Request for comment from LEWG

Request for comment from LEWG

[Soul Studios]

Hi all,
was hoping to hear back from Jonathan but haven't so far, so just 
posting here for comment too - this is Matt Bentley.

std::hive (previously: colony) has had a preliminary meeting under LEWG 
- there were a couple of questions raised by the chairs based on the 
concern that the secondary option (memory_use) in the 'hive_priority' 
template parameter might not end up being implemented by implementors, 
which then would get frozen into ABI. They recommended I contact 
yourselves, amongst others.
If someone could get back to me on the following before april 5th (the 
next hive lewg meeting) that'd be appreciated:

a). Would libstdc++ be more likely to use the reference implementation 
(https://github.com/mattreecebentley/plf_hive, which has the memory_use 
implemented) or write it's own? The reference is under a zLib license, 
which is compatible with GPL, but as far as I know I could 'grant' a 
version to libstdc++ under the GPL, if this were a problem (the original 
would retain it's zlib license).

b). If libstdc++ were to roll it's own implementation, would the 
memory_use priority be likely to be implemented? In terms of the 
reference implementation, this option merely changes the skipfield type 
from unsigned short to unsigned char - which reduces skipfield memory 
use while limiting maximum block capacities, which in turn limits 
performance above ~500 elements. There are more memory-saving designs 
possible, which are described in the 'Design Decisions' section of the 
paper 
(http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p0447r19.html), 
but to date I have not tried these, have only indicated that they are 
possible.

Thanks in advance-
Matt