C++ Logo


Advanced search

Re: [std-proposals] Forwarding refs for class template arg deduction

From: Tom Honermann <tom_at_[hidden]>
Date: Tue, 26 Mar 2024 10:17:21 -0400
On 3/26/24 8:03 AM, Frederick Virchanza Gotham via Std-Proposals wrote:
> On Monday, March 25, 2024, Jonathan Wakely wrote:
> At some point we do have to accept that this poster is just
> spamming the list with half baked nonsense.
> This mailing list isn't exclusively for finalised papers.
> Every idea starts out as a 'half-baked idea', although there are other
> terms for such embryonic thinkings.
> Sometimes an idea should be kept secret until it has been researched
> and developed. And sometimes an idea should be shared right at the
> very beginning.
> Not trying to get all biblical on you here but let's consider the
> rudder of a ship. If the rudder is off by a few degrees at the very
> beginning of the journey, then the error over a long distance can
> result in significant deviation. Ideas can be like this too. Some
> ideas should be shared at the 'half-baked idea' stage, so they can be
> floated for people to pass remark -- and that's what I intended here
> with having a new syntax for '^template^'. It seems my only oversight
> on this occasion was that I hadn't seen an explicit deduction guide
> before.
> Some ideas should be floated here on the mailing list in their
> pre-baking phase to get people's thoughts and opinions. Maybe you want
> a separate mailing list just for finalised papers though, or for ideas
> with at least X hours of research and development.
> Not every idea should be shared prematurely though. I'm currently
> writing a paper to add self-modifying code to the C++ standard. I
> don't want to curtail my own creativity and so I've purposely not
> shared my idea with anyone, nor have I researched any pre-existing
> library that does self-modifying code nor 'just in time compilation'.
> I don't know if my idea is original, or if it's been done
> independently 7 times in the past 10 years, but I want to polish it
> off all by myself before asking for opinions or comparing my own
> technique to how other people did it.
> I can't find the original mission statement for this mailing list, but
> it was posted here a while back, and I believe it contained the term
> 'float an idea'.

The "mission statement" (or equivalent) is described at
https://isocpp.org/std/submit-a-proposal. Quoting:

> Post an initial brief description of your feature on the
list, *including especially the problem it solves and alternatives

There is nothing wrong with floating your ideas here. I think what
people are objecting to is the frequent posting of ideas that do not
include any motivation or description of the intended improvements they
are expected to provide. I highlighted some of the quoted text above;
this is what I think people frequently find missing in your posts.

For example, a post that consists of "What if we were to do X?" falls
short of the criteria above. What we would prefer to see are posts like,
"Allowing X would enable code that, today, has to be written in this way
... to instead be written in this way and to run more efficiently. What
do you think?". In general, demonstrating implementability is not
necessary (and is a distraction) here unless there is push back on an
idea based on implementability concerns. That being said, providing a
link to godbolt.org <https://godbolt.org> or similar that demonstrates
the idea and problems it solves are quite welcome.



Received on 2024-03-26 14:17:22