Re: [boost] [metal] Formal Review - Request for Endorsement

Subject: Re: [boost] [metal] Formal Review - Request for Endorsement
From: Louis Dionne (ldionne.2_at_[hidden])
Date: 2017-04-10 17:29:21

> On Mon, Apr 10, 2017 at 9:39 AM, Michael Caisse via Boost <

> boost_at_.boost

>> wrote:
> 
>> On 4/9/17 09:43, Bruno Dutra via Boost wrote:
>> > Dear Community,
>> >
>> > I'd like to formalize what has been known to many for some time now and
>> > request the formal review of Metal, a modern C++14 library designed to
>> make
>> > template metaprogramming intuitive and effectively replace Boost.MPL.
>> > Klemens Morgenstern has kindly offered to manage the review process.
>> >
>>
>> 
> <snip>
>>
>> There has been some talk/encouragement to review several of the
>> competing metaprogramming libraries at the same time. Did this idea fall
>> apart?
>> &lt;http://lists.boost.org/mailman/listinfo.cgi/boost&gt;
>>
> 
> There just hasn't been a consensus yet on how to go about it AFAICT, but
> it
> does look like the discussions have stalled a little bit, so maybe this is
> a good opportunity to catch up.

I think this post [1] by Odin Holmes does a good job at summarizing
the different tradeoffs between the libraries. A few weeks ago, we talked
about different tradeoffs on the C++ Slack for quite some time, but we
never reached a consensus. I was taking notes, and I think the only thing
everybody agreed on was:
1. speed is an important criteria
2. eager evaluation is the way to go

Given that all the libraries are eager, and all of them are pretty fast
(by MPL standards at least), this is not groundbreaking.

I'd like to reiterate that while we clearly need a pure-type TMP library, I
think it would be a disservice to the community to have more than one.
The libraries are very similar to an end user that's not a TMP expert, and
having multiple libraries would just cause confusion.

I'm not sure what's the path forward if the library authors can't reach a
consensus. We could perhaps start talking about the contentious issues
on this list and try to see what the greater community thinks too.

Louis

[1]:
http://odinthenerd.blogspot.com/2017/03/tradeoffs-of-tmp-mpl-design.html

--
View this message in context: http://boost.2283326.n4.nabble.com/metal-Formal-Review-Request-for-Endorsement-tp4693574p4693614.html
Sent from the Boost - Dev mailing list archive at Nabble.com.