Thanks very much for all your responses. It's pretty clear now that
what I thought I could somehow do is not "pythonic" and for good
reason.
Kent Johnson says it well: "Module dependencies should be one-way" I
can buy that.
On Mon, Jun 29, 2009 at 2:47 PM, Kent Johnson<ken...@tds.net> wrote:
> On Mon, Jun 29, 2009 at 1:23 PM, Robert Lummis<robert.lum...@gmail.com> wrote:
>> Here's an example that seems not possible in python. I'm probably
>> missing something so please enlighten me. I only tried doing this as
>> an exercise to show myself how name references work. I'm not saying
>> it's needed or that it's good practice.
>>
>> I can write the following as a single file and it works as expected:
>>
>> ===snip===
>> #!/usr/bin/python
>>
>> def show(*args):
>> print
>> for arg in args:
>> print arg + ':',
>> exec('print ' + arg)
>>
>> a=15
>> b='hello'
>> x=['bob',3]
>>
>> show('a')
>> show('a','b')
>> show('a','b','x')
>> ===snip===
>>
>> The calls to 'show' output lines like "a: 15" which could be useful
>> for debugging or some such purpose.
>>
>> However, it seems that I can't put the function definition in a file
>> and import it because I can't find a way to refer to an object in the
>> main program file from within a module file.
>
> Right. We did recently discuss ways to implement this function:
> http://www.mail-archive.com/tutor@python.org/msg35873.html
>
>> I understand that it's a
>> good thing to contol which namespaces are referenced by which code but
>> isn't there sometimes a need for code in a module to access the main
>> program file's namespace? My example may be a little contrived but
>> isn't this ability legitimately needed at times?
>
> Generally no, that would be a design smell. Module dependencies should
> be one-way; if main needs module foo, then foo should not have to know
> about main.
>
> Kent
>
--
Robert Lummis
_______________________________________________
Tutor maillist - Tutor@python.org
http://mail.python.org/mailman/listinfo/tutor
