From: David Abrahams (david.abrahams_at_[hidden])
Date: 2001-11-29 16:59:50
Instead of taking an approach which is "likely to annoy pretty much
everybody", why not try something else?
My suggestion: bring up your arguments on the list now and build consensus
around your point-of-view. Instead of working out all these ideas in private
and then presenting an entirely written product to boost on an
all-or-nothing basis, discuss the issues with the author(s) of the
library(s) with which your submission will overlap, and invite them to
cooperate with you in a redesign of the existing facilities. That approach
is much more likely to result in a positive outcome which not only uses your
ideas but respects the design goals of the original facilities. Also, it has
proven workable in the past.
I think your perception that it would annoy everyone is probably correct.
It's not because of the technical issues, though: it has lots more to do
with the approach.
----- Original Message -----
From: "Andrei Alexandrescu" <andrewalex_at_[hidden]>
Sent: Thursday, November 29, 2001 4:38 PM
Subject: Re: [boost] Submission: typelist
> From: "Jeff Garland" <jeff_at_[hidden]>
> > > 2. How would boost users have benefitted from our hosting of that
> > Singleton, Multimethods, Vistor, and Factories. Real tools for building
> > that Boost doesn't have now. The benefits are going to be less where
> > already has solutions: smart pointers, small object allocations, and
> This brings up a delicate point. It is actually exactly where I thought
> there's going to be much discussion; now, in light of the discussion that
> already took place, it seems like that is going to be a _very_ long debate
> (if I don't give up on it from the start).
> The delicate point is that I plan on submitting things from Loki that
> overlap with stuff that already is approved and used in boost. Take for
> example functors - I am going to basically propose a total replacement of
> boost's functors with an improved version of Loki's Functor (that I'm
> working on), and I will bring arguments on why I think Loki's version is
> superior. This is likely to annoy pretty much everybody: (1) the users of
> the current boost functors; (2) the creators of the boost functors, who
> obviously spent a great deal of time working on them; (3) the whole boost
> community, who spent time on reviewing and improving the functors. So I
> no idea on how that is going to go.
> Now that you brought smart pointers - my (and some boosters') belief is
> that, on the contrary, Loki's smart pointers are perhaps the most
> interesting addition to boost.
> Info: http://www.boost.org Unsubscribe:
> Your use of Yahoo! Groups is subject to http://docs.yahoo.com/info/terms/
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk