[Python-Dev] Simpler reformulation of C inheritance Q.

Christian Tismer tismer@tismer.com
Tue, 06 Aug 2002 11:51:38 +0200

David Abrahams wrote:

[Greg Ewing, adding a level of indirection]

[David Abrahams]
>>>2. The scheme you propose is more costly in memory and cycles than I'd

>>It's only one memory cycle more than it takes to access
>>the existing sub-structures. And it's a lot better than
>>the alternative, which is doing a Python dict lookup!

Of course it is better. But since it it possible to
do more better, a sub-optimal solution will not
make me forget about it.

> When I spoke of memory, I was talking about the extra pointer per level of
> inheritance.
> When I spoke of cycles, I was talking about the cycles to manage that
> memory (probably moot).

Since we are talking of types and meta-types, I believe
memory issues are of minor interest.
There will not be more then a few hundred classes,
and they will be created just once.
The reason why I want to have extra data and function
caches in the types is that this is *very* memory
efficient, in comparison to stuffing things into the
instances (which would be easy to implement).

> It's not too terrible, but I'd like it a lot better if types would just use
> tp_basicsize to find the beginning of the variable stuff so we could embed
> the memory in the type itself. 'Course, I've forgotten more than I knew
> about that code, so I might be barking up the wrong banyan.

That's exactly what I want to do, but I have to find
out how the variable part of types is used at the moment,
and I admit I didn't understand it, yet.

The place where user stuff should go is where instances
have their slots. With meta-types, it now happens that
types become instances, but types refuse to have slots.
This needs to be changed, everything else is a workaround.

regards - chris

Christian Tismer             :^)   <mailto:tismer@tismer.com>
Mission Impossible 5oftware  :     Have a break! Take a ride on Python's
Johannes-Niemeyer-Weg 9a     :    *Starship* http://starship.python.net/
14109 Berlin                 :     PGP key -> http://wwwkeys.pgp.net/
work +49 30 89 09 53 34  home +49 30 802 86 56  pager +49 173 24 18 776
PGP 0x57F3BF04       9064 F4E1 D754 C2FF 1619  305B C09C 5A3B 57F3 BF04
      whom do you want to sponsor today?   http://www.stackless.com/