|
Boost Users : |
From: Jean-Noël Rivasseau (elvanor2004_at_[hidden])
Date: 2005-06-17 14:46:04
If you want to make the BGL serializable, how would you proceed?
Intrusively, meaning writing the serialize function as a class
member, or non intrusively ?
You are saying that "One only needs access to the vertices, edges,
and properties of the graph to serialize it."
However, that doesn't seem easy to get access to these vertices,
since it would maybe mean looking at the implementation details for
a *long* time. Do you think I could write the serialization patches
in a reasonable amount of time?
If you think so, I could give it a try, with some intial help.
Jean-Noël
Also, I didn't get a reply to my question about the >> and <<
operator, and if possible, I'd like to understand that.
>
> On Jun 17, 2005, at 10:47 AM, Robert Ramey wrote:
>
>
>> The serialization library is designed to make the possible - and
>> hopefully
>> easy - without modification of the BGL library. This depends on the
>> BGL
>> library exposing enough information in order to save and load back
>> the
>> state
>> of the objects it uses. I can't know that it does this. Note that
>> serialization is implemented totally non-intrusively for all std
>> collections, variant and now shared_ptr. So what you want to do
>> should be
>> possible.
>>
>
> The BGL graph types do expose enough information for serialization.
> One
> only needs access to the vertices, edges, and properties of the graph
> to serialize it. We would love to make the BGL serializable (it's been
> on by TODO list for ages), but it has never percolated to the top of
> the stack. Of course, we'd be happy to review and integrate any
> patches
> to make the BGL graph types serializable.
>
> Doug
>
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>
>
Boost-users list run by williamkempf at hotmail.com, kalb at libertysoft.com, bjorn.karlsson at readsoft.com, gregod at cs.rpi.edu, wekempf at cox.net