Blobs are subclassable

[Jim Fulton]

Blobs do not support subclassing, but people are tempted to subclass them.

At a minimum we need to make very explicit in their documentation that subclassing isn't supported. It might be worthwhile going further, Marius suggested raising an error in __init__ if self is of the wrong class. That should be easy and might provide adequate protection.

[Christian Theune (ctheune)]

Letting Sebastian handle this.

[Sebastian Wehrmann (sw-gocept)]

I've worked on this. Here is my patch.
Can somebody have a look on this, please?

[Jim Fulton (jim-zope)]

This looks fine, although I don't like the extra indirection through
_init_state. I'd rather just have __init__ call the original
__setstate__.

Jim

On Sep 6, 2007, at 5:10 AM, Sebastian Wehrmann wrote:

> I've worked on this. Here is my patch.
> Can somebody have a look on this, please?
>
> ** Attachment added: "C:\Users\sweh\Desktop\bug127182.patch"
> http://launchpadlibrarian.net/9146563/C%3A%5CUsers%5Csweh%
> 5CDesktop%5Cbug127182.patch
>
> --
> Blobs are subclassable
> https://bugs.launchpad.net/bugs/127182
> You received this bug notification because you are a direct subscriber
> of the bug.

--
Jim Fulton mailto:<email address hidden> Python Powered!
CTO [phone] http://www.python.org
Zope Corporation http://www.zope.com http://www.zope.org

[Sebastian Wehrmann (sw-gocept)]

Jim Fulton wrote:
> This looks fine, although I don't like the extra indirection through
> _init_state. I'd rather just have __init__ call the original
> __setstate__.
>
Should I change it or leave it?

--
Sebastian Wehrmann

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

[Jim Fulton (jim-zope)]

On Sep 6, 2007, at 10:25 AM, Sebastian Wehrmann wrote:

> Jim Fulton wrote:
>> This looks fine, although I don't like the extra indirection through
>> _init_state. I'd rather just have __init__ call the original
>> __setstate__.
>>
> Should I change it or leave it?

I don't feel strongly about it. I'd prefer that you change it, but I
can do it later if it really bugs me. :)

Thanks for working on this!

Jim

--
Jim Fulton mailto:<email address hidden> Python Powered!
CTO [phone] http://www.python.org
Zope Corporation http://www.zope.com http://www.zope.org

[Sebastian Wehrmann (sw-gocept)]

Ok, here another patch with code Jim might prefer.

Now someone privileged should decide, which one to check-in :)

Regards,
  Sebastian

[Sebastian Wehrmann (sw-gocept)]

Me again,

there was a little typo in a comment, changed atomically to automatically. Don't ask me why.

Switched it back.

Sorry for this.

[Christian Theune (ctheune)]

Am Donnerstag, den 06.09.2007, 18:13 +0000 schrieb Sebastian Wehrmann:
> Me again,
>
> there was a little typo in a comment, changed atomically to
> automatically. Don't ask me why.
>
> Switched it back.
>
> Sorry for this.

I think you got caught up with the method signature of __init__().

It doesn't need a "state" argument and can simply pass "None" to
__setstate__.

Christian

[Sebastian Wehrmann (sw-gocept)]

Christian, your right. Here the corrected version.

[Jim Fulton (jim-zope)]

On Sep 7, 2007, at 1:12 AM, Sebastian Wehrmann wrote:

> Christian, your right. Here the corrected version.

Looks good. Thanks again.

Jim

--
Jim Fulton mailto:<email address hidden> Python Powered!
CTO [phone] http://www.python.org
Zope Corporation http://www.zope.com http://www.zope.org

[Christian Theune (ctheune)]

Applying Basti's patch.

[Christian Theune (ctheune)]

Fix on trunk means released.

[Christian Theune (ctheune)]

released in b4