Re: [Boost-users] [serialization] counting the xml cost

Subject: Re: [Boost-users] [serialization] counting the xml cost
From: Michael Powell (mwpowellhtx_at_[hidden])
Date: 2013-04-20 16:16:42

On Sat, Apr 20, 2013 at 2:58 PM, Michael Powell <mwpowellhtx_at_[hidden]> wrote:
> Hello,
>
> I am trying to follow the code in
> http://www.boost.org/doc/libs/1_53_0/libs/serialization/doc/serialization.html
> and finding non-member function void
> boost::serialization:save(Archive&, ...). What am I doing wrong?

I am also studying this post for answers:
http://boost.2283326.n4.nabble.com/boost-Serialization-V1-42-0-Question-about-BOOST-SERIALIZATION-SPLIT-FREE-and-non-public-members-td2670917.html

Seems the issue is still (potentially) somewhat intrusive, at least if
you want to not expose a lot of fields for public inspection?

Still getting C++ errors though. I am attempting to friend
boost::serialization methods across namespaces into dchem::model
namespace. Any problems with that, per se?

> As the name suggests, I would like to split the load and save
> concerns. From what I can tell, needs to be in the
> boost::serialization namespace? Anything else?
>
> Thank you...
>
> Regards,
>
> Michael Powell
>
>
> On Sat, Apr 20, 2013 at 11:03 AM, Michael Powell <mwpowellhtx_at_[hidden]> wrote:
>> Hello,
>>
>> Basically, I'd like to find out how well the boost serialization
>> facilities handle Xml and to what degree, robustness, etc.
>>
>> I am counting the cost of serialization for our model. We'll need do
>> some basic CRUD type operations on our model objects throughout our
>> system, and I want to use a nicely loosely coupled solution. Xml is
>> the thought at first; I've also considered a lightweight database like
>> SQLite which transfers easily for patches, upgrades, and other
>> transfers.
>>
>> Looking at the boost serialization facilities, one keyword was
>> emphasized in the tech-agnostic serialization documentation. Tech
>> meaning whether this is a binary file, text, JSON, Xml, whatever. The
>> keyword being that serialization and load occur in the same order.
>>
>> Enter the desire for Xml. One common problem in .NET C# land, at any
>> rate, is that reads are commonly known as being asymmetric from
>> writes. That is, not guaranteed to happen in the same order,
>> composition, whatever. That is, you may see attributes in one
>> instance, where as in another the element might have been broken out.
>>
>> Such as: <myObject name="something" /> and
>> <myObject><name>something</name></myObject> are technically the same.
>> The serializer might scoff at that, but I'm not sure the W3C suggest
>> otherwise. That's my loose understanding of the specification.
>>
>> Regards,
>>
>> Michael Powell