Boost logo

Boost :

Subject: Re: [boost] [serialization] Warnings from spirit header in XML archive headers (non_terminal/rule.hpp)
From: Number Cruncher (number.cruncher_at_[hidden])
Date: 2010-05-26 05:00:39


On 25/05/10 17:46, Robert Ramey wrote:
> I spent some time with a similar patch. I found no way a makiing it
> compatible with one of the compilers which serialization supported at the
> time.
> (borland). So I just left things as they were.
>
> In the meantime the borland compiler version of the serialization library
> has been broken in any case so my original motivation for not doing
> something
> like this non longer exists. On the other hand, the warning is totally
> benign
> and only appears when the library is built - not in user programs so I
> don't think it's worth any effort to address.
>

Except when you want to build on the xml_iarchive e.g. to implement
different header handling. Then I need to invoke the
boost/archive/impl/xml_iarchive_impl.ipp template instantiations which
pulls in the basic_xml_grammar.hpp header.

Neither of these are library-compiled-only; they are part of boost
include directory and are the recommended way of extending some of the
serialization archives (see e.g
http://www.boost.org/doc/libs/1_43_0/libs/serialization/doc/derivation.html
)

> If one want's to do something truely worthwhile, it would be
> to upgrade the xml_serialization to latest version of Qi/Karma.
> It's purported to be much faster. The serialization library uses
> spirit to parse only one line of xml at a time so it's actually
> a very simple application. That is, it's a good excercise for
> someone who want's to experience spirit but doesn't have
> a real application available.
>
>

I agree that migrating to Qi/Karma is a great idea; I may yet get round
to it some day! However, in the mean time, the spirit authors have
provided a clear upgrade path by first deprecating some old header
locations and changing some paths. Surely the first step to an upgrade
is to be compliant with the new structure by using the symbols from
spirit::classic?

At the end of the day, I can also patch my copy of boost locally with
the non-borland-compliant fix, but in light of your response about
borland problems, couldn't you just commit the patch to the trunk? The
spirit authors state "The forwarding headers are expected to be removed
in the future.", i.e. boost::serialization *will break*. (
http://www.boost.org/doc/libs/1_43_0/libs/spirit/doc/html/spirit/what_s_new/spirit_1_x.html
)

Thanks for incredibly useful library,
Simon.

> Robert Ramey
>
>
>
>
> _______________________________________________
> Unsubscribe& other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
>


Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk