that (whether via
a single dict's d[name] mapping or the infinite d.get(name, default) mapping,
or the mapping implicit in the search, e.g., for a bare name using a function's
local name lookup rules (local, closure, global, builtin) -- not to mention the
mapping names undergo when used as attribute names in various ways.

Compared to other languages, python's name spaces (in my sense) seem to have
evolved interestingly ;-) I am wondering whether the future holds a more unified
model of namespaces and their access mechanisms, so most inspect and sys._getframe
name access hacks will be obviated.

Re defintions, perhaps it is time to reinvigorate Skip's wiki or something like that?

There was an interesting thread re improving python doc functionality at

http://mail.python.org/pipermail/python-list/2004-May/219682.html

An online extension of help() maybe taking a syntax clue from google ,like

    help('define:xxx')

might be an interesting hook into a glossary of definitions.

Regards,
Bengt Richter



More information about the Python-list mailing list