I can write aggregating allocation for a C with my eyes closed, in a tiny number of lines. These days I mostly wouldn't. If the target system thinks it's a good idea to add a header to every 32 byte allocation, that's their problem. No worthwhile malloc implementation does that, other than in a debugging configuration, where it adds "red zones".

The calculations in my other post show that it's not as simple as that next pointer being pure overhead compared to open addressing.

https://news.ycombinator.com/item?id=35416520