[boost] Re: TR2 and C++0x heads up

From: Jonathan Turkanis (technews_at_[hidden])
Date: 2005-04-27 15:41:15

Edward Diener wrote:
> David Abrahams wrote:
>> <snip entire quoted thread>
>>
>> Edward Diener <eddielee_at_[hidden]> writes:
>>
>>
>>>>> as I think the Boost one is better than anything I could do and
>>>>> better than anything else I have ever seen using standard C++. I
>>>>> will be glad to work with Mr. Ramey on the writing up a proposal
>>>>> part, since I am a fairly good writer, but it is obviously up to
>>>>> him and not me whether he wants to propose the library to the
>>>>> committee.
>>>>
>>>> Not really.  If you really want this to happen, *you* should commit
>>>> yourself to proposing it, and then invite Mr. Ramey to join you.
>>>
>>> I disagree on this. It is clearly wrong, in my eyes, to propose any
>>> library to the C++ standard committee when that library is someone
>>> else's work.
>>
>>
>> Why do you think so?
>>
>>
>>> I greatly respect Mr. Ramey for the work he has done and is doing to
>>> make the Boost serialization library as good as it is but I would
>>> never want to propose such a library for inclusion as a standard
>>> library unless the creator of the library wanted to promote it
>>> also. It is also unfair to put that creator in such a position of
>>> extra work which standardizing a library might entail.
>>
>>
>> If that's the only reason, it's easily dealt with.  The creator of
>> the library isn't obliged to do anything to enable standardization.
>> It can be your effort entirely.
>>
>
> Let me try to understand this. Are you saying that you ( the
> proverbial 'you', not the personal 'you' ) would take a library that
> someone else had developed, even if it is admittedly in the public domain, and
> propose it as a standard library to the C++ committee, making whatever
> changes might be necessary to get it accepted ?

I believe Robert said some time ago that he didn't think Boost.Serialization was
(then) a good candidate for standardization, because it was large and could use
more support from the core language.

If you disagree, I don't see anything wrong with your proposing
Boost.Serialization for standardization. Suppose it is accepted into a TR with
some small changes. Then Robert will naturally make the corresponding changes to
the Boost version, and drop the Boost version entirely when the standard version
is universally available. On the other hand, if the library is accepted with
major changes, then assuming the Boost version retains some advantage over the
standard version, the libraries can coexist peacefully.

Jonathan