zope.i18nmessageid.Message doesn't have an interface

[Christian Zagrodnick]

The zope.i18nmessageid.Message class doesn't provide a specific interface.

Is this intentional? I think it should provide an interface so we don't need to do isinstance(bla, Message) any more.

[Philipp von Weitershausen (philikon)]

On 21 Apr 2008, at 08:22 , Christian Zagrodnick wrote:
> The zope.i18nmessageid.Message class doesn't provide a specific
> interface.
>
> Is this intentional? I think it should provide an interface so we
> don't
> need to do isinstance(bla, Message) any more.

In a way it's intentional: only the C implementation is immutable and
therefore secure against manipulation further down the line. That
said, we also have a Python implementation which takes over when the C
implementation isn't available (which makes it really hard to test as
well...).

[Christian Zagrodnick (zagy)]

On 21.04.2008, at 08:58, Philipp von Weitershausen wrote:
> On 21 Apr 2008, at 08:22 , Christian Zagrodnick wrote:
>> The zope.i18nmessageid.Message class doesn't provide a specific
>> interface.
>>
>> Is this intentional? I think it should provide an interface so we
>> don't
>> need to do isinstance(bla, Message) any more.
>
> In a way it's intentional: only the C implementation is immutable and
> therefore secure against manipulation further down the line. That
> said, we also have a Python implementation which takes over when the C
> implementation isn't available (which makes it really hard to test as
> well...).

Does C-Implementation mean we cannot declare an interface?

--
Christian Zagrodnick

gocept gmbh & co. kg · forsterstrasse 29 · 06112 halle/saale
www.gocept.com · fon. [phone] · fax. [phone]

[Philipp von Weitershausen (philikon)]

On 21 Apr 2008, at 09:17 , Christian Zagrodnick wrote:
> On 21.04.2008, at 08:58, Philipp von Weitershausen wrote:
>> On 21 Apr 2008, at 08:22 , Christian Zagrodnick wrote:
>>> The zope.i18nmessageid.Message class doesn't provide a specific
>>> interface.
>>>
>>> Is this intentional? I think it should provide an interface so we
>>> don't
>>> need to do isinstance(bla, Message) any more.
>>
>> In a way it's intentional: only the C implementation is immutable and
>> therefore secure against manipulation further down the line. That
>> said, we also have a Python implementation which takes over when
>> the C
>> implementation isn't available (which makes it really hard to test as
>> well...).
>
> Does C-Implementation mean we cannot declare an interface?

Sure we can: z.i.classImplements(Message, ISomeIface)

[Christian Zagrodnick (zagy)]

On 21.04.2008, at 11:28, Philipp von Weitershausen wrote:
>
>> Does C-Implementation mean we cannot declare an interface?
>
> Sure we can: z.i.classImplements(Message, ISomeIface)

So why are we not doing this? It's still unclear to me what the
intention of not giving it an interface was.

--
Christian Zagrodnick

gocept gmbh & co. kg · forsterstrasse 29 · 06112 halle/saale
www.gocept.com · fon. [phone] · fax. [phone]

[Philipp von Weitershausen (philikon)]

On 21 Apr 2008, at 15:08 , Christian Zagrodnick wrote:
>
> On 21.04.2008, at 11:28, Philipp von Weitershausen wrote:
>>
>>> Does C-Implementation mean we cannot declare an interface?
>>
>> Sure we can: z.i.classImplements(Message, ISomeIface)
>
>
> So why are we not doing this? It's still unclear to me what the
> intention of not giving it an interface was.

Because reimplementation is in fact undesired. We really want people
to use the immutable (and therefore secure) C implementation.

[Christian Zagrodnick (zagy)]

Ah, now I got it. So we'll leave everything as is. Fine with me.