Subject: Re: [boost] Boost.Experimental Re: [variant] Maintainer
From: Gottlob Frege (gottlobfrege_at_[hidden])
Date: 2015-07-02 13:47:31
On Thu, Jul 2, 2015 at 11:33 AM, Robert Ramey <ramey_at_[hidden]> wrote:
> On 7/2/15 7:44 AM, Gottlob Frege wrote:
>> On Sun, Jun 28, 2015 at 10:29 AM, AgustÃn K-ballo BergÃ©
>> <kaballo86_at_[hidden]> wrote:
>>> Perhaps you'd wish to bring back the Sandbox
>>> (http://www.boost.org/community/sandbox.html), which has been superseded
>>> standalone repositories since the move to Git.
>> I think we need *something* for experimental libraries. Even if it is
>> just someone's own git repository stamped (but not endorsed) with the
>> Boost "brand".
>> Or a page in boost.org listing current experiments.
>> Or... something.
>> I'd like some kind of experimental area that would help find the best
>> possible libraries.
> Why isn't the boost library incubator considered ideal here? In combination
> with github it already provides everything necessary - and significantly
> more already. And it's ridiculously easy to use.
Yes, of course; sorry. I wanted to mention blincubator in here somewhere.
Maybe that is the right answer; I don't know enough to know.
I think the other problem, or the real problem, that none of this
solves, is getting wide adoption of experimental features.
A constexpr branch is something bleeding edge users might use, and can
maybe be done without much breakage - it is hopefully mostly addition.
Changing something like boost::optional to have conversion to tribool
instead of bool, or removing operator*() and forcing users to use
get(), etc. Those are big breaks. I have no idea how to get users to
actually be guinea pigs for those experiments. Should we have 3
different optionals to choose from? (Should we randomly select which
optional you get in your boost distribution so as to get good A/B
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk