overriding setting

[Francesco Guerrieri]

On 6/6/07, Marc 'BlackJack' Rintsch <bj_666 at gmx.net> wrote:
> In <mailman.8748.1181124863.32031.python-list at python.org>, Francesco
> Guerrieri wrote:
>
> > Now the question is this:
> > I would like to initialize such an object in this way:
> > a = myList()
> > a = [[1, 2, 3], [4, 5, 6, 7]]
> > a.pad()
> > # and now a _should_ contain [[1, 2, 3, ""], [4, 5, 6, 7]]
> >
> >
> > Obviously this doesn't work, because when at the second line I do the
> > initialization, type(a) becomes <type 'list'>, and so I get the
> > expected AttributeError since pad cannot be found.
>
> You don't initialize in the second line, you just rebind `a` to a
> completely different object.  Names don't have types in Python, objects do.
>
> `list()` takes an optional argument.  Just make sure your derived type
> does to and passes this to the base class `__init__()`.  Then you can
> create an instance like this:
>
> a = MyList([[1, 2, 3], [4, 5, 6, 7]])
>

yes it's true that it is not an initialization :-) It's that I hoped
that there was a way to do an init rather than a rebinding of the
name.
Your suggestion is exactly what I have implemented for the time being...
I subclass the builtin list type, I have a pad method which adds the
requested whitespaces at the end, an append method which invokes the
base class append AND calls the pad method, and finally a __call__
which calls the append. Furthermore, I check that the input is
valid... So everything works fine :-)

The only problem is that while coding I have the temptation to write a
= [[...], [...]) rather than a([1, 2, 3], [5,6, 7, 8]).  Plus I find
it uglier :-) but if there isn't a reasonable way, I'll give up :-)

thanks,
Francesco