On Tue, Nov 28, 2023 at 6:34 PM Peter Dimov via Boost
This hasn't been working very well for WG21. The trend is towards proposals named P00xxR38 with quality often not matching the average rejected Boost library.
To be fair, most of the R38 libraries are large and difficult. But obviously I think something important has been lost here and I'd like to restore it. Mostly I think that's libraries in the hands of lots of users, including committee members as users.
LEWG is not an efficient mechanism for designing libraries, because iterating here is vastly faster and better than iterating at WG21 meetings.
Indeed. And importantly, we have actual real users to find problems...
(But to ensure smooth sailing one would nowadays need to remember to keep extensive design notes on how the library came to be this way and not some other, in order to produce these on demand when asked in LEWG.
I don't see this part as a bad thing -- even Boost is hit and miss and documenting decisions. It's a lot of work.
This used to be less needed in the past when LWG and Boost significantly overlapped.)
I assume you mean when LEWG basically didn't exist and it was *just LWG*. For sure. But the community has expanded past that -- and so we need to find new ways in my view to do the work that can't realistically be done in LEWG in a place like Boost. Apparently new authors don't find being in Boost of value -- I'm hoping we can change that going forward, we'll see. Jeff