Re: [DISCUSSION] FILE* slowness

[apache2]

6) I forgot to mention the suggestion to try out mmap-backed I/O.
My thoughts on it are:
  6.1) mmap(2) would be pretty easy and convenient for the read-case, where we 
are reading immutable files.
  6.2) This paper is pretty interesting on the topic of mmap: 
https://www.cidrdb.org/cidr2022/papers/p13-crotty.pdf
  6.3) I think mmap would quickly become very complicated when dealing with r/w 
files, especially when we'd have
       to shrink/grow files, and *particularly* when considering portability.
  6.4) It's worth a shot / worth benchmarking, but I don't think I'd have the 
time to be a main driver on this.