dictionary interface

[Antoon Pardon]

Op 2005-10-05, Steve Holden schreef <steve at holdenweb.com>:
> Antoon Pardon wrote:
>> 
>> I have been searching some more and finally stumbled on this:
>> 
>> http://docs.python.org/ref/comparisons.html
>> 
>>   Mappings (dictionaries) compare equal if and only if their sorted
>>   (key, value) lists compare equal. Outcomes other than equality are
>>   resolved consistently, but are not otherwise defined.
>> 
>> This seems to imply that the specific method to sort the dictionaries
>> is unimported (as long as it is a total ordering). So I can use whatever
>> method I want as long as it is achieves this.
>> 
>> But that is contradicted by the unittest. If you have a unittest for
>> comparing dictionaries, that means comparing dictionaries has a
>> testable characteristic and thus is further defined.
>> 
>> So I don't need a full implementation of dictionary comparison,
>> I need to know in how far such a comparison is defined and
>> what I can choose.
>> 
> The dict unit tests are probably trying to ensure that the dictionary 
> ordering doesn't change from version to version, which is probably a 
> good idea in case someone (foolishly?) deciess to rely on it.

I doubt that. Just to check I tried the following:

  class Tree:

    def __lt__(self, term):
      return len(self) < len(term)

And the test passed.

> I can't help wondering, though, under what conditions it actually makes 
> sense to compare two dictionaries for anything other than equality.

Yes that is part of the problem, because I can't think of such a
condition it is hard to think of what extra constraints could be
usefull here.

Anyway, I have searched the source of the test for all testing
with regards to < and after some browsing back and fore it seems
it all boils down to the following two tests.

   self.assert_(not {} < {})
   self.assert_(not {1: 2} < {1L: 2L})

-- 
Antoon Pardon