Request for feedback on WG14 proposal N3849 (alloc_at_least)

[Paul Eggert]

On 2026-04-23 16:15, Charles Munger wrote:
> Part of the goal of this proposed standard API is that programs could use
> it with nearly-zero overhead regardless of the libc implementation,

I understand the motivation, but we're not seeing performance figures 
supporting our often-fallible performance intuition, ideally from 
measurements of realistic C applications as opposed to microbenchmarks.

> it would be straightforward for libcs to implement without exposing
> implementation details (if they don't want to) or adding extra storage
> overhead.

Might this have some unintended effects? Suppose, for example, the 
change boosts performance for rewritten apps on implementations where 
realloc performs poorly, but also increases complexity and worsens 
performance for unmodified apps on implementations where realloc 
performs well.

Since the main motivation for the proposal is performance, it would be 
helpful to have realistic measurements to see what the performance 
effects are in practice. Otherwise we're kinda in the dark about the 
proposal's costs and benefits.