Jack Lloyd

Thanks. Right now we rely heavily on transport layer (ie whatever the IO callback layer is doing) to coalesce writes. That works ok for TCP most of the time but...

Looking at it further `get_cpu_total` isn't something we even need or use, except for this test. We can simplify things down to just `get_cpu_available`, which returns a best guess at...

Apologies. I had been under the impression this was unused/unusable because bakefile (or so I thought) only supported up to Visual C++ 2013 which is no longer supported in 3.x...

No update or immediate development plans. Hash2curve is implemented now, which is the necessary first step for OPAQUE and/or CPace support. Partially thing haven't moved because for a long time...

This would be nice to have. I don't have a ton of free time right now but would be happy to accept patches for this. Actually though I think our...

@prashanttholia Absolutely. If you have questions or etc feel free to open an issue or leave them here.

@prashanttholia of course

The problem is a race between static destructors, so I think it not happening with libstdc++ just by chance. The problem the pooling allocator (which hands out memory from a...

One option which is a little unappealing but does have the advantage of not crashing :) would be to not `munmap` the memory in the destructor of ~locking_allocator (and just...

It seems like a GCC/Clang `destructor(0)` attribute would allow freeing 'mostly' last, though with shared objects it seems very hard to say what exactly will happen with static destructors. Additionally...