simple integer subclass

Jean-Michel Pichavant jeanmichel at sequans.com
Tue Aug 3 15:44:15 CEST 2010


Jean-Michel Pichavant wrote:
> Andreas Pfrengle wrote:
>> On 3 Aug., 03:22, Carl Banks <pavlovevide... at gmail.com> wrote:>
>>  
>>> You are creating an object that differs from a built-in, int, in a
>>> highly misleading way that only makes sense in a very limited context,
>>> and this object's modified behavior gives no clue that it's been
>>> modified in such as way.  (That is, it's not possible to tell if the
>>> object's not a regular int just by looking at __str__()'s return
>>> value.)  To make matters worse, you want to program this object to
>>> coerce other integers, so there's a risk of these objects escaping
>>> from the context where they make sense.
>>>
>>> This is just a bad idea.  The type is not the place to implement
>>> behavior that makes sense only in a limited context.  Instead, do
>>> something like this:
>>>
>>> print "Item %d is %s." % (i+1, s[i])
>>>     
>>
>> I see your concerns. I started with the approach to add +1 directly
>> before displaying the int. However, since there are some variables
>> that shall be displayed normally and others that are indices I want to
>> show starting at 1, I thought the easiest way would be to define a
>> type that does the job, then I would only need to define it once and
>> not take care everywhere whether I have a normal variable or a
>> displayed index.
>> Thinking about it, it might really be dangerous to coerce always to
>> int1, since sometimes I might want a normal int as result (I can't
>> tell yet for sure).
>> I'm just thinking about only overloading the operations if the int1 is
>> on the left-hand side (so __op__ coerces to int1, while __rop__
>> doesn't). This would make operations non-commutative - but I also
>> would need to put more brains in every calculation, which could
>> finally take more effort than only "upgrading" the display :-???
>> Seems I end up with your suggestion - if noone else has an idea ;-)
>>
>> The application will be a browsergame, and most gamers start counting
>> at 1, so they would probably wonder about a "level 0 item" ;-)
>> If there didn't already exist lots of code, I would redesign the whole
>> data-structure - I think that's "lessons learned" for the next project
>> -.-
>>   
> The level of an item is attribute of this item, not an index in a list.
> You may have an issue with your design, using an improper structure.
>
> In a more general manner, given your example I don't see why you 
> should expose the index of an element in an internal list to the user.
>
> JM
>
"If there didn't already exist lots of code, I would redesign the whole
data-structure"

sorry I didn't get this one at first read, my comment is pretty much 
useless.

JM



More information about the Python-list mailing list