[boost] Re: shifted_ptr: why?

From: David Abrahams (dave_at_[hidden])
Date: 2003-10-01 17:46:34

"Rozental, Gennadiy" <gennadiy.rozental_at_[hidden]> writes:

>> If every library for which a more-general framework could be 
>> envisioned had to wait around for that more general framework 
>> to appear, we'd never approve anything.
>
> Let's not over generalize here. We are talking about specific
> library and specific generic framework that was extensively
> discussed here and not something will couldn't even envision.

Yes, it can be envisioned, but has not been implemented, at least not
to Boost standards.  That's exactly what I meant.  Let's go back from
the general to the specific.  Why should this specific library be
treated differently from other libraries for which a more generic
framework might be envisioned?

>> I'm not saying we should approve shifted_ptr necessarily, 
>> but the mere fact that it isn't a policy in a PBSP framework 
>> (which we don't > have) shouldn't be grounds for rejecting it.
>
> In this specific case I disagree. We know more or less that there is an
> ability to design generic framework for smart pointer  class of solutions
> (There are several attempts already to formalize it). 

Yes, I think the fact that so far none of those several attempts have
(to my knowledge) been successful in dealing with the exception-safety
issues is good grounds for not delaying other libraries based on the
speculation that it can be done.

> No need to produce numerous custom solution while we better spent
> our efforts on generic one (and any new library is some effort -
> effort to test, effort to publish, effort to support and so on).

And you claim that having a PBSP would significantly reduce the
testing, publishing, and support efforts for this library?  Have you
taken a close look at the implementation, or are you basing this
assertion just on the fact that it's a smart pointer?

> I still feel that in this particular case we already in a position to accept
> only generic solution as a new submission. Any other step would be a waste
> of efforts.

Whose effort are you claiming would be wasted?  

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com